Handbook of Smart Energy Systems 9783030979393, 9783030979409

155 89 105MB

English Pages 3382 Year 2023

Table of contents :
Preface
Contents
About the Editors
Contributors
Part I Reliability Enhancement of Energy Systems
The Need for Self-Sufficiency and Integrated Water and Energy Management
Contents
1 Highlights
2 Introduction
3 The Nexus Water-Energy
3.1 Energy
3.2 Water
3.3 Energy and Water Systems
4 The Development and Installation of the System in the Small Mountain Town of Covilha, Portugal
5 The 4Es and the Integrated Energy and Water System Installed in Covilha
6 Conclusion
7 Glossary
References
Rethinking Renewable Energy Development in the Republic of Kazakhstan from the Perspectives of International Relations
Contents
1 Introduction
2 Electricity Infrastructure and International Relations
3 Regionalism and Energy Regionalism in Central Asia
4 Renewable Energy Development in Kazakhstan
5 The Central Asia Power System and Extra-regional Powers
6 Energy Future for Kazakhstan and Central Asia
6.1 Energy Security: National Versus Regional
6.2 Involvement of Extra-regional Entities: Pros and Cons
7 Conclusion
8 Cross-References
Glossary
References
A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations
Contents
1 Introduction
2 Modeling Aggregate of TCL Populations
2.1 Dynamics of TCLs Under Thermostat-Based Deadband Control with Forced Switching
2.2 A PDE Aggregate Model
2.3 An Equivalent PDE Model
3 Aggregate Power Tracking Control Design
4 Well-Posedness and Stability of the Closed-Loop System
4.1 Well-Posedness Assessment
4.2 Stability Analysis
5 Simulation Study
6 Concluding Remarks
A: Notations on Function Spaces
B: Proof of Proposition 2
C: Proof of Theorem 3
References
Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid
Contents
Nomenclature
1 Introduction
2 Methodology
2.1 US Electric Power Demand Curves
2.2 US Climate Zones and New Residential Building Permits
2.3 Multifamily Residential Prototype Building (Baseline)
2.4 Multifamily Net-Zero Energy (NZE) Residential Building
2.5 Calculation of Net-Electricity Demand Differences
3 Results and Discussion
3.1 NZE Balances of Multifamily Models in US Climate Zones
3.2 Net-Electricity Impact of NZE Residential Buildings on a National Scale
3.3 Potential Aggregate Impact on the Actual US Electricity Demand Profile
4 Conclusions
References
Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network
Contents
1 Introduction
2 System Details
2.1 Wind Turbine
2.2 Photovoltaic Panel
2.3 Fuel Cell and Its Associated Components
3 Reliability
4 Objective Function and Its Constraints
5 Particle Swarm Optimization (PSO) Algorithm
6 Results
7 Conclusions
Appendix
Nomenclature
References
Virtual Power Plants and Integrated Energy System: Current Status and Future Prospects
Contents
1 Introduction
1.1 Scope and Key Contributions of the Chapter
2 VPPs, Microgrids, and Energy Systems
3 Technical Challenges in a VPP
3.1 Mitigation of Power Quality Disturbances
4 VPP with IES: An Energy Informatics Approach
5 Value of VPP with an IES
6 Sector Coupling Options for VPPs
7 Storage Options for VPPs
8 Scale of a VPP in an Electricity Market
9 VPPs of the Future
9.1 Toward Intelligent VPP Solutions
10 Conclusion
References
Reliability Analysis of Smart Grids Using Formal Methods
Contents
1 Introduction
1.1 Reliability Modeling Methods
1.2 Reliability Analysis Methods
2 Problem Statement
3 Proposed Solution
4 State of the Art
4.1 Event Tree Reliability Analysis
4.2 Cause-Consequence Diagram Reliability Analysis
4.3 Functional Block Diagram Reliability Analysis
4.4 Monte Carlo Simulation Reliability Analysis
4.5 Reliability Analysis in HOL4 Theorem Proving
4.6 Formal Reliability Analysis of Smart Power Grids
5 Proposed Methodology
6 Smart Grid Power System Applications
7 Conclusions
References
Multi-objective Reliability-Based Design Optimization of Shell-and-Tube Heat Exchangers Using Combined Subset Simulation Method and Naive Bayes Algorithm
Contents
1 Introduction
2 Mathematical Models for Shell-and-Tube Heat Exchanger Design
2.1 STHE Modeling
2.1.1 Shell Side
2.1.2 Tube Side
2.1.3 Cost Estimation
2.2 Reliability Analysis
2.3 Proposed Reliability Assessment Method
2.4 Multi-Objective Model for STHE Design
3 Result
4 Conclusion
5 Cross-References
References
Key Technologies for the Energy Internet
Contents
1 Introduction
2 Key Technologies for the Energy Internet
2.1 Energy Router
2.1.1 SST-ER-Based Energy Router
2.1.2 Multiport Converter Energy Router
2.1.3 Power Line Communication Energy Router
2.2 Distributed Energy Resources
2.3 Advanced Metering Infrastructure
3 Information and Communications Technology
3.1 Fifth-Generation (5G) Communication Networks
3.2 Pervasive Edge Computing
4 Conclusion
References
How to Simulate If We Only Have Partial Information but We Want Reliable Results?
Contents
1 Need for Reliable Simulations Under Uncertainty: Formulation of the Problem
1.1 Need for Smart Energy Systems: A Brief Reminder
1.2 Smart Energy Systems: Need for Simulations
1.3 Need for Simulations Under Uncertainty
1.4 Need for Reliable Simulation Results
1.5 How This Problem Is Resolved Now and Why New Methods Are Needed
1.6 What We Do in This Chapter and Why This Problem Is Important Beyond Smart Energy Systems
2 Towards Formulation of the Problem in Precise Terms
2.1 What We Want to Estimate: First Approximation
2.2 What We Want to Estimate: A Realistic Description
2.3 What We Have
2.4 Let Us Take Uncertainty into Account
2.5 Resulting Precise Formulation of the Problem
2.5.1 First (Main) Problem
2.5.2 Second (Auxiliary) Problem
2.6 Mathematical Comment
3 Analysis of the Problem
3.1 Linearization Helps Us Simplify the Problem
3.2 Formula for the Range under Linearization
3.3 How We Can Estimate y0
3.4 But How Can We Estimate Δ: Brainstorming
3.5 How to Simulate Cauchy Distribution
3.6 Caution
3.7 How to Reconstruct Δ from the Final Expression
4 Resulting Algorithm
4.1 The Problem: Reminder
4.2 First Step: Computing y0
4.3 Second Step: Preparing Cauchy Distributed Random Variables
4.4 Third Step: Re-Scaling, to Enable Linearization
4.5 Fourth Step: Running Simulations
4.6 Final Steps
4.7 Caution: The Proposed Algorithm Is NOT a Usual True-to-Life Simulation
4.8 How Accurate and How Efficient Is This Algorithm
4.9 Experimental and Theoretical Confirmation
References
Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data
Contents
1 Introduction and Fundamentals
2 The Pseudo Code of the Goodness-of-Fit Algorithms
3 The Mathematica Codes of Cramér–von Mises and Anderson-Darling Tests for Weibull and Lognormal Distributions
4 Illustrative Examples
5 Conclusions and Suggestions
References
Bayesian Network for Composite Power Systems Using Hybrid Mutual Information Measure
Contents
1 Introduction
2 Introducing the HMI Measure and Using it to Construct the BN Associated with Composite Power System
2.1 The BN Construction Procedure
2.1.1 Training Data Generation
2.1.2 BN Construction
2.2 Mutual Information and Normalized Mutual Information Measures
2.3 A New Hybrid Dependency Measure
2.4 Component Ranking
2.5 Case Study
2.5.1 Comparison of MI, NI, and HMI Measures
2.5.2 Final Structure of BN Associated with MRTS
2.5.3 Numerical Evaluation of Component Ranking
3 Conclusion
References
Optimal Management of Smart Home Appliances Considering Stochastic Behavior of Wind Turbine
Contents
1 Introduction
2 Home Energy Management System
3 Wind Turbine
4 Objective Functions
4.1 Electricity Bill
4.2 Peak Demand
5 Constraints
6 Energy Management Algorithm
7 Numerical Results and Discussion
7.1 Case 1
7.2 Case 2
8 Conclusion
References
Novel Data-Driven Methods for Evaluating Demand Response Programs in a Smart Grid
Contents
1 Introduction
2 Project Background and Problem Statement
3 Two Data-Driven Demand Response Evaluation Methods
3.1 Notation
3.2 A DR Evaluation Method for Coincident Load Reduction
3.3 Evaluating Annual Energy Savings
4 Computational Results
4.1 Effect on Reducing Coincident Load
4.2 Effect on Annual Energy Savings
4.3 Effect of Critical Peak Pricing
5 Conclusions
6 Cross-References
References
On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis
Contents
1 Introduction
2 Features of Electricity Markets and the Evolution of Thermal Plants
3 Market Structure Versus Market Forces
4 Revisiting the Convergence Issue
5 Club Convergence
References
Recent Developments in the Smart Energy Systems
Contents
1 Introduction
2 Concept of Smart Energy Systems
3 Key Technological Developments for Smart Energy Systems
3.1 Intelligent Novel Energy Power Generation Forecast Technology
3.2 Advanced Energy Storage Technologies
3.3 Digital-Energy Integrated Market and Service Mechanism
4 Application of IoT in Smart Energy Systems
4.1 Application in Policies for Energy Operations
4.2 Application in Energy Storage
4.3 Application in Grid Control and Power Generation
4.4 Application in Actuation and Sensing
4.5 Application in Data Visualization
4.6 Application in Forecasting
4.7 Application in Real-Time Environment Monitoring
4.8 Recent Advancement in IoT
5 Case Study: Smart Energy Towns in China
5.1 Constraints to Build China's SET
5.2 Driving Force for China's SET
5.3 Recommendations to Build China's SET
6 Challenges in Smart Energy Systems
7 Future Prospects in Smart Energy Systems
8 Conclusions
References
Constructing Renewable Energy Systems Using Big Data Applications
Contents
1 Introduction
1.1 Renewable Energy
1.2 Big Data Technology
2 An Integrated Big Data Architecture in Renewable Energy
2.1 Blockchain for Smart Grid
2.2 Forecasting models
2.2.1 Short-Term Forecast
2.2.2 Long-Term Forecast
3 Conclusion
References
Control and Monitoring of Wind Farms Based on IoT Application for Energy Conversion
Contents
1 Introduction
2 Current Status (Before Utilization of IoT)
3 Analytics Platform
4 Conclusion
References
Smart Grid and Resilience
Contents
Abbreviations
1 Introduction
2 Introducing the Challenge and Necessity of the Study
3 Literature Review
4 Part I: Investigation of the Effect of Charging Electric Combined Vehicles on Power Losses and Voltage Deviations in the Distribution Network
4.1 Uncoordinated Charging
4.2 Coordinated Charging
5 Part II: Investigating the Effect of Increasing the Number of Vehicles on Losses and the Cost of Investing in the Power Network
6 Reduce Investment in Peak Hours with Smart Charging Strategy
7 Transfer of Charging Time to Non-peak Hours
7.1 Increased Energy Losses
8 Part III: Optimal Charging of Electric Vehicles by Observing the Restrictions of the Maximum Distribution and Transmission Network
8.1 Variables Required to Operate the System under Study
8.2 Analysis of the Proposed Method
9 Conclusion
10 Verify of Results by Conventional Methods
11 Suggestions for Future Work
References
Grid Integration of Wind Energy Using Fuzzy Logic Algorithm
Contents
1 Introduction
2 Permanent Magnet Synchronous Generator – Wind Turbine System Modeling
2.1 Simulation Result Analysis
2.2 Modeling of Permanent Magnet Synchronous Generator (PMSG)
3 Rotor Side Control Structure
3.1 Fuzzy Logic Controller
4 Grid Side Control of WECS
5 Simulation and Results
6 Conclusion
References
Quantitative Methods for Data-Driven Next-Generation Resilience of Energy Systems and Their Supply Chains
Contents
1 Introduction
2 Supply Chain Resilience: Definitions and Tenets
3 Performance and Structural Metrics for Resilience Assessment
3.1 Performance Metrics
3.2 Structural Metrics
4 A Framework for Next-Generation Data-Driven Resilient Energy Supply Chains
5 Conclusions and Future Research Directions
6 Cross-References
References
Hybrid Attack Modeling for Critical Energy Infrastructure Protection
Contents
1 Introduction
2 Recent Works
3 Definitions
4 Energy Infrastructure Model
5 Hybrid Operation Model
5.1 Spatial Model
5.2 Time Model
5.3 Goal Hierarchy
5.4 Hybrid Attack Model
6 Hybrid Threat and Risk Assessment
6.1 Events
6.2 Scenario
6.3 Scenario-Case Approach to Threat and Risk Assessment
7 Conclusion
8 Cross-References
References
Social Cost-Benefit Analysis of Emission Norms: A Conceptual Framework
Contents
1 Introduction
2 Literature Review
3 General Equilibrium Analysis
4 Deadweight Loss
5 Substitution Effect
6 Social Cost Benefit Analysis
7 Conceptual Framework
8 Benefit of Emission Norms
9 Direct Benefits
9.1 Direct Reduction in Emissions
9.2 Regular Update on Vehicle Health
9.3 Indian Emissions Norms at Par with Global Standards
10 Wider Impact (WI)
10.1 Improvement in Health
10.2 Reduction in Pollutants in the Air
10.3 Regular Update of Emission Standards
10.4 Push for Alternates Such as Elective Vehicles and Biofuels Like Ethane
11 Costs of Emission Norms
12 Compliance Costs (C.C)
13 Production Costs
13.1 Refined Engines
13.2 Higher Labor Costs
13.3 Modification of Auxiliary Parts
13.4 New and Improved Gauge Installation
13.5 Increased Production Time
13.6 Assembly Line Reconfiguration
14 Operational Costs
14.1 Replacement of BSIV Vehicles at the Retail End
14.2 Gradual Removal of Noncompliant Vehicles by Government
15 Administrative Costs
15.1 Training of Workers for New Engine Assembly
15.2 Awareness Drive in Public by NGO and Government Agencies
16 Cost to Consumers (C. Co.)
16.1 Increase in Vehicle Costs
16.2 Emission Norms Compliance Rates Such as Regular Vehicle Pollution Checks, Vehicle Urea Level Maintenance
17 General Equilibrium
18 Substitution Effect
19 Conclusion, Limitations, and Future Directions
A.1 Glossary
References
Graphene-Based Wearable Sensors
Contents
1 Introduction
2 Graphene Patterning Method
2.1 Hard Mask
2.2 Inkjet Printing
2.3 Photolithography and Plasma Etching
2.4 Transfer Printing
2.5 Other Methods
3 Graphene Wearable Devices
3.1 Graphene Electrophysiological Sensor
3.1.1 EMG
3.1.2 ECG
3.1.3 EOG
3.1.4 EEG
3.2 Graphene Mechanical Sensor
3.3 Graphene Gas Sensor
3.3.1 Humidity Sensors
3.3.2 Acetone Sensors
3.3.3 NO2 Sensor & NH3 Sensor
3.4 Graphene Fluids Sensor
3.4.1 Glucose
3.4.2 Micromolecular Organic
3.4.3 Macromolecular Organic
4 Perspectives and Challenges
5 Conclusion
References
Piezoelectric Polymer Composites for Energy Harvesting
Contents
1 Introduction
2 Piezoelectricity and Energy Harvesting Technology
3 Advances in Piezoelectric Polymer Composites
4 Material Classification
4.1 Piezoelectric Polymers
4.2 Piezoelectric Ceramics
4.3 Piezoelectric Single Crystals
4.4 Piezoelectric Polymer Composites and Nanocomposites
4.5 Polylactic Acid
5 Applications
6 Present Scenario of Piezoelectric Energy Harvesting
7 Future Prospects
8 Summary
References
Practical Applications of Gaussian Process with Uncertainty Quantification and Sensitivity Analysis for Digital Twin for Accident-Tolerant Fuel
Contents
1 Introduction
2 Digital Twin Framework for Advanced Nuclear Fuel Development
3 Gaussian Process (GP)
4 Applications of a GP to Nuclear Fuel Development
5 Conclusions
References
Energy Consumption Management System for Smart Cities Using IoT Cloud
Contents
1 Introduction
2 Smart Cities
2.1 Migrating the Urban Areas to Smart Cities
3 Energy Consumption
3.1 Managing Energy Using Microgrids
4 Microservices
4.1 Design of Microservices
4.2 Running Microservices on Many Cores
5 IoT Cloud
5.1 How Does the IoT Cloud Work?
5.2 How Does the IoT Cloud Manage the Power?
6 Workflow Architecture
7 Functionality of the Application
7.1 Authentication
7.2 Collection of Information
7.3 Collection of Characteristics
7.4 Advising Services
7.5 Implementation Use Case
8 Future Uses Cases
8.1 Failure Probability Modeling
8.2 Outage Detection and Prediction
8.3 Dynamic Energy Management & Demand Response Management
8.4 Smart Grid Security and Theft Detection
9 Conclusion
References
An Overview of Drone Energy Consumption Factors and Models
Contents
1 Introduction
2 Factors Affecting Energy Consumption of Drones
2.1 Drone Design
2.2 Environment
2.3 Drone Dynamics
2.4 Delivery Operations
3 Modeling of Drone Energy Consumption
3.1 Optimal Path Planning
3.2 Path Following Control
3.3 Battery-Aware and Battery Performance
3.4 Target Tracking
3.5 UAV-Enabled Mobile Edge Computing
3.6 UAV-Enabled Multicasting
3.7 Drone's Component Models
3.8 Joined Models
4 Cross-References
References
Big Data Applications for Improving the Reliability of the French Electricity Distribution Grid
Contents
1 Introduction
2 A New Technological Range of Equipment, with More Communication Capabilities for a Better Knowledge of the Network Status
2.1 New Communicating Chains
2.1.1 Linky Smart Meter, Concentrator, and Communicating Chain
2.1.2 Smart Secondary Substations
2.1.3 Fault Indicators Connected to a Concentrator
2.2 IoT Sensors and IoT Platform for the Electricity Grid
2.2.1 IoT Platform
2.2.2 Fault Indicators on Overhead Network
2.2.3 Flood Sensors in Secondary Substation
2.2.4 IoT Surge Arrester
2.2.5 Transformer Monitoring Sensor Stations
3 Technical Tools and Solutions
3.1 Big Data Platform
3.1.1 Shared File System
3.1.2 Big Data Processing Engine
3.1.3 Event Streaming Platform
3.2 Leverage on Data Preparation
4 Dealing with Network Aging
4.1 LV Network Replacement Optimization
4.1.1 Method
4.1.2 Results
4.1.3 Overhead Feeders
4.2 MV Network Replacement Optimization
4.2.1 Data Pipeline and Transformation
4.2.2 Underground Networks
4.2.3 Overhead Networks
4.3 From Statistic Modeling to Value Creation
5 Climatic Resilience of the Electricity Grid Network
5.1 Storms
5.1.1 Dataset Description
5.1.2 Model Features
5.1.3 Test Methodology and Result
5.1.4 Links with the Overhead MV Network Targeting Model
5.2 Flooding
5.2.1 Flood Scenarios
5.2.2 Flood Impact Tool
5.2.3 Flood Impact Underlying Model
5.3 Heat Waves
5.3.1 Statistical Models
5.3.2 Operational Model
5.3.3 Crisis Management Organization
6 Conclusion
7 Cross-References
References
Structural Global Reliability Analysis of Floating Offshore Wind Turbines
Contents
1 Introduction
2 Modeling of Stochastic Dynamic Excitations of FOWTs
2.1 Simulation of Random Wind Speed Field
2.2 Simulation of Random Irregular Waves
2.3 Joint Probabilistic Modeling of Combined Environmental Excitations
3 Integrated Dynamic Modeling of FOWTs
3.1 Integrated Structural Dynamic Modeling
3.2 Load Calculations of FOWTs
4 Reliability Analysis Method Based on PDEM
4.1 Generalized Probability Density Evolution Equation (GDEE)
4.2 Dynamic Reliability Evaluation Based on GDEE
5 Dynamic Reliability Analysis of FOWTs: An Example
6 Conclusion
References
A Physics-Regularized Degradation Model for Cooling System Health Management
Contents
1 Introduction
2 A Physics-Informed System Health Degradation Model
2.1 Motivating Example
2.2 Modeling Strategy and Some Practical Considerations
2.3 The Modeling
2.4 Numerical Illustration
3 Conclusion
References
Part II Simulation and Optimization of Energy Systems
A Stackelberg Game-Theoretic Model of Fee-and-Rebate Pricing in a Load-Reduction Emergency Demand Response Program
Contents
1 Introduction
1.1 Emergency Demand Response Program
1.2 Development of Game-Theory-Based Model
2 Principles of Rebate and Scarcity Pricing Determination in a Demand Response Program
2.1 Scarcity Level
2.2 Rebate Determination
2.3 Scarcity Pricing Determination
3 Stackelberg Game Formulation for an Emergency Demand Response Program
3.1 Model
3.2 Solution Technique
3.2.1 Primal and Dual
3.2.2 Theorem 1: Weak Duality
3.2.3 Theorem 2: Strong Duality
3.2.4 Exemplar Primal and Dual Pair of LPs
3.3 Implementation of Procedures
4 Performance of Stackelberg Game Model: A Case Study
4.1 Agent Participation
4.2 Distribution of Rebates
4.3 Resolution for Scarcity Price
5 Conclusions
References
Energy Simulation Optimization for Building Insulation Materials
Contents
1 Introduction
2 Energy Simulations
3 An Energy Simulation Application for Thermal Insulation
4 Conclusion
References
Machine Learning for Building Energy Modeling
Contents
Nomenclature
1 Introduction
2 Machine Learning for Generating Building Energy Models (BEMs)
3 Climate Change Impacts on Building Energy Change
3.1 Projected Average Global Temperatures from CMIP5 RCP
3.2 Projected Average Global Temperatures from CMIP6 SSP
3.3 Comparison of Average Global Temperatures from CMIP5 RCP with CMIP6 SSP
3.4 Impacts of Rising Temperatures on Building Energy Demands
4 A Practical Guide to Developing BEMs and Forecasting the Impact of Climate Change on Buildings Using Python
5 Future Directions
6 Key Takeaways
7 Cross-References
References
Machine Learning to Facilitate the Integration of Renewable Energies into the Grid
Contents
1 Introduction
2 Literature Review
2.1 Prediction Methods
2.2 Storage
2.2.1 Pumped Storage Hydropower
2.2.2 Compressed Air Storage
2.2.3 Flywheel
2.2.4 Battery
2.3 Solar Cells for a Good Production
3 Development of Renewable Energies
3.1 Renewable Energy
3.2 The Importance of Making Smart Energies
3.3 Advantages of Artificial Intelligence
4 Machine Learning
4.1 Supervised Learning
4.2 Unsupervised Learning
5 Data Description
6 Methods
6.1 Linear Regression
6.2 Gaussian Process Regression
6.3 Artificial Neural Networks
6.4 Error Calculation
7 Results and Comparison
8 Conclusion
References
Energy Management of Smart Homes by Optimizing Energy Consumption Scheduling
Contents
1 Introduction
1.1 Household Energy Management Without the Use of Renewable Energy
1.1.1 Time of Using
1.1.2 Real-Time Pricing
1.1.3 Critical Peak Pricing
1.1.4 Peak Time Rebates
1.2 Household Energy Management Using Renewable Energy
2 Smart Home
2.1 Energy-Consuming Appliances in a Smart Home
2.2 Energy Storage Systems in Smart Homes
3 Optimization Models
3.1 Model 1
3.2 Model 2
3.3 Model 3
3.4 Model 4
3.5 Model 5
4 Discussion and Concluding Remarks
References
Applications of Machine Learning for Renewable Energy: Issues, Challenges, and Future Directions
Contents
1 Introduction
2 Machine Learning in Renewable Energy Prediction
2.1 Machine Learning in Predicting Wind Energy
2.2 Machine Learning in Predicting Solar Energy
2.3 Machine Learning in Predicting Hydropower
3 Conclusion
4 Cross-References
References
Multi-criteria Optimization for System Integration of Decentralized Off-Grid Hybrid Renewable Polygeneration
Contents
1 Introduction
2 Inputs to the Polygeneration
3 Optimization of Polygeneration Systems
4 Few Examples
4.1 Case Study 1
4.2 Case Study 2
5 Conclusions
References
Optimal Risk Management of Electric Power Systems with CLOUD Simulation and Security Meter Algorithms
Contents
1 Introduction, Motivation, and Literature Survey
2 Software-Based Methods for Optimizing Electric Power Generation Risk Management
2.1 Repair Rate Planning
2.2 Load Planning
2.3 Crew Planning
2.4 Product Planning
2.5 Security Meter Risk Assessment and Risk Mitigation Planning
3 Applications of the Reviewed Methods to Electric Power Systems for Optimal Risk Management with Cost and Benefit
3.1 Interpretation of Table 4's CLOURAM Software Results with Cost and Benefit Analysis
4 Discussions and Conclusion
A.1 Appendix A
B.1 Appendix B
B.1.1 How to Install CYBERRISKSOLVER to RUN the CLOUD Assessment Derated and Security Meter APPLETS
5 Cross-References
References
A Novel Mathematical Model for Infrastructure Planning of Dynamic Wireless Power Transfer Systems for Electric Vehicles
Contents
1 Introduction
2 Infrastructure Allocation
3 System Architecture, Assumptions, and Limitations
4 Optimization Model
4.1 Model Transformation
4.2 Renewable Energy Integration
4.3 Numerical Example
5 Discussion
References
Application of Machine Learning for Energy-Efficient Buildings
Contents
1 Introduction
2 Mathematical Definition of M&V
3 Existing M&V Method for Building
4 Data and Problem Description
5 Overview of One-Dimensional Convolution Network and Long Short-Term Memory (LSTM) Cell
5.1 Convolution Neural Network and LSTM Cell
6 Metrics Development
6.1 Kolmogorov-Smirnov Test Integral (KSI) and OVER Metrics
6.2 Root Mean Square Error (RMSE) and Root Mean Quartic Error (RMQE)
6.3 Maximum Absolute Error (MXAE), Mean Absolute Error (MAE), and Mean Bias Error (MBE)
6.4 Coefficient of Determination (COD)
7 Proposed Deep Network Architecture with Novel Two-Stepped Training Process
7.1 Proposed Deep Network Architecture
7.2 Two-Stepped Training Process
8 Results
9 Conclusion
References
A Simulation-Based Optimization Framework Applied to Assess the Resilience of Energy Distribution Center
Contents
1 Introduction
2 Literature Reviews
3 Resilience Management
3.1 Deterministic Optimization
3.2 Stochastic Optimization
3.3 Simulation-Based Optimization
4 Resilience Framework
4.1 Energy System Structure
4.2 The Main Framework
4.2.1 Resilience Assessment
4.2.2 Resilience Management
4.3 Scenario Generation
4.4 Strategy Planning
5 Conclusion and Future Directions
References
Mathematical Modeling and Simulation Validation in Optimizing Multi-objective Energy Systems Performance
Contents
1 Introduction and Background
2 Methodology
2.1 Model Formulation
2.2 Uncertainty
2.3 Multi-objective Decision-Making
3 Case Study
4 Results
5 Conclusion
References
Machine Learning Methods for Estimating Energy Performance of Building Facade Systems
Contents
1 Introduction
1.1 Objectives
1.2 The Novelty Element and Rationale
1.3 The Key Questions
1.4 Outline
2 Background
2.1 Building Energy Performance Assessment Methods
2.2 Building Energy Performance Estimation Using Predictive Machine Learning Algorithms
2.2.1 K-Nearest Neighbors
2.2.2 Linear Regression Models
2.2.3 Tree-Based Regression Models
2.2.4 Support Vector Machines
3 Methodology
3.1 The Database for Building Energy Savings of UHP-FRC Facade Systems
3.2 Predictive ML Model Development, Parameters Tuning, and Regularization
4 Results and Discussion
4.1 Descriptive Summary
4.2 Results of Hyperparameter Tunings
4.3 The Performance of Machine Learning Models in Predicting Total, Heating, and Cooling Energy Savings of Implementation of UHP-FRC Facade Panels
4.3.1 Best Model Parameters and Performance of K-NNs
4.3.2 Best Model Parameters and Performance of Regression Models
4.3.3 Best Model Parameters and Performance of Tree-Based Models
4.3.4 Best Model Parameters and Performance of SVMs
5 Conclusion and Recommendations for Future
References
Systematic Literature Review of Multi-criteria Decision-Making Applied to Energy Management
Contents
1 Introduction
2 Background
2.1 Energy Management
2.2 Multi-Criteria Decision-Making
3 Methodology
4 Results
5 Conclusions
References
Game Theory Modeling of Energy Systems
Contents
1 Introduction
2 What Is Game Theory?
3 The Transition in Energy Systems and the Application of Game Theory Modeling
3.1 Factors Shaping the Energy Transition
3.2 The Need for a New Modeling Framework
3.3 Game Theory Models in Energy System Problems
4 Areas of Game Theory Application in Energy Systems
4.1 Game Theory Modeling of Electric Vehicle Charging Problems
4.2 Development, Control, and Optimization of Smart Energy Systems
4.2.1 Development of Energy Storage Capacity in Energy Systems
4.2.2 Hybrid Energy System Planning
4.2.3 Game Theory Modeling and Control of Microgrids
4.3 Modeling Large-Scale Energy System
4.4 Energy Trading, Demand Response, and Bidding Strategy
4.5 Generation Expansion Planning
4.6 Energy Policy Analysis
5 Limits of Game Theory Modeling
6 Conclusion
References
Risk Assessment of Electric Power Generation Systems by Stochastic Simulation Using CLOUD Computing
Contents
1 Introduction and Motivation
2 Theoretical History, Notation and Methods, Two- and Three-State Units, Interdependence, and Start-Up Failure, and Start-Up Delays
2.1 Theoretical History
2.2 Notation and Methods
2.3 Two-State (Up, Down) and Three-State (Up, Derated, Down) Units when FOR (Forced Outage Rate) Nonconstant r.v.
2.3.1 Two-State Units
2.3.2 Three-State Units
2.3.3 Interdependence of Units, Start-Up Failures and Start-Up Delay
3 Input Management, Applications to Large Power Systems, Outcomes and Conclusions
3.1 Types of Generation Data, Collected Samples and Other Input Materials
3.2 Case Studies, Tabulations, and Outcomes
3.3 General Discussion, Interpretations of Outcomes, and Conclusions
3.3.1 General Discussion
3.3.2 Interpretations of Outcomes and Conclusions
4 Cross-References
Appendix A
References
Green Scheduling of a Complex Flexible Manufacturing Problem
Contents
1 Introduction
2 Literature Review
3 Methodology and Structure
3.1 Simulated Annealing (SA) Algorithm
3.2 Harmony Search (HS) Algorithm
4 Analysis of Two Algorithms SA/HS Algorithms in Classical Mode
4.1 Comparison Between the Outputs of Two Algorithms in Classical Mode
4.2 Analysis of Two Algorithms SA/HS Algorithms in DRC Mode
4.3 Comparison Between the Outputs of Two Algorithms in DRC Mode
5 Conclusion
6 Cross-References
References
A Simulation-Based Framework for the Adequacy Assessment of Integrated Energy Systems Exposed to Climate Change
Contents
1 Introduction
2 The Case Study
3 Climate Change Modeling
4 Adequacy Assessment
5 Results
5.1 Base Case Scenario
5.2 Demand Increase Scenario
5.3 Demand and Supply Increase Scenario
6 Conclusion
A.1 Climate Change Impact on the IES Adequacy
A.1.1 Solar PV Power
A.1.2 Wind Power
A.1.3 Thermal Power Plants
A.1.4 Electrical Grid
A.1.5 Demand
A.1.6 List of Acronyms
A.1.7 List of Symbols
References
A Brief Guide on the Modeling of Green Vehicle Routing Problems
Contents
1 Introduction and Historical Notes
2 Problem Definition
3 Formulations
3.1 Defining and Bounding the Variables
3.2 Reducing the Search Space Through Preprocessing Rules
3.2.1 Preprocessings I
3.2.2 Preprocessings II
3.2.3 Preprocessings III
3.2.4 Preprocessings IV
3.3 Modeling the Problem Through Objective Function and Constraints
3.3.1 Defining the Objective Function
3.3.2 Degree Constraints
3.3.3 Connectivity Constraints
3.3.4 Fuel Capacity Constraints
Exponential Size Constraints
Polynomial Size Constraints
3.3.5 Model Improvements
3.3.6 Time Constraints
3.3.7 Complete Formulation
4 Final Remarks
References
US Natural Gas Consumption Analysis via a Smart Time Series Approach Based on Multilayer Perceptron ANN Tuned by Meta-heuristic Algorithms
Contents
1 Introduction
2 Material and Methodology
2.1 Multilayer Perceptron (MLP)
2.2 Particle Swarm Optimization Algorithm (PSO)
2.3 Imperialist Competitive Algorithm (ICA)
3 Illustration Example
4 Conclusion
5 Cross-References
References
Simulation and Optimization of Energy Systems
Contents
1 Introduction
2 Modeling and Simulation
2.1 Physics-Based Models
2.2 Data-Driven Models and Digital Twining
2.3 Limitations of Each Modeling Approach
2.4 Statistical Meteorological Data
2.5 Existing Open-Source Simulation Models
3 Optimization
3.1 Optimization Methods
3.1.1 Linear Programming and Nonlinear Programming
3.1.2 Evolutionary Algorithms: Genetic Algorithm
3.1.3 Evolutionary Algorithms: Particle Swarm
3.2 Cost Minimization
3.3 Optimal Integration of Energy Systems
4 Control
4.1 Model Predictive Control
4.2 Fuzzy Logic Control
5 Conclusion
References
Data Clustering Method for Probabilistic Power Flow in Microgrids
Contents
1 Introduction
1.1 Background
1.2 Literature Review
2 Problem Formulation
2.1 Objective Functions
2.2 Constraints
2.2.1 Nonlinear Deterministic Power Flow Model
2.3 Constraints for the Convex Deterministic Power Flow Including Reconfiguration
3 Uncertainties Modeling Via Sampling
3.1 Uncertainty Modeling for Wind Power Generation
3.2 Uncertainty Modeling for Solar Power Generation
3.3 Uncertainty Modeling for Power Demand
4 Lightning Attachment Procedure Optimization-Based K-means Data Clustering
5 Case Studies and Results
5.1 Case Study I
5.2 Case Study II
5.3 Case Study III
5.4 Case Study IV
6 Cross-References
References
Material Supply Network Optimization in the Energy and Utility Industry
Contents
1 Introduction
2 Literature Review
2.1 Energy Distribution Network Design
2.2 Material Supply Chain Design
3 Basic Optimization Framework
4 Extensions and Variants
4.1 Fixed-Charge Flow
4.2 Location-Inventory Decision
4.3 Supply Disruption
5 A Case Study
6 Conclusions and Future Research
References
Application of Digital Twin Technology on Simulation and Optimization of Prime Movers in Energy Systems
Contents
1 Introduction
2 Modeling and Simulation
2.1 Model Zooming Example
3 Model-based Techniques
3.1 Gas Turbine Systems
3.2 Performance Monitoring
3.3 Performance Tracking
3.4 Diagnostics (Detection and Isolation)
3.5 Prognostics (Prediction and Life Estimation)
3.6 Compressor Fouling Example
4 Gas Turbine Faults and Degradation Modes
4.1 Faults and Degradation Classification
4.2 Compressor Turbine Degradation Example
5 Integrated Control and Health Monitoring
5.1 Accommodation
5.2 Life Extension
5.3 Model-Based Control
5.4 Soft Sensor Example
6 Implementation and Deployment of Digital Twins
6.1 Performance Tracking Example
References
Digital Twin for Multi-criteria Decision-Making Framework to Accelerate Fuel Qualification for Accident Tolerant Fuel Concepts
Contents
1 Introduction
2 Fuel Pellets
2.1 Monolithic Fuel
2.1.1 UN
2.1.2 U3Si2
2.2 Composite ATF Fuels
2.2.1 UN-UO2
2.2.2 U3Si2-UO2
3 Integration of Experiments with Computational Research Efforts for ATF
3.1 Thermal Hydraulics (TH)
3.2 Fuel Performance
4 What Is Digital Twin/Material Twin?
4.1 Applications of Digital and Material Twin for Nuclear
4.1.1 Cooling System
4.1.2 Material Applications of DT
4.1.3 Support of Management and Control
4.1.4 Ansys Digital Twin Builder: Applicability of Digital Twin for ATF
5 Conclusion
References
Model Predictive Control and Distributed Optimization in Smart Grid Applications
Contents
1 Introduction
2 Microgrid Modeling and Problem Formulation
2.1 Microgrid Modeling
2.2 Demand-Side Management and Selection of an Objective Function
3 Model Predictive Control
4 Distributed Optimization
4.1 Distributed Optimization: Requirements, Challenges, and the Communication Structure in Smart Grid Applications
4.2 Dual Ascent as an Example of a Lower Level Optimization Scheme
4.3 Coordinated Descent as an Example of Algorithm 3
5 Concluding Remarks
References
High-Performance Solar Cells by Machine Learning and Pareto Optimality
Contents
1 Introduction
2 Solar Cell Structure
3 Optimization of Solar Cells
3.1 Pareto Optimality
3.2 Machine Learning
4 New Trends
References
Drone Delivery Systems and Energy Management: A Review and Future Trends
Contents
1 Introduction
2 Drone Energy Suppliers
2.1 Battery-Based Supplying Techniques
2.1.1 UAVs Powered by Batteries
2.1.2 Swapping
2.1.3 Laser-Beam Inflight Recharging
2.1.4 Tethered UAVs
2.2 Fuel Cell Powered UAVs
2.2.1 The Drawback of Fuel Cells Efficiency Issues
2.2.2 Fuel Cell Types
2.2.3 Fuel Storage
2.3 Hybrid Power Sources
2.3.1 Fuel Cell and Battery
2.3.2 The Use of Solar Cells as an Auxiliary Power Source
2.3.3 The Use of Supercapacitor as an Auxiliary Power Source
3 Energy Management Strategies
3.1 Rule-Based Strategies
3.1.1 Fuzzy Logic Strategies
4 Conclusions
5 Cross-References
References
Applications of Machine Learning in the Planning of Electric Vehicle Charging Stations and Charging Infrastructure: A Review
Contents
1 Introduction
2 Review of Different Machine Learning Used for EV Charging Station Location Problem
3 Present Research on Charging Station Network Design
3.1 Traditional Mathematical and Probabilistic Models
3.2 Heuristic and Data-Driven Algorithms
3.3 Optimization Models
4 Conclusion and Future Directions
5 Cross-References
References
Surrogate Modeling-Driven Physics-Informed Multi-fidelity Kriging for the Prediction of Accident-Tolerant Fuel Properties
Contents
1 Introduction
2 Multi-fidelity Kriging (MFK)
3 Material Property Modeling
4 Conclusion
References
Robust Design Optimization Method for Engineering System
Contents
1 Introduction
2 Optimization Under Uncertainty
3 Test Case and Analysis
4 Conclusion
References
Data-Driven Multi-scale Modeling and Robust Optimization of Composite Structure with Uncertainty Quantification
Contents
1 Introduction
2 Robust Optimization and Multiscale Problem
3 Case Study: Multiscale Modeling with Uncertainty and Sensitivity
4 Conclusion
References
Data Analysis Applications in Optimizing the Smart Grid System
Contents
1 Introduction
2 Problems of Smart Grid Systems Optimization and Development
2.1 Problem 1: Transmitted Power Maximization
2.2 Problem 2: Optimization of the Distribution Network Configuration
2.3 Problem 3: Power Deficit Minimization
2.4 Problem 4: Power Loss Minimization
3 Assessment of the Dependencies Between Power Flows and the Generation/Consumption Parameters in the Power Grid Nodes
4 Assessment of the Dependencies Between Power System State Parameters and FACTS Parameters
5 Smart Grid Clustering
6 Conclusions
7 Cross-References
References
Data Analysis Applications in the Optimal Integration of Energy Supply Chain
Contents
1 Introduction
2 What Is an Energy Supply Chain?
3 How the Energy Supply Chain Can Be Optimized?
3.1 Understand and Digitize the Whole Supply Chain and Business Operation
3.2 Avoiding the Assets' Overpayments
3.3 Innovation Sustainability
4 Using Big Data in Supply Chains
4.1 Consumer Behavior and Usage Patterns
4.2 Enhanced Storage Management
4.3 Accurate Forecasting to Meet Client Demand
4.4 Collection of Energy Data
4.5 Analytics from Energy Data
5 Using Data Analytics in Energy Supply Chain
5.1 Scalability of Energy Analytics
5.2 Modeling the Probability of Failure
5.3 Detection and Forecast of the Outages
5.4 Dynamic Energy Management
5.5 Theft Detection and Smart Grid Security
5.6 Preventive Equipment Maintenance
5.7 Demand Response Management
5.8 Real-Time Customer Billing
5.9 Improvement of Operational Efficiency
5.10 Optimization of Asset Performance
5.11 Improvement of Customer Experience
5.12 Machine Learning Algorithms in Energy Distribution
5.13 Prediction of Energy Prices
5.14 Prediction of Energy Demand
5.15 Prediction of Energy Price Merit Order
5.16 Improvement in Trading
5.17 Understanding the Unsatisfied Customers
5.18 Calculating the Customer Lifetime Value
5.19 Calculating the Chances of Winning a Customer
5.20 Customized Offers
5.21 Understanding the Energy Usage at the Device Level
6 Conclusion
References
Developing Energy Demand Forecasting Methods
Contents
1 Introduction
2 Brief Literature Review on Load Forecasting
3 Ranges of Load Forecasting
4 Load Forecasting
4.1 Factors Affecting the Load
4.2 Properties of the Load Curve
5 Time Series Models
5.1 Moving Average
5.2 Exponential Smoothing
5.3 ARIMA Models
6 Artificial Intelligence Models
6.1 Regression Model
6.2 Expert Systems
6.3 Neural Networks
6.3.1 Feedforward Neural Network
6.3.2 Recurrent Neural Network
Long Short-Term Memory Cell
Gated Recurrent Unit Cell
7 Performance Assessment
8 Load Forecasting Process
9 Conclusion
References
Data-Driven Energy Waste Minimization at Energy Distribution Networks
Contents
1 Introduction
2 Smart Energy Distribution Networks
3 Data-Driven Analytics
4 Data-Driven Energy Waste Minimization in Energy Networks
4.1 Current Trends in Data-Driven Energy Waste Minimization
4.2 Challenges and Future Research
5 Conclusion
6 Cross-References
References
Applications of Data Envelopment Analysis (DEA) for Optimizing Energy Consumptions
Contents
1 Introduction
2 DEA Models
3 Applications of DEA in Agricultural Productions
4 Applications of DEA in Renewable Energy Systems
4.1 Applications of DEA in Biofuel and Bioenergy Systems
4.2 Applications of DEA in the Static Renewable Energy Systems
4.2.1 Solar Energy Systems
4.2.2 Wind Energy Systems
5 Applications of DEA in Non-renewable Energy Systems Including Oil, Gas, Electricity, Coal, and Thermal Power Industries
6 Applications of DEA in Other Systems Including Dairy Farms, Industries, Transportation Systems, Airlines and Energy Efficiency Evaluation Among Different Countries
7 Discussion
8 Conclusion
References
Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-objective Energy System Performance
Contents
1 Introduction
1.1 Decision Making
1.2 Decision-Making Process
1.3 Choosing the Best Strategy
1.4 Decision-Making Models
2 Multi-Objective Optimization
2.1 Feasible Set and Decision Space
2.2 Feasible Space Map and Criterion Space
2.3 Order Relations
2.4 Cone
2.5 Pointed Cone
2.6 Efficient Solutions
2.7 Ideal Point
2.8 Nadir Point
2.9 Utopia Point
2.10 Optimal Solution
2.11 Strictly Optimal Solution
2.12 Scientometric Analysis
3 Multi-Objective Optimization Algorithms
3.1 Epsilon-Constraint Technique
3.1.1 Applications
3.2 Normal Boundary Intersection
3.2.1 Applications
4 Multi-Criteria Decision-Making Techniques
4.1 Decision-Making Criteria
4.2 Measurement Criteria
4.3 MCDM Models
4.4 Scientometric Analysis
4.5 MCDM Methods
4.5.1 Fuzzy Satisfying Method
Applications
4.5.2 VIKOR Decision Maker
Applications
4.5.3 TOPSIS
Applications
4.5.4 AHP
Applications
4.5.5 ANP
Applications
5 Conclusion
References
Simulation Applications in Analyzing the Trade-Off Between Climate Change and Energy Consumption
Contents
1 Introduction
2 Simulation Methodology
3 Global Warming
4 Biodiversity
5 Fossil Fuel Consumption
5.1 Urbanization
5.2 Agricultural and Livestock Products Demand
5.3 Industrial Products Demand
5.4 Energy Efficiency
6 Deforestation
7 Simulation
8 Solutions
8.1 Public Awareness
8.2 Policies, Laws, Guidelines
9 Conclusion
References
Analysis of Energy Transition Pertaining to the Future Energy Systems
Contents
1 Introduction
2 Boundary and Framework of the Review
3 An Overview of Energy Transition
4 Trends on Energy Management in the Power System
5 A Brief Overview on Energy Storage Systems
6 Strengths, Challenges, and Future Research Needs
7 Conclusion and Future Perspectives
References
Data Mining Applications in Smart Grid System (SGS)
Contents
1 Introduction
2 Data Mining Applications in SGS
2.1 Data Mining in Demand Response
2.2 Data Mining in Transmission and Distribution
2.3 Data Mining in Security
2.3.1 Anomaly Detection
2.3.2 False Data Injection
2.3.3 Denial-of-Service (DoS)
2.4 Data Mining in Predictive Maintenance and Fault Management (PM&FM)
2.4.1 Power Outage
2.4.2 Equipment Breakdown
2.4.3 Electrical SGS Faults
2.4.4 Arc Faults
3 Processing Software of Data Mining
3.1 Data Mining Tools
3.2 Data Visualization Tools
4 Challenges of Using Data Mining in SGS and Future Works
5 Conclusion
6 Cross-References
References
Part III Intelligent Development of Energy Systems
The Role of Blockchain and Cryptocurrency in Smart Grid: Renewable Energy Trading, System Security and Privacy Preservation
Contents
1 Introduction
2 Smart Grid Key Challenges
2.1 Renewable Energy Trading
2.2 How to Make the Economics Work?
2.3 How to Deal with Self-Interests and Concerns when Making a Political Decision?
2.4 Where to Build and how Big to Build?
2.5 How to Manage Decentralized Generation, Distribution, and Redistribution?
2.6 How to Supply on Demand?
2.7 How to Cope with the Existing Infrastructure?
2.8 How to Build a Healthy Ecosystem in the Energy Sector?
3 System Security
4 Privacy Preservation
5 Cryptocurrencies and Blockchain Technologies
6 Blockchain-Based Renewable Energy Trading
7 Energy Web Token (EWT)
8 Blockchain-Based Smart Grid System Security
9 Blockchain-Based Smart Grid Privacy Preservation
10 Conclusion
References
Energy Harvesting for Smart Energy Systems
Contents
1 Introduction
2 Literature Review
3 Smart Grids
4 The Main Characteristics of Smart Grids
5 Conscious and Active Participation of Consumers in Smart Grids
6 Modification of Production and Amount of Reserves
7 Provide the Required Power Quality
8 Flexibility in the Face of Disasters and Natural Disasters
9 New Products, New Services, and New Markets
10 Intelligent Networks for Optimizing Equipment Use and Increasing Performance Efficiency
11 Rearrangement Goals in Distribution Networks
12 Network Rearrangement in the Presence of Loss Reduction
13 Rearrangement in the Network in Order to Balance the Load of Feeders
14 Rearrangement in Order to Recover the Load and Increase Reliability
15 Rearrangement in Order to Melt the Ice of the Lines
16 Methods of Reducing Network Losses
17 Improved Frog Mutation Algorithm
18 Analysis of Results and Simulation
19 The Following Scenarios Are Considered in Solving this Problem
19.1 Scenario 1: Basic Scenario
19.2 Scenario 2: Rearrangement Alone
19.3 Scenario 3: Placement of Scattered Products in the Network
19.4 Scenario 4: Locating a Capacitive Bank in the Network
19.5 Scenario 5: Rearrangement with Simultaneous Placement of Distributed Generation Sources and Optimal Capacitance in the Network
20 Conclusion
21 Offers:
References
Dynamic Bayesian Network Based Approach for Modeling and Assessing Resilience of Smart Grid System
Contents
1 Introduction
2 Resilience of Smart Grid
2.1 The Perspective of Resilience
2.2 Resilience Capacities
2.2.1 Absorptive Capacity
2.2.2 Adaptive Capacity
2.2.3 Restorative Capacity
3 Dynamic Bayesian Network
4 Approach Details
5 Case Study
6 Discussion
6.1 Sensitivity Analysis of the Model
7 Conclusion
References
Application of Machine Learning in Occupant and Indoor Environment Behavior Modeling: Sensors, Methods, and Algorithms
Contents
1 Introduction
2 Literature Review on Data-Driven Occupancy Estimation
3 Literature Review on Machine Learning-Based Window State Estimation
4 Literature Review on Machine Learning-Based Window Opening/Closing Action Estimation
5 Literature Review on Machine Learning-Based Indoor Temperature Prediction
6 Literature Review on Machine Learning-Based Occupancy Status Prediction and Implementation of Occupant-Centered HVAC Control
References
Software Engineering Smart Energy Systems
Contents
1 Introduction
2 Background
2.1 Software Engineering
2.2 Service-Oriented Computing
2.3 AI Planning
2.4 Demand-Side Management
3 Software Engineering Future-Proof Energy Systems
3.1 Smart Energy Systems' General Requirements
3.1.1 Functional Requirements
3.1.2 User-Related Requirements
3.1.3 Non-functional Requirements
3.2 The Blueprint
4 Case Study: Smart Energy Offices
4.1 Design
4.1.1 Formal Approach
4.1.2 Separation of Concerns
4.2 Implementation
4.3 Benefits
4.3.1 Energy Efficiency
4.3.2 Accuracy
4.3.3 Scalability
5 Case Study: Following Smart Grid Signals
5.1 Optimal Operation Scheduling
5.1.1 Objectives
5.1.2 Constraints
5.2 Cost-Efficiency and Sustainability
6 Conclusions
7 Cross-References
References
An Intelligent Decision Support System for an Integrated Energy Aware Production-Distribution Model
Contents
1 Introduction
2 Mathematical Model and Assumptions
3 Solution Methodology
3.1 GA and BBO Algorithms
3.2 Solution Representation and Decoding
3.3 Migration and Crossover Operator
3.4 Mutation Operator
4 Results and Discussion
4.1 Hypothesis Test
4.2 Discussion and Implications
5 Conclusion
6 Cross-References
References
Design and Operational Strategies for Grid-Connected Smart Home
Contents
1 Introduction
2 Smart Devices and Sensors at Grid Level
3 Smart Devices and Sensors at Home Level
4 Designing of Existing Household Devices in PSCAD
5 Data Flow Between Smart Grid and Smart Homes
References
The Role of Data Collection, Storage, and Processing in the Intelligent Energy Systems of Tomorrow
Contents
1 Introduction
2 Energy Meters and Data Collection
3 Data Analysis Methods in the Evolution of ISE
4 Application of Data Analysis in the Progress of ISE
5 Conclusions, Challenges, and Further Research Directions
References
Adaptive Decentralized Under-frequency Load Shedding in Smart Grid
Contents
1 Introduction
2 Proposed Method
3 Conclusion
References
Photovoltaic Rooftops in Smart Energy Systems
Contents
1 Introduction
2 Load Consumption in Dwellings: Recording Interval in Smart Meters
3 Photovoltaic Power Estimation
4 Photovoltaic Self-Consumption and Self-Sufficiency in PV Rooftops
4.1 Direct Self-Consumption and Self-Sufficiency Curves. ZEB Point
5 Discussion
5.1 PV Rooftops in Residential Sector
5.2 PV Rooftops in Industrial Sector
5.3 Impact of the Recording Interval When Estimating Photovoltaic Self-Consumed Energy
6 Conclusion
References
Smart Grid Resilience
Contents
1 Introduction
2 Distributed Cyber-Physical Systems in Smart Grid
3 Reliability and Resilience
3.1 Reliability
3.2 Resilience
3.3 Electric Grid Resilience in Relation to Electric Grid Reliability
3.4 Electric Grid Resilience Metrics
4 Disruptive Events and Threats to Smart Grid
4.1 Extreme Natural Events
4.2 Accidental Disruptive Events
4.3 Man-Made Disruptive Events
5 Enhancing Smart Grid Resilience
5.1 Electric Grid Resilience Framework
5.2 Strategies for Smart Grid Resilience
5.2.1 Planning and Design
5.2.2 Operation Capabilities
6 Concluding Remarks
References
Smart Energy Systems, Infrastructure Financing, and the Wider Economy
Contents
1 Introduction
2 Smart Energy Systems
2.1 An Overview of the Scientific World
3 Smart Power Grids
4 Electric Power Transmission Networks
5 Smart Grid
6 Smart Energy Systems, Infrastructure Financing, and the Economy
7 Conclusion
8 Cross-References
References
Multi-step Ahead Power Demand Forecasting in Smart Grid
Contents
1 Introduction
2 The Proposed Model
3 Data Analysis and Computational Results
4 Conclusion
References
Social Acceptance of Smart Energy Systems
Contents
1 Introduction
2 Smart Grids
3 Smart Grids for Electricity System
4 Components of the Smart Grid
4.1 Smart Generation
4.2 Smart Stations
4.3 Smart Distribution
4.4 Smart Meters
4.5 Integrated Communication
4.6 Advanced Control Methods
5 The Benefits of the Smart Grids for the Energy Consumers
6 Smart Grid Concept and Implementations in the World
6.1 Europe
6.1.1 Italy Telegestore Project (2000–2005)
6.1.2 Malta Enemalta Project (2008–2013)
6.1.3 Germany DENA I and II Smart Grid Projects
6.2 The United States
6.3 China
7 Impacts of the Smart Systems on the Consumption Behavior
8 Conclusion
References
Edge Computing in Smart Grids
Contents
1 Introduction
2 State of the Art
2.1 Efficient Energy Management
2.2 Enhanced Fault Detection
2.3 Smart Charging of Electric Vehicles
2.4 Enhanced Privacy and Security
3 The Role of IoT, Edge Computing, and Big Data in Smart Grids
4 Challenges and Future Directions
References
Resilience of Smart Integrated Energy Systems
Contents
1 Introduction
1.1 Design Principles for a Resilient Smart Grid
2 Resilience Evaluation Frameworks
2.1 Resilience Metrics
2.2 Emerging Techniques for Resilience Evaluation
3 Resilience Strategies
3.1 Introduction to Resilience Strategies
3.2 Cyber-Physical Situational Awareness
3.3 Adaptive Mechanism
3.3.1 Dynamic Partitioning and Reconfiguration
3.3.2 Dynamic Resource Allocation
4 Outlook
References
Smart Buildings in the IoT Era: Necessity, Challenges, and Opportunities
Contents
1 Introduction
2 What is IoT?
3 Smart Building
3.1 Sensing Capabilities
3.1.1 Sensor Types
3.1.2 Software Technologies
3.2 Targeted Systems and Applications
3.2.1 HVAC
3.2.2 Plug Loads
3.2.3 Lighting
3.2.4 Water Recycling Systems
4 Challenges
5 Opportunities
5.1 Digital Twin
5.2 Artificial Intelligence and Machine Learning
6 Way Forward
6.1 Interoperability
7 Summary
References
Smart Grids in the IoT Era: Necessity, Challenges, and Opportunities
Contents
1 Introduction
2 Background
2.1 Internet of Things (IoT)
2.2 Smart Grids
3 IoT in Smart Grids
3.1 Smart Grids Challenges
3.1.1 Inadequacies in Grid Infrastructure
3.1.2 Cyber Security
3.1.3 Storage Concerns
3.1.4 Data Management
3.1.5 Communication Issues
3.1.6 Stability Concerns
3.1.7 Energy Management and Electric Vehicle
3.2 IoT-Based Smart Grids
3.3 Architecture of IoT-Based Smart Grids
3.3.1 Physical Implementation
3.3.2 IoT Models in Smart Grids
3.3.3 Computing Platforms and Web Services
3.3.4 Application Protocols
4 Challenges of IOT-Based Smart Grids
4.1 Implementation Challenges
4.1.1 Connection Stability and Communication
4.1.2 Cost
4.2 Technical Challenges
4.2.1 High Heterogeneity and Lack of Proper Standards
4.2.2 Network Congestion and the Challenge of Communication Delay
4.2.3 Cyber-Security and Privacy
4.2.4 Contextual Data and Information
4.2.5 Big Data Analytics
4.2.6 Provision of Service Quality
4.2.7 Protocol Supporting
4.2.8 IPv4 Addressing Limit
4.2.9 Legal Aspects and Social Facets
4.2.10 Service Detection
4.3 Security and Privacy Challenge Domains
4.3.1 Customer Domain
4.3.2 Information and Communication Domain
4.3.3 Challenges in the Grid Domain
5 Conclusion
6 Cross-References
References
Governance of Smart Energy Systems
Contents
1 Introduction
2 Smart Energy Systems
3 Governance of Smart Energy Systems
4 Conclusion: Open Questions and Future Directions
References
A Hybrid Smart Neural Network Model for Short-Term Prediction of Energy Consumption
Contents
1 Introduction
2 Methodology Mythologies and Structure
2.1 ANFIS
2.2 GA Structure
2.3 Fitness Value Function
3 Numerical Examples (Case Study)
4 Conclusion
5 Cross-References
References
Processing Smart Meter Data Using IoT, Edge Computing, and Big Data Analytics
Contents
1 Introduction
2 Smart Meter
2.1 Working of Smart Meter
2.2 Functionality of Smart Meter
2.3 Communication Network in Smart Meter System
2.4 Benefits of Smart Meter
3 IoT and Big Data
3.1 Working of IoT and Big Data Together
3.2 Requirements of Big Data and Analytics in IoT Environment
4 Working of IoT and Big Data in Smart Grid
References
Energy-Smart Transportation Systems
Contents
1 Introduction
2 Alternative Fuels in Transportation
3 Connected and Automated Vehicles Energy Efficiency
4 Digital Twin and Smart Grids
5 Smart Mobility Services
6 Green Vehicle Routing Problem
7 Eco-driving
8 Big Data and Machine Learning Role
9 Conclusions and Future Direction
References
Energy Efficient Mission Control of Unmanned Intelligent Swarm Systems
Contents
1 Introduction
2 Technical Background
2.1 Flight Dynamics of Quadrotors
2.2 Telecommunications
2.3 Decentralization Levels
2.4 Proportional-Integral-and-Derivative (PID) Controller
3 Computational Problems in Swarm Missions
3.1 Formation
3.2 Mapping
3.3 Trajectory Following
3.4 Task Assignment
3.5 Data Routing
4 Softwares
4.1 Design and Animation
4.2 Simulation
4.3 Libraries
5 Conclusions
References
Building the Remote Surveying System of Energy Consumption in Maritime Transportation Using Internet of Things (IoT) Technique
Contents
1 Introduction
2 Literature Review
2.1 Related Works
2.2 Data Acquisition System (DAS)
2.3 Remote Measures of Energy Consumption
3 Materials and Methodologies
3.1 Target Ship
3.2 Internet of Things (IoT)
4 Results and Discussion
4.1 The Proposed Model of Supervising the Energy Consumption
4.1.1 Data Collection Module
4.1.2 Data Analytics Module
4.2 Experimental Investigation on Target Ship
5 Conclusion
References
AI for Electricity Market Design
Contents
1 Introduction
2 What Is Artificial Intelligence (AI)?
3 Impact and Examples of AI?
4 What Is Cognitive Computing?
5 Different Types of Artificial Intelligence?
6 Reactive Machines
7 Limited Memory
8 Theory of Mind
9 Self-Awareness
10 Artificial Intelligence and Related Topics and Terminology?
11 What Is Machine Learning?
12 What Is Deep Learning?
13 What Is Artificial Neural Network (ANN)?
14 Artificial Intelligence in Electricity Sector?
15 How Does AI Help to Improve Electricity Sector?
16 Applications of AI in Power Sector?
17 Looking into the Future with AI in the Power Sector?
18 Conclusion
References
Predicting US Energy Consumption Utilizing Artificial Neural Network
Contents
1 Introduction and Literature Review
2 Variable Selection and Datasets Used
2.1 Selection of Predictor Variables
2.2 Datasets
3 Methodology
3.1 Backpropagation Algorithm
3.2 Error Estimation
3.3 Proposed Feed-Forward Backpropagation Network
4 US Energy Consumption Prediction Results
5 Conclusion
6 Cross-References
References
A Survey of Cyber-physical Systems Applications (2017–2022)
Contents
1 Introduction
2 CPS Applications
2.1 CPS and Smart Agriculture
2.2 CPS and Education
2.3 CPS and Energy Management
2.4 CPS and Environmental Monitoring
2.5 CPS and Transportation
2.6 CPS and Health Care
2.7 CPS and Process Control
2.8 CPS and Security
2.9 CPS and Smart City/Smart Home
2.10 CPS and Smart Manufacturing
3 Discussions
4 Conclusions and Implications for Future Research
5 Cross-References
References
Using Artificial Intelligence for Nuclear Nonproliferation and Commercial Nuclear Applications
Contents
1 Introduction
2 Potential Areas of Nonproliferation Concern
3 Early Warning Systems
4 Autonomous Defenses
5 Integrated Decision Support
6 NonMilitary Applications
7 Discussion
8 Conclusion
References
Machine Learning and Artificial Intelligence-Driven Multi-scale Modeling for High Burnup Accident-Tolerant Fuels for Light Water-Based SMR Applications
Contents
1 Introduction
2 Accident Tolerant Fuels for Light Water Reactors
3 ATF Research for LWR-Based SMRs
4 Machine Learning and Nuclear Power Industry
5 Machine Learning and AI in Nuclear Materials Research
6 AI-Driven Research in LWR-Based SMRs
7 Machine Learning and Artificial Intelligence-Based Multiscale Modeling Incorporating Digital Twin
7.1 Neutronics Calculation and Optimization
7.2 Thermal-Hydraulics
7.3 Fuel Performance Code
7.4 Digital Twin Surrogate Models
7.5 Robustness Analysis Using Machine Learning
8 Discussions
9 Conclusion
References
Blockchain and Smart Grids: Opportunities, Open Issues, and Future Prospects
Contents
1 Introduction
2 The Concept of Blockchain
2.1 Architecture of Blockchain
2.2 Characteristics of the Blockchain Network
2.3 Consensus Algorithm
2.4 Approaches to Consensus
2.5 Smart Contracts
3 Applications of Blockchain in Smart Grids
4 Challenges of Blockchain-Based Smart Grids
4.1 Scalability
4.2 Cost of Implementation
4.3 Lack of Legal and Regulatory Frameworks
4.4 Privacy and Security
4.5 Lack of a Fair Reward and Punishment System
5 Practical Initiatives
5.1 Energy Tokens
5.1.1 SolarCoin
5.1.2 NRGcoin
5.1.3 Charg Coin
5.1.4 Efforce (WOZX)
5.1.5 Energy Web Token (EWT)
5.1.6 Echo Tech Coin (ECOT)
5.1.7 Rox Utility Token (ROX)
5.2 P2P Energy Trading
5.2.1 Enerchain
5.2.2 Power Ledger
5.2.3 Lightency
5.2.4 LO3 Energy
5.2.5 ETIBLOGG
5.2.6 Suncontract
6 Concluding Remarks and Future Directions
6.1 Future Directions
6.1.1 Energy Nonfungible Tokens (NFTs)
6.1.2 Higher Investment by Governments and Public Institutions
6.1.3 Greener Blockchain
References
A Blockchain-Based Smart Grid to Build Resilience Through Zero-Trust Cybersecurity
Contents
1 Introduction
2 Prior Work
3 The Proposed Model
3.1 Blockchain-Based Smart Contract and Energy Resilience
3.2 Peer-to-Peer Trading and Smart Grids
3.3 Cybersecurity and Smart Grids
4 Discussion
4.1 Managerial
4.2 Future Research
5 Conclusions
References
Blockchain and Open Energy Markets
Contents
1 Introduction
2 Open Energy Market
3 Blockchain Technology
3.1 Economics ofBlockchain
4 Smart Grids and Blockchain
4.1 Decentralization
4.2 Supply Chain Management
4.3 Cyber Security
4.4 Transparency
4.5 Good Governance
5 Summary
6 Cross-References
References
Leveraging Industry 4.0: Deep Learning, Surrogate Model, and Transfer Learning with Uncertainty Quantification Incorporated into Digital Twin for Nuclear System
Contents
1 Introduction
2 Industry 4.0 in the Context of Nuclear System
3 Surrogate Modeling, Deep Learning, and Transfer Learning
4 Digital Twins
5 Uncertainty Quantification
5.1 Polynomial Chaos Method
5.2 Monte Carlo Method
5.3 Neural Network Method
5.4 Sensitivity Analysis Method
5.5 Dimension Reduction Method
5.6 Inverse Uncertainty Quantification
6 Uncertainty Quantification and Nuclear Industry
7 Conclusions
References
Using AI and Classical Controllers for Improving the Renewable Energy Systems
Contents
1 Introuction
2 Wind Turbine
3 Wind Turbine Generators
3.1 Wind Turbine Categories
3.2 Wind Turbine Control Methods
4 Photovoltaic Systems
4.1 Sun Trackers
4.2 Control Methods
5 The Most Recent Researches Review
6 Conclusion
7 Cross-References
References
Uncertainty Quantification and Sensitivity Analysis for Digital Twin Enabling Technology: Application for BISON Fuel Performance Code
Contents
1 Introduction
2 Polynomial Chaos for Uncertainty Quantification
2.1 Multidimensional or Multivariable Polynomials
2.2 Regression Method to Estimate PC Coefficients
3 Sensitivity Analysis
4 Demonstrations to Fuel Performance Code: BISON
5 Conclusion
References
Wind Turbine Anomaly Detection Based on SCADA Data
Contents
1 Introduction
2 Wind Turbine SCADA Parameters and Signal Processing
2.1 Wind Turbine SCADA Parameters
2.2 Signal Processing of Wind Turbine SCADA Parameter
3 SCADA-Based Anomaly Detection
3.1 Control Chart-Based Approach
3.2 Change Point Detection Approach
3.3 Power Curve-Based Approach
3.4 Normal Behavior Model Approach
3.5 Other Approaches
4 A Brief Discussion on Advanced Analysis
5 Conclusion
References
Future Energy System Analyses
Contents
1 Introduction
2 Electrical Infrastructure Futures
2.1 Smart Grid and Energy Internet
2.1.1 Defining the Smart Grid
2.1.2 Smart Grid Studies
2.1.3 The Energy Internet
2.2 Super Grid and Supersmart Grid
2.2.1 Defining the Super Grid
2.2.2 Super Grid Studies
2.2.3 The Supersmart Grid
2.3 Further Options
2.3.1 Integrated/Multienergy/Smart Energy Systems
2.3.2 Sector Coupling
3 Assessing New Technologies
4 Energy Data Analytics
4.1 Data Categories and Uses
4.2 Data-Related Challenges
5 Modeling, Simulation, and Testbeds
5.1 Energy Systems Modeling
5.2 Modeling of Critical Energy Infrastructures
5.3 Energy System Testbeds
6 Socio-Technical-Economic Perspectives
6.1 Energy Security and the Energy Trilemma
6.2 Transition Research
7 Concluding Remarks
8 Cross-References
References
Hybrid Threats to the European Union's Energy Sector: An Overview
Contents
1 Introduction
2 Hybrid Threats to the Energy Sector in Europe
2.1 Hybrid Threats, the Infrastructure Domain, and the Energy Sector
2.2 State Actors and Their Strategic Interests
2.2.1 Russia
2.2.2 Algeria
2.2.3 Nigeria
2.2.4 China
2.2.5 Morocco
2.3 Tools Directly Targeting the Energy Sector
2.3.1 Priming Phase
2.3.2 Destabilization and Coercion Phases
2.4 Indirect Influence on the Energy Sector
2.4.1 Intra-Domain Dependencies
2.4.2 Inter-Domain Dependencies
3 Conclusion and Future Challenges
4 Cross-References
References
Green Production
Contents
1 Introduction
2 Green Production Strategy
2.1 Market
2.2 Standard and Regulations
3 Sustainability in Green Production
4 Modeling and Framework
4.1 Performance Assessment
4.2 Planning
4.3 Implementation
4.4 Monitoring
5 Advances in Energy Resources and Storage
5.1 Solar Energy
5.2 Wind Energy
5.3 Bioenergy
5.4 Geothermal Energy
5.5 Hydropower and Tidal Energy
5.6 Supercapacitor
6 Conclusion
References
Integration of Renewable Energy Sources
Contents
1 Reliability and Adequacy Indices
1.1 Introduction
1.2 Important Reliability Indicators
1.3 The Effect of Renewable Resources on Reliability Indicators
1.3.1 Case Study 1
1.3.2 Case Study 2
1.4 Evaluation of EDNS and LOLP Indices
1.5 Conclusion
2 Integration of Renewable Energy with an Emphasis on Short Circuiting, Island Mode, Contingency Analysis, and Energy Storage Systems
2.1 Introduction
2.2 Impact of Renewable Sources of Energy on Short Circuit Current
2.3 Island Mode Operation
2.4 Contingency Ranking
2.5 Evaluate the Reduction of Wind Power Plant Capacity
2.6 Conclusion
3 Harmonic Indices in Unbalanced and Balanced Modes
3.1 Introduction
3.2 Evaluation of Power Quality Indicators
3.3 Comparison of the Effect of Unbalanced and Balanced States
3.4 Conclusion
References
Integration of Renewable Energy Systems
Contents
1 Introduction
2 The Significance of Transition to Renewable Energy
3 Environmental Sustainability
4 Political and Social Aspect
5 Economic Development and Growth
6 Technical Aspect
7 Challenges
7.1 Voltage Imbalances
7.2 Harmonic Distortions
7.3 Frequency Fluctuations
8 Solutions to Challenges
8.1 Smart Grid
8.2 Smart Grid and Cyber Security Issues
9 Energy Storage Systems
10 Conclusion
References
Part IV Sustainable Development of Energy Systems
Planning to Incorporate Energy Conservation Practices, Renewable Energy Production Systems, and Eco-friendly Building Design Practices to Support Sustainability in US Public Schools
Contents
1 Introduction
1.1 Business Problem
2 Conservation Strategies for Reducing School Energy Use
2.1 Energy-Efficient Buildings and the Built Environment
2.2 Permaculture and Aquaculture
2.2.1 Additional Components of Permaculture to Reduce Energy Use
2.3 Incorporating Aquaculture to Improve Energy and Environmental Sustainability
2.3.1 Measuring the Value of Permaculture or Aquaculture Using Odum's Concept of an Energy Evaluation
2.4 Calculating Energy Savings from Permaculture and Aquaculture
3 Sustainable School Power Generation
3.1 Case for Sustainable Power Generation with Schools
3.2 Modeling Estimated Energy and Financial Savings
3.2.1 Commercial Solar Viability
Initial Cost Components of Commercial Solar
Model School Solar Installation
Permitting and Fees
Deciding Whether to Incorporate Solar
3.3 Commercial Wind Generation for Schools
3.3.1 Cost Components of Commercial Wind Turbines
Wind Equipment
Installation
Permitting and Fees
Maintenance Costs with Wind Energy Generation
3.3.2 Deciding Whether to Incorporate Wind
4 Renewable Energy Returns and Total Cost of Ownership
5 How School Energy Production Contributes to Community Sustainability
6 Conclusion
References
An Integrated AI-Multiple Criteria Decision-Making Framework to Improve Sustainable Energy Planning in Manufacturing Systems: A Case Study
Contents
1 Introduction
2 Literature Review
3 Problem Definition
4 The Proposed Methodology
4.1 Simulation by AnyLogic Software
4.2 Life Cycle Assessment by SimaPro Software
4.3 Make a Final Decision by TOPSIS Method
4.4 Machine Learning by Multilayer Perceptron Neural Network
5 Results
6 Conclusion
References
Novel Integrated Membrane Auto-Thermal Reactors (NIMATRs) for Energy Efficiency and Sustainability
Contents
Abbreviations
1 Introduction
2 The Environmental and Sustainable Development Engineering, System Theory
2.1 Main Components of SD
3 Dehydrogenation (Dehydro) and Hydrogenation (Hydro) Reactions
3.1 Dehydro
3.2 Dehydro Reaction and the Generated Hydrogen Consumption
3.3 Endothermic (Endo) Dehydro Coupled with Exothermic (Exo) Reactions
3.4 The Thermal Coupling and the Auto-Thermal (Auto) Reactors
3.4.1 Recuperative Coupling Reactor
Recuperative Coupling Reactors Design
Examples of Recuperative Reactions Coupling with Membrane
3.4.2 Regenerative Coupling Reactor
3.4.3 Direct Coupling Reactor
3.5 The Membrane Reactor
4 Membrane Reactors for the Fischer-Tropsch Synthesis
4.1 Fischer-Tropsch Reaction
4.1.1 Introduction
4.1.2 Gas-to-Liquid Fuel
4.2 High-Temperature Fisher-Tropsch (HTFT) Processes
4.3 HTFT Reactor Type and Developments
4.3.1 Fixed-Bed Reactor
4.3.2 Fluidized-Bed Reactor
4.3.3 The Fluidization Principle
4.3.4 Slurry Bubble Column Reactors
4.3.5 Product Intensifier Reactor (Dual-Type Membrane Reactor)
5 Conclusions
5.1 Important Terminologies
References
Environment Trade-Off in Using Renewable Energy in Oil-Rich Countries
Contents
1 Definitions
2 Introduction
3 Energy Systems in Selected Countries
3.1 Venezuela
3.2 Saudi Arabia
3.3 Canada
3.4 Iran
3.5 Iraq
3.6 Russia Federation
3.7 Kuwait
3.8 United Arab Emirates
3.9 The USA
4 Renewable Energy Policies in Selected Countries
4.1 Venezuela
4.2 Saudi Arabia
4.3 Canada
4.4 Iran
4.5 Iraq
4.6 Russian Federation
4.7 Kuwait
4.8 United Arab Emirates
4.9 The USA
5 Barriers for Renewable Energy Development in Different Countries
5.1 Extensive and Subsidized Distribution of Fossil Fuels
5.2 Market Failures
5.3 Lack of Appropriate Financial Mechanism
5.4 Lacks of Various Renewable Energy Infrastructures
5.5 Policy Barriers
6 Cost-Benefit Analysis of Renewable Energy Development in Different Countries
6.1 Climate and Environment-Related Co-benefits
6.2 Economic Co-benefits
6.3 Social Co-benefits
6.4 Political/Institutional Co-benefits
7 Introducing Socioeconomic Mechanisms to Increase Renewable Energy Use in Oil-Rich Countries
7.1 Tax Measures
7.2 Rebates, Grants, and Performance-Based Incentives
7.3 Loan Programs, Guarantees, and Credit Enhancements
7.4 Feed-in-Tariff
8 Conclusion
References
Data Analytics Applications in the Energy Systems Concerning Sustainability
Contents
1 Introduction
2 Data Analytics Applications in Smart Grids
2.1 Data Collection
2.2 Results
2.2.1 Forecast of Demand and Prices
2.2.2 Change Consumer's Behavior
2.2.3 Increasing Adoption of Green Resources
2.2.4 Increasing EVs (Electric Vehicles) Integration
2.2.5 The Energy Market
2.2.6 Smart Grid Security and Theft Detection
2.2.7 Remote Monitoring
2.2.8 Improve Power Quality
2.3 Discussion
2.3.1 Economic
2.3.2 Environmental
2.3.3 Social
3 Conclusion
4 Cross-References
References
Challenges of Smart Grid Technology Deployment in Developing Countries
Contents
1 Introduction
2 Smart Grid Technology
3 Smart Grid Technology Deployment in Developing Countries: Challenges and Considerations
4 Challenges of Smart Grid Technology Deployment Through the Lens of Technological Innovation System (TIS)
5 Research Methodology
6 Smart Grid Technology in Iran: From the Early Days of Electricity in the Nineteenth Century Up to Now
7 Challenges of Smart Grid Technology Deployment in Iran
8 Conclusion and Policy Recommendation
9 Cross-References
References
Cost-Sustainability Trade-Off Solutions for the Optimal Planning of Local Integrated Energy Systems from Nanogrids to Communities
Contents
1 Introduction
1.1 The Optimal Planning of Local Integrated Energy Systems
1.2 Planning Tools for Local Integrated Energy Systems
1.2.1 DER-CAM
1.2.2 HOMER
1.2.3 EnergyPRO
1.2.4 EnergyPLAN
1.2.5 Calliope
1.3 Gaps and Novelty of the Proposed Approach
2 Modelling of Energy Technologies in Local Integrated Energy Systems
2.1 Design Constraints
2.2 Operation Constraints
2.2.1 Operation Constraints for Generation Technologies
Combined Heat and Power Systems
Natural Gas Boilers
Solar Technologies
2.2.2 Operation Constraints for Conversion Technologies
Reversible Heat Pumps
Absorption Chillers
2.2.3 Operation Constraints for Storage Technologies
Batteries
Thermal Energy Storage Systems
2.2.4 Operation Constraints for Heating Network
3 A Methodology for the Optimal Planning of Local Integrated Energy Systems
3.1 Short-Term Objective
3.2 Long-Term Objectives
3.3 System Balances
3.3.1 Electricity Balance
3.3.2 Thermal Balance
3.4 Optimization Method
4 Case Studies from Nanogrids to Interconnected Energy Hubs in an Energy Community
4.1 Case Study 1: Multi-energy Nanogrid
4.2 Case Study 2: Integrated Energy Community
5 Conclusions
References
A Detailed Analysis of the Barriers of Using Renewable Energies and Their Roles in Sustainable Development in Iran
Contents
1 Introduction
2 Smart Buildings and Green Architecture
3 Introducing Renewable Energies and Their Roles in Iranian Economy
3.1 Solar Energy
3.2 Wind Energy
3.2.1 Measured Values in Meteorological Stations of Iran
3.3 Geothermal Energy
3.4 Biomass
3.5 Marine and Tidal Energy
3.6 Hydropower
3.7 Thermal Power Plants
4 Development of Renewable Energies
4.1 Reliable Energy Resource
4.2 Sustainable Electricity Production
4.3 Diversification of the Country's Energy Basket
4.4 Decentralization
4.5 Economic Development Potential
4.6 Greater Employment Possibility
5 Evaluation of the Country's Development Programs in the Field of Renewable Energy
6 Major Challenges in the Development of RE Sources in Iran
6.1 Suggested Solutions
6.2 Applying Technologies to Find Best Strategies
7 Conclusion
References
Proposing a New Hedging Strategy Based on Considering the Efficiency of Energy Markets in Crises
Contents
1 Introduction
2 Methodology
2.1 AMIM Measure
2.1.1 Normalizing the Coefficients of the AR(p) Model
2.1.2 The Magnitude of Market Inefficiency
2.1.3 Building Confidence Intervals
2.1.4 The Adjusted Market Inefficiency Magnitude
2.2 Hurst Exponent
2.3 Agglomerative Hierarchical Clustering
2.4 Minimum Variance Hedge Ratio (Optimal Hedge Ratio)
3 Data Description and Preliminary Analysis
3.1 Descriptive Statistics
4 Empirical Results
4.1 Hurst Exponent Results and AMIM
4.2 Hierarchical Clustering
5 Conclusion
A.1 Appendix
References
Analyzing the Solar Energy System Investment in Turkey
Contents
1 Introduction
2 Literature Review
3 Methodology
4 The Evaluation of Solar Energy in Turkey
4.1 Specifying Criteria
4.1.1 Economic Criteria
Maintenance Cost
Substructure Setup Cost
Being in the Incentive Zone
Energy Transfer Cost
4.1.2 Environmental Criteria
Work Safety Risk
Amount of Productivity in Areas
Ruggedness Rate of Areas
Suitability for Installation in Areas
4.1.3 Technical Criteria
Substructure Setup Time
Sunshine Duration
The Angle of The Sun Daylight
Amount of Wind Blowing
4.1.4 Energy Criteria
Productivity
Amount of Energy Demand
4.2 Choosing Districts from Turkey
4.2.1 Assumptions
4.3 Implementation
4.3.1 Iterations with AHP
4.3.2 Iterations with ELECTRE
4.3.3 Iterations with PROMETHEE
5 Results
6 Sensitivity Analysis
7 Conclusion
References
The Influence of Circular Economy in Renewable Energy Systems: EoL Solar Panel Management
Contents
1 Introduction
1.1 Circular Economy
1.2 Renewable Energy Systems
1.3 Solar Panels
1.4 Circular Economy of Solar Panels
2 Popularity of PV Panels
2.1 Proposed Solutions of End-of-Life PV Management
2.2 Regulation
2.3 Reverse Logistics
2.4 Life Cycle Assessment
2.5 End-of-Life Recycling and Management
3 PV Recycling (Methods)
4 PV Recycling Responsibility
4.1 Society
4.2 Consumer
4.3 Producer
5 PV Recycling Environmental, Economic Perspective
6 PV Recycling Circular Approach
6.1 Cradle-to-Grave (C2G) Approach
6.2 Cradle-to-Cradle (C2C) Approach
6.3 Designing Circular Business Models (CBMs)
7 Worldwide Macromanagement Status
8 Conclusion
References
Integration Strategies of Renewable Energy Sources in a Conventional Community
Contents
1 Introduction
1.1 Evolution of Smart Grid
1.2 IoT in Smart Grid
1.3 AI Implementation in Smart Grids
1.4 Net Metering
1.5 Advanced Metering Infrastructures
2 Need for Renewable Energy and Its Impact on Environment
3 Strategies of Renewable Energy Integration
4 Case Study
References
IEC-61850 Performance Evaluation in a 5G Cellular Network: UDP and TCP Analysis
Contents
1 Introduction
1.1 The Evolution on Power System Industry
1.2 The Evolution on Cellular Network Industry
1.3 State of the Art
2 Test Bed Description
2.1 Wireless Communication
2.2 LVDC Microgrid Pilot
2.2.1 Architecture
2.2.2 Application
2.3 Test Bed Objectives
3 IEC-61850
3.1 MMS
3.2 Goose
3.3 SV Messages
3.4 R-SV, R-Goose
3.5 Summary of the Standard IEC-61850
4 Test Bed Setup and Tools
4.1 Deployment
4.1.1 Database Description
5 Measurement Procedure and Results
5.1 Throughput Performance Evaluation
5.2 ICMP Performance Evaluation
5.3 Performance Evaluation of the Application
5.4 Other Results
6 Discussion
References
Analysis of the Renewable Energy Generation Capability for Attending a National Renovation Fleet Through Ethanol-Cell Electric Vehicles in a South American Market
Contents
1 Introduction
2 Background
2.1 Current Energetic Demand for the Fleet in Brazil
2.2 Sugarcane Production and Climatic Requirements
2.3 Data Envelopment Analysis (DEA)
3 Method
3.1 Data Collection
3.2 Data Treatment
3.3 Slack-Base Measure (SBM) Data Envelopment Analysis (DEA) Model
3.4 Scenarios Analysis
4 Results and Discussion
5 Final Remarks
6 Cross-References
References
Practical Application of the Decision-Making Grid (DMG) for Supporting Maintenance Strategy Decisions in a Small Hydroelectric Power Plant (SHPP)
Contents
1 Introduction
2 Theoretical Background
2.1 Decision-Making Grid (DMG)
2.2 Analytic Hierarchy Process (AHP)
2.3 Hydroelectric Power Plant Maintenance
3 Method
3.1 DMG: Problematic Equipment Identification
3.1.1 Data Collection
3.1.2 Data Treatment
3.2 AHP: Best Maintenance Alternative Identification
3.2.1 Criteria Definition
3.2.2 Alternatives Definition
4 Results and Discussion
4.1 DMG: Problematic Equipment Identification
4.2 AHP: Best Maintenance Alternative Identification
5 Final Remark
6 Cross-References
References
A Good Practice in Urban Energetics in a Hungarian Small Town, Kaposvar
Contents
1 Introduction
2 History and Situation of District Heating in Kaposvár from the 1990s till 2013
3 The History of District Heating from 2013 till 2021
4 Current Situation of District Heating Services in 2021
5 Operation of CNG Buses
6 Cooperative Utilization of Biogas Produced by the Sugar Factory
7 Daily Deviations, Fluctuations in Energy Consumption
8 The Electric Vehicle Charging Stations
9 The Possibilities for Waste Heat Recovery in District Heating, Based on a Study
10 Urban Energy Outlook, Smart Metering
11 Current Developments, Toward Smart District Heating
12 Conclusion
References
Digital Twin and Artificial Intelligence Incorporated with Surrogate Modeling for Hybrid and Sustainable Energy Systems
Contents
1 Introduction
2 Surrogate Modeling
3 Artificial Intelligence-Driven Surrogate Modeling
3.1 Least Squares
3.2 Inverse Distance Weighting (IDW)
3.3 Kriging
3.4 Radial Basis Functions (RBF)
3.5 Artificial Neural Network
3.6 Support Vector Regression
4 Application in Hybrid and Sustainable Energy Systems
5 Digital Twin Surrogate Models and Associated Uncertainties
6 Conclusion
References
Evaluation of Energy Efficiency for Smart Manufacturing: Applications and Future Scopes
Contents
1 Introduction
2 Performance Evaluation of Energy Efficiency
3 Conclusion
4 Cross-References
References
The Role of Big Data in Building Renewable Energy Systems: The Case of Emerging Markets
Contents
1 Introduction
2 Big Data as an Important Resource for Organizations
3 Smart Renewable Energy Systems
4 Building Smart Renewable Energy Systems with Big Data
5 Conclusion
References
The Relation Among Ownership, Environmental, Social, Governance (ESG), and Corporate Social Responsibility (CSR) in China
Contents
1 Introduction
2 Background and Literature Review
2.1 Chinese Firms' Ownership Structure Background
2.2 China's Environment and Energy Challenges
2.3 State Ownership and Non-state Ownership
2.4 SOEs Work on ESG and CSR
2.5 Institutional Investors Work on ESG and CSR
2.6 Qualified Foreign Institutional Investors (QFIIs) Work on ESG and CSR
2.7 The Other Relevant Researches
3 Discussion
4 Cross-References
References
Energy Semantic Data Management and Utilization in Smart Grid Networks with Focus on Circular Economy
Contents
1 Introduction
2 Literature Review
2.1 Smartness Definition
2.2 Classifications of Smartness
2.3 Semantic Interoperability in Smart Grid Network
2.4 Interoperability in Smart Grid Network
2.4.1 Multi-agent System
2.4.2 Levels of Exchanged Information
2.4.3 Semantic Interoperability
2.4.4 Grid Management
2.4.5 Cyber Security
2.4.6 Consumer Management
2.4.7 Electricity Market
3 Proposed Conceptual Model
3.1 Decentralized Technologies
3.2 Circular Economy Versus Linear Economy
3.3 Knowledge Graph for Smart Grids
3.3.1 Data Acquisition Layer
3.3.2 Knowledge Graph Construction Layer
3.3.3 Knowledge Computation and Management Layer
3.3.4 Intelligent Applications Layer
4 Conclusion
References
Analyzing the Effect of Solar, Wind, Nuclear, and Total Renewable Energy Intensities on Economic Growth in BRICS
Contents
1 Introduction
2 The Future – Solar, Wind, and Nuclear Energy
3 BRICS: The Changemaker and Its Stand on Solar, Wind,and Nuclear Energy Globally
3.1 The Stand of Brazil in Terms of Renewable Energy and Economic Growth
3.2 The Stand of Russia in Terms of Renewable Energy and Economic Growth
3.3 The Stand of India in Terms of Renewable Energy and Economic Growth
3.4 The Stand of China in Terms of Renewable Energy and Economic Growth
3.5 The Stand of South African Terms of Renewable Energyand Economic Growth
4 Data and Methodology
4.1 Augmented Dickey-Fuller Tests
4.2 Johansen Cointegration Test
4.3 Granger Causality Tests
4.4 Vector Autoregressive Tests
4.5 Impulse Response Functions
5 Results and Findings
5.1 Analysis for Brazil
5.2 Analysis for Russia (Table 165Tab55)
5.3 Analysis for India
5.4 Analysis for China (Table 165Tab1212)
5.5 Analysis for South Africa (Table 165Tab1616)
6 Conclusion
References
Supply Chain Resilience Along with Assessment of Sustainable Development: Experimental Data for the Integration of Supply Sources
Contents
1 Introduction
2 Past Works
3 Method
3.1 Experimental Case: Water Heating
3.2 Data and Variables
4 Results
4.1 Environmental Impacts: Life Cycle Assessment (LCA)
4.2 Economic Impact: Life Cycle Cost Savings
4.3 Social Impacts: Social LCA (S-LCA)
5 Discussion
References
The Food-Energy-Water Nexus in Sustainable Energy Systems Solutions
Contents
1 Introduction
2 Problem Statement and Challenges
3 FEWN State-of-the-Art Methods
4 FEWN Optimization Studies
4.1 Food-Centric FEWN Study
4.2 Energy-Centric FEWN Study
4.3 Water-Centric FEWN Study
5 Remarks and Conclusion
6 Cross-References
References
Hydrogen-Based Dense Energy Carriers in Energy Transition Solutions
Contents
1 Introduction
2 Problem Statement
3 Technology Pathways
3.1 Methane Reforming
3.2 Electrolytic Production of Hydrogen
3.3 Hydrogen Storage
3.4 Hydrogen-Based DEC Production
4 Modeling Considerations
4.1 Short-Term Temporal Variability
4.2 Long-Term Temporal Variability
4.3 Scenario Reduction
5 Illustrative Examples
5.1 Integrated Planning and Scheduling with Graph-Based Mode Transition: Hydrogen Economy
5.2 Deterministic Approaches for Integrated Process Synthesis and Global Optimization: Ammonia Production
5.3 A Forecasting Framework to Design Energy-Intelligent Tax Policies: Carbon Tax
6 Discussion
References
A Q-Learning-Based Demand Response Algorithm for Industrial Processes with Operational Flexibility
Contents
1 Introduction
2 The Flexibility Function (FF)
3 System Model
3.1 Industrial Customer Model
3.2 Service Provider Model
4 Q-Learning Policy
4.1 Markov Decision Process
4.2 Q-Learning Policy
5 Numerical Simulation
5.1 Parameter Settings
5.2 Results
6 Conclusion
Nomenclature
References
Indicators Framework for Sustainability and Circular Economy Implementation
Contents
1 Introduction
2 The Structure and Implementation of Circular Economy
3 The Benefits and Barriers of Circular Economy Implementation
4 The Review of Circular Economy and Sustainability Indicators Framework
5 Conclusions and Recommendations
References
Optimal Balancing of Wind Parks with Virtual Power Plants in the Market Environment
Contents
1 Introduction
2 Problem Description
2.1 Power Markets and the Balancing Task
2.2 Features of the Intra-day and Day-Ahead Markets
2.3 Description of the Assets
2.4 Notation
2.5 The Objective
2.6 Constraints for Turbines
2.6.1 Power Constraints
2.7 The Storage Constraints
2.8 The Timing Constraints
2.9 Timing Constraints Tuning
2.10 The Constraints for Batteries
2.10.1 Power Constraints
2.10.2 Storage Constraints
2.10.3 Tightening Constraints
2.10.4 Switches of the Batteries
2.10.5 Components in the Objective Value
2.11 The Cumulative Variables
2.12 Constraints Associated with Hydro Storage Facilities
2.13 The Coupling Constraints
2.14 The Optimization Problem Formulation
2.15 Notation Related to the Output of the Optimization
2.16 A Structural Asset-Based Problem Formulation
3 Description of Gradual Increase
3.1 Inspiration for GI
3.2 Systems of Subsets
3.3 The Initial Problem
3.4 Independent Problems
3.5 Intermediary Problems
3.6 The Feasibility of {Γk(j), k{1,2,…,#A }}
3.7 Pseudocode for Gradual Increase
4 Generalization of Gradual Increase
4.1 Tetris Notation
4.2 Injection of Feasible Solutions
4.3 Sub-pools with Multiple Starts
4.4 Cold Warmstart vs Warm Warmstart
4.5 Upper Bound and MIPgap
4.6 Reformulation of Gradual Increase
4.7 New Possible Facets of GI
5 Operation of the VPP Conducted on Historical Data
5.1 Model Predictive Control
5.2 Robust Model Predictive Control (RMPC)
5.3 RMPC and the Philosophy of Gradual Increase
6 Data Inputs into Optimization
6.1 Price Inputs
7 Nuances of the Optimization and Implementation
7.1 Practical Aspects of Gradual Increase
7.2 Inputs in Code 1 and Code 2
7.3 Scripts and Modules in Code 1 and Code 2
7.4 Hardware and Solvers
8 Results
8.1 Analysis of Price Forecasts
8.2 Optimality
8.3 Speed
8.3.1 Maximum Improvements in Calculation Times
8.3.2 Improvements in Average Time
8.4 Results of Code 1 and Code 2
9 Conclusion
References
Recent Trends in Sustainable Supply-Chain Optimization
Contents
1 Introduction
2 Methodology
2.1 Identification of Research Gap
2.2 Formulation of Research Questions
2.3 Locating Studies
2.4 Screening and Selection
2.5 Analysis of the Literature
3 Results
3.1 Sustainability Concerns
3.2 Decision Levels
3.3 Optimization Methods
4 Conclusions
References
Energy Technology RD&D Budgets, Environmental Sustainability, and Energy Transition: A Review of Emerging Trends, Policies, and Pathways
Contents
1 Introduction
2 Energy Technology RD&D Budgets
2.1 World Energy Investment: An Overview
2.2 Global Investments in Energy Technology RD&D
2.3 Investment Trends Across Energy RD&D
2.3.1 Energy RD&D Investments by Technology
2.3.2 Energy RD&D Investments by Country
3 Environmental Sustainability and Principal Determinants
3.1 Economic Growth
3.2 Financial Development
3.3 Infrastructure
3.4 Energy Transition and Other Determinants
4 The Four-Factor Theoretical Framework for Modeling Energy, Environment, and Climate Change
5 Conclusion and Policy Implications
6 Cross-References
Appendices
Appendix 1 Description – Abbreviation
Appendix 2 List of 136 Countries by Region
References
Green Synthesis of Carbon Nanomaterials
Contents
1 Introduction
2 Traditional Routes for Carbon Nanomaterials Synthesis
2.1 Synthesis Methods for Graphene
2.1.1 Mechanical Exfoliation
2.1.2 Graphene Oxide Reduction
2.2 Synthesis Methods for CNTs
2.2.1 Electric Arc Discharge
2.2.2 Chemical Vapor Deposition
2.2.3 Spray Pyrolysis
2.3 Synthesis Methods for Carbon Nano-Dots
2.3.1 Electrochemical Route
2.3.2 Exfoliation
2.3.3 Laser Ablation
2.3.4 Pyrolysis
3 Green Methods for Synthesis of Carbon Nanomaterials
3.1 Graphene and its Oxide
3.2 CNTs
3.3 Carbon Nano-Dots
4 Applications
4.1 Energy Storage
4.2 Water Treatment
4.3 Food Sector
5 Challenges
6 Conclusions
References
Graphene and Graphene-Based Sustainable Composites
Contents
1 Introduction
2 Fabrication of Graphene-Based Composites
3 Properties of Graphene-Based Composites
4 Applications
4.1 Energy-Related Devices
4.2 Proficient Separation Technologies
4.3 Biomedical Engineering
5 Perspective
6 Conclusions
References
Social Innovation Approaches for Sustainable Living: A Review on Research Methods and Social Problems
Contents
1 Introduction: Understanding the Social Innovation Approaches
2 Social Innovation and Design Thinking
3 Social Innovation Approaches for Social Goods
4 Conclusion
5 Cross-References
References
Internet of Things Value Creation for Sustainable Energy
Contents
1 Introduction
1.1 IoT Applications Taxonomy
1.1.1 IoT in Industry
1.1.2 IoT in Smart City Concept
1.1.3 IoT in Intelligent Grid
1.1.4 IoT in Agriculture
1.1.5 IoT in Health Care
1.1.6 IoT in Transportation
1.2 Economic Impact of IoT on Electric Power and Energy Systems
2 Privacy Challenges Emerged from IoT Devices
2.1 Information Privacy Smart Grid
2.2 Privacy Concerns Implicated by Smart Meter Data Services
2.2.1 Individual Patterns
2.2.2 Real-Time Surveillance
2.2.3 Digital Detritus of Personal Data
2.2.4 Physical Secureness
2.3 Unfolding Privacy Intrusion
3 Conclusion
References
Big Data Analytics in Smart Energy Systems and Networks: A Review
Contents
1 Introduction
2 Descriptive Analysis
3 Diagnostic Analysis
4 Predictive Analysis
5 Prescriptive Analysis
6 Conclusion
7 Cross-References
References
Sustainability in a Circular Economy Context: Investigation of the Market-Based Instruments Potential for Energy Efficiency Improvement (Case Study: Iran)
Contents
1 Introduction
2 Financial Mechanisms for Energy Efficiency Improvement Programs
3 Introducing Market-Based Instruments for Energy Efficiency Improvement
4 Introducing the Iranian Energy Efficiency and Environment Market (IREEEM)
5 SWOT Analysis for the Implementation of the Iranian Energy Efficiency and Environment Market (IREEEM)
5.1 SO1: Developing IREEEM to the Transportation Sector and Including Oil Products
5.2 SO2: Establishing an Evolving Fund to Provide Low-Interest-Rate Loans for Investors
5.3 ST1: Establishing Proper Financial Mechanisms to Support IREEEM Implementation Considering Economic
5.4 ST2: Encouraging the Government, the Key Stakeholder, to Support this Market
5.5 ST3: Establishing a Detailed Mechanism to Specify the Role of Involved Organizations and Better Cooperation
5.6 WO1: Enforce IREEEM Implementation
5.7 WO2: Providing an Opportunity for the Private Sector to Export Energy Equal to Energy-Saved Certificates
5.8 WO3: Preparing Guidelines for Improving Energy Intensity in Different Sectors
5.9 WO4: Support Establishing New ESCOs
5.10 WO5: Trading GHGs Emissions Reduction As a By-Product of Energy Efficiency Projects
5.11 WT1: Labeling Energy Consumers Based on Their Energy Performance
6 Conclusion
References
Wireless Charging of Electric Vehicles Through Pavements
Contents
1 Introduction
1.1 SAE J2954 Standard
2 Review
2.1 Wireless Charging Using Electromagnetic Induction or Magnetic Resonance
2.2 Wireless Charging by Unique Pavement Structures, Materials, or Charging Systems
2.3 Wireless Charging Using the Electric Field
3 Conclusion and Future Research Directions
3.1 Shortcomings of WCTs
3.2 Future Scope
References
Cost-Environment Trade-Off in Using Renewable Energy in an Energy System
Contents
1 Introduction
1.1 The Trend in Renewable Energy Capacity Development
1.2 Policies to Accelerate Renewable Energy Development
2 Advantages and Challenges of Developing Zero-Emission Electricity Generation Capacity
2.1 Advantages of Renewable Energy Technologies
2.2 Challenges of Renewable Energy Development
2.3 Alternative Low and Zero-Emission Technologies
3 Cost-Environmental Analysis of Renewable Technologies
3.1 The Capital Cost of Renewable Technologies
3.2 The O&M Cost of Renewable Energy Technologies
3.3 Climate and Health Impact of Energy Consumption
3.4 Levelized Cost of Electricity
4 Feasibility of 100% Renewable Energy Systems
5 Conclusion
References
Data Analytics Applications in Reducing the Emission Footprint of an Energy System
Contents
1 Introduction
2 Literature Review
3 Pillars of Data Analytics
3.1 Descriptive Analytics
3.2 Diagnostic Analytics
3.3 Predictive Analytics
3.4 Prescriptive Analytics
4 Conclusion
References
Logistics Processes Optimization with Regard to Sustainability Concerns
Contents
1 Introduction
2 Problem Statement
3 Problem Formulation
4 Problem Reformulation
5 Solution Methodology
6 A Case Study and Sensitivity Analysis
7 Conclusions
References
Data-Driven Techniques for Optimizing the Renewable Energy Systems Operations
Contents
1 Introduction
2 Fault Detection
3 Data-Driven Fault Detection Techniques Used in Wind Turbines
3.1 Types of Faults in the Wind Turbine System
4 Data-Driven Fault Detection Techniques Used in Photovoltaic Systems
4.1 Artificial Neural Network Based Fault Detection
4.2 Recurrent Neural Network Based Fault Detection
4.3 Convolutional Neural Network Based Fault Detection
4.4 Machine Learning-Based Fault Detection
5 Data-Driven Fault Detection Techniques Used in Hydroelectric Systems
6 Data-Driven Fault Detection Techniques Used in Geothermal Power Plants
7 The Optimal Placement to Establish Renewable Energy Sources
7.1 Data-Driven Techniques for Optimal Placement and Proper Sizing of Wind Turbine Systems
7.2 Data-Driven Techniques for Optimal Placement and Proper Sizing of PV Systems
8 The Accurate Prediction of the Power of the Renewable Energy Systems
8.1 Wind Turbines Energy
8.2 Photovoltaic Energy
8.3 Forecasting Approaches
8.4 Historical Data
9 Conclusion
10 Cross-References
References
Applying Optimization Techniques to Develop a Renewable Energy Supply Map
Contents
1 Introduction
2 Quasi-Dynamic and Renewable Generation Time Pattern
2.1 The Effect of Renewable Resources
2.2 Optimal Operation Summary
3 Optimal Operation of the Real Grid-Connected RES in Iran
3.1 Load Flow
3.2 Short Circuit
3.3 Harmonic Load Flow
3.4 Transient Stability
4 Conclusion
References
Sustainable Process Intensification for Biomass Valorization
Contents
1 Introduction
2 Biomass Pretreatment
3 Separation of Bioproducts
4 Biomass Valorization Using Process Intensification
4.1 Sustaining Molecular Functioning Structures
4.2 Utilizing Compound Property
4.3 Combining Phenomena and Process Integration
5 Systematic Exploration of Intensification Design Space
5.1 Fast and Accurate Property Prediction
5.2 Performance Evaluation of Process Intensification
5.3 Representation and Generation of Intensification Pathways
6 Conclusion
References
Index

Recommend Papers

Handbook of Electrical Energy Systems

592 57 9MB Read more

Smart Grid Technologies in Electric Systems for Renewable Energy

The book includes six chapters in which represented integrated Smart Grid systems for coordination of production and con

255 39 12MB Read more

SMART Automatics and Energy: Proceedings of SMART-ICAE 2021 (Smart Innovation, Systems and Technologies, 272) 9811687587, 9789811687587

This book gathers selected papers presented at the International Conference on SMART Automatics and Energy (SMART-ICAE 2

125 69 22MB Read more

IoT and Analytics in Renewable Energy Systems, Volume I: Sustainable Smart Grids & Renewable Energy Systems 9781032362816, 9781032362823, 9781003331117

Smart grid technologies include sensing and measurement technologies, advanced components aided with communications and

209 107 21MB Read more

Handbook of Optimization in Electric Power Distribution Systems (Energy Systems) 3030361144, 9783030361143

This handbook gathers state-of-the-art research on optimization problems in power distribution systems, covering classic

110 46 12MB Read more

Design of Smart Power Grid Renewable Energy Systems [3rd ed.] 1119573327, 9781119573326

The Updated Third Edition Provides a Systems Approach to Sustainable Green Energy Production and Contains Analytical Too

1,231 86 6MB Read more

Handbook of Renewable Energy 9783642394874

215 93 3MB Read more

Smart Grid and Sustainable Energy 1774695278, 9781774695272

This book covers different topics from smart grids and sustainable energy, including: automation and control of smart gr

168 103 42MB Read more

Intelligent Paradigms for Smart Grid and Renewable Energy Systems [1st ed.] 9789811599675, 9789811599682

This book addresses and disseminates state-of-the-art research and development in the applications of intelligent techni

604 99 17MB Read more

Blockchain and Artificial Intelligence Technologies for Smart Energy Systems [1 ed.] 0367771276, 9780367771270

Present energy systems are undergoing a radical transformation, driven by the urgent need to address the climate change

112 87 39MB Read more

Handbook of Smart Energy Systems
9783030979393, 9783030979409

Author / Uploaded
Michel Fathi
Enrico Zio
Panos M. Pardalos

0 0 0
Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up

File loading please wait...

Citation preview

Michel Fathi Enrico Zio Panos M. Pardalos Editors

Handbook of Smart Energy Systems

Handbook of Smart Energy Systems

Michel Fathi • Enrico Zio • Panos M. Pardalos Editors

Handbook of Smart Energy Systems With 1329 Figures and 366 Tables

Editors Michel Fathi Information Technology and Decision Sciences University of North Texas Denton, TX, USA

Enrico Zio Centre for Research on Risks and Crises (CRC) Mines Paris-PSL University Sophia Antipolis, France

Panos M. Pardalos Industrial & Systems Engineering University of Florida Gainesville, FL, USA

ISBN 978-3-030-97939-3 ISBN 978-3-030-97940-9 (eBook) https://doi.org/10.1007/978-3-030-97940-9 © Springer Nature Switzerland AG 2023 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG. The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

Considering the importance of energy production, transmission, and distribution systems for the current and future states of human living, while coping with climate change, this handbook focuses on providing a suite of solutions enabling energy access to industry development and the general world population in reliable, secure, and sustainable ways, by improving the performance of energy systems. The book consists of four parts. • Reliability Enhancement of Energy Systems This part describes the developments in reliability engineering of energy systems including the reliability of the elements in an energy distribution system, reliability of data collection procedures, reliability of the logistics for every system supply, etc. The major topics covered are as follows: • Importance of data reliability in energy demand forecast • Influences of data collection and analysis on the reliability assessment of energy systems • Reliability analysis of offshore renewable energy facilities • Influences of mobile data on energy systems reliability • Big data applications for improving the reliability of energy systems • Applications of data analytics in reliable energy logistics processes • Intelligent Development of Energy Systems Intelligent energy systems for fast and rapid decision-making, with online monitoring, are needed for the development of distributed complex smart grids under multiple resources planning. This part introduces advanced development and usage of new technologies in smart energy systems and the application of smart energy management for industrial development. The major topics covered are as follows: • Smart grids • Future energy system analysis v

vi

• • • • • •

Preface

Energy data management and utilization Energy efficiency for smart manufacturing Green production Smart energy aware systems Integration of renewable energy sources Smart network control of distributed energy systems

• Simulation and Optimization of Energy Systems Operations research methods and search algorithms can be used to optimize the performance of energy systems and the ways energy can be distributed to consumers. Methodologies such as Lagrangian relaxation, game theory, stochastic programming, multi-objective optimization, heuristics, and metaheuristics are methods that can be used to optimize the performance measures of an energy system, such as cost, customer satisfaction, delivery time, and balanced supply-demand. As it is not always possible to use real system data or best a system for observing its response to certain conditions, simulation is a common methodology to observe the anticipated behavior of a system in predefined conditions. Applying simulation and optimization to observe and improve the performance of an energy system is the focus of this part. The major topics covered are as follows: • • • • • • • • • • •

Data analysis applications for optimizing smart grid systems Data analysis applications for optimal integration of energy supply chains Optimizing energy mobile data collection networks Developing energy demand forecasting methods Data-driven energy waste minimization in energy distribution networks Applications of data envelopment analysis (DEA) for optimizing energy consumption Multiple-criteria decision making (MCDM) applications for optimizing multiobjectives energy system performance Using simulation to analyze the performance of an energy distribution center Novel applications of simulation to optimize an energy system Simulation applications for analyzing the trade-off between climate change and energy consumption Simulation applications for analyzing the reliability of an energy system

• Sustainable Development of Energy Systems Global warming and negative environmental experiences have become a major concern of the twenty-first century. Most of the negative impacts can be mitigated by changing energy consumption practices. This part describes the methods which can reduce the production of emissions by transforming the industrial and urban consumption of fossil fuels to renewable energies consumption.

Preface

vii

The major topics covered are as follows: • Cost-environment trade-off for using renewable energy in an energy system • Data analytics applications for reducing the emission footprint of an energy system • Logistics processes optimization with regards to sustainability concerns • Influences of waste management practices on energy emission footprints • Data-driven techniques for optimizing renewable energy systems operations • Applying optimization techniques to develop a renewable energy supply map • Novel applications of data-driven techniques for moving from using traditional energy to renewable energy • Constructing renewable energy systems using big data applications The main intention and goal of this book has been to bring leading experts in this most important area of smart energy systems to present their novel models and their applications. We hope that this book will prove useful to researchers, students, and engineers in the different domains of smart energy systems, and encourage them to undertake research and development in this exciting and practically important field. We want to thank all the authors involved in this project for their exceptionally valuable contributions. We also want to thank the reviewers who have helped us to review and improve several chapters of this book. In our journey toward accomplishing this book, we have had great support from conferences (The International Conference on System Reliability and Safety (ICSRS), The European Safety and Reliability Conference (ESREL), F&R Energy conference), societies (European Safety and Reliability Association, National Science Foundation, United States Department of Energy, IEEE Power & Energy Society, IEEE Reliability Society, American Energy Society, The International Society of Global Optimization, Institute for Operations Research and the Management Sciences, Decision Sciences Institute, Production and Operations Management, Texas A&M Energy Institute), editorial boards of journals (Operations Research, European Journal of Operational Research, Operations Research Forum, Reliability Engineering and System Safety, Energy Systems, Journal of Risk and Reliability, Journal of Risk and Uncertainty), and universities (University of North Texas, Politecnico di Milano, MINES Paris – PSL, University of Florida). Denton, USA Sophia Antipolis, France Gainesville, USA June 2023

Michel Fathi Enrico Zio Panos M. Pardalos

Contents

Volume 1 Part I Reliability Enhancement of Energy Systems . . . . . . . . . . . . . . . . The Need for Self-Sufficiency and Integrated Water and Energy Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ana R. C. Silva, Ricardo M. Silva, Gerardo J. Osorio, Fernando Charrua-Santos, and Antonio Espirito-Santo

1

3

Rethinking Renewable Energy Development in the Republic of Kazakhstan from the Perspectives of International Relations . . . . . . . . . Ka Wai Christopher Hor

27

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jun Zheng, Guchuan Zhu, and Meng Li

47

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dongsu Kim, Heejin Cho, and Pedro J. Mago

77

Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reza Gharoie Ahangar and Hani Gharavi

101

Virtual Power Plants and Integrated Energy System: Current Status and Future Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sambeet Mishra, Chiara Bordin, Madis Leinakse, Fushuan Wen, Robert J. Howlett, and Ivo Palu Reliability Analysis of Smart Grids Using Formal Methods . . . . . . . . . . . Mohamed Abdelghany and Sofiène Tahar

115

147

ix

x

Contents

Multi-objective Reliability-Based Design Optimization of Shell-and-Tube Heat Exchangers Using Combined Subset Simulation Method and Naive Bayes Algorithm . . . . . . . . . . . . . . . . . . . . . Sima Ohadi, Jafar Jafari-Asl, Oscar D. Lara Montaño, and Naser Safaeian Hamzehkolaei Key Technologies for the Energy Internet . . . . . . . . . . . . . . . . . . . . . . . . . . Hafiz Majid Hussain, Arun Narayanan, and Pedro H. J. Nardelli

165

183

How to Simulate If We Only Have Partial Information but We Want Reliable Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vladik Kreinovich and Olga Kosheleva

195

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Omar Kittaneh

215

Bayesian Network for Composite Power Systems Using Hybrid Mutual Information Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tahereh Daemi, Mohammad Reza Salehizadeh, and Miadreza Shafie-khah

247

Optimal Management of Smart Home Appliances Considering Stochastic Behavior of Wind Turbine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Masoud Alilou, Hossein Shayeghi, and Behrouz Tousi

267

Novel Data-Driven Methods for Evaluating Demand Response Programs in a Smart Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lihui Bai and Arnab Roy

287

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marios Tsioufis and Thomas A. Alexopoulos

307

Recent Developments in the Smart Energy Systems . . . . . . . . . . . . . . . . . Adil Wazeer and Apurba Das

327

Constructing Renewable Energy Systems Using Big Data Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nassim Sohaee

347

Control and Monitoring of Wind Farms Based on IoT Application for Energy Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohammad Niyazi and Adel Nazemi Babadi

361

Smart Grid and Resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zahra Zand, Muhammad Reza Ghahri, Soheil Majidi, Mostafa Eidiani, Morteza Azimi Nasab, and Mohammad Zand

367

Contents

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm . . . . . . . C. Anuradha, S. Vijayalakshmi, Viswanathan Ganesh, and P. S. Ramapraba Quantitative Methods for Data-Driven Next-Generation Resilience of Energy Systems and Their Supply Chains . . . . . . . . . . . . . . . . . . . . . . . Natasha J. Chrisandina, Shivam Vedant, Mahmoud M. El-Halwagi, Efstratios N. Pistikopoulos, and Eleftherios Iakovou

xi

391

409

Hybrid Attack Modeling for Critical Energy Infrastructure Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Maryna Zharikova, Volodymyr Sherstjuk, and Stefan Pickl

429

Social Cost-Benefit Analysis of Emission Norms: A Conceptual Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elizabeth Varsha Paul

449

Graphene-Based Wearable Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adil Wazeer, Apurba Das, Arijit Sinha, and Amit Karmakar

473

Piezoelectric Polymer Composites for Energy Harvesting . . . . . . . . . . . . Adil Wazeer, Apurba Das, Arijit Sinha, and Amit Karmakar

489

Practical Applications of Gaussian Process with Uncertainty Quantification and Sensitivity Analysis for Digital Twin for Accident-Tolerant Fuel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kazuma Kobayashi, Dinesh Kumar, Matthew Bonney, and Syed Alam Energy Consumption Management System for Smart Cities Using IoT Cloud . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ravi Varma Kumar Bevara An Overview of Drone Energy Consumption Factors and Models . . . . . Pedram Beigi, Mohammad Sadra Rajabi, and Sina Aghakhani

503

515 529

Big Data Applications for Improving the Reliability of the French Electricity Distribution Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jérémie Merigeault and Odilon Faivre

549

Structural Global Reliability Analysis of Floating Offshore Wind Turbines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jianbing Chen, Yupeng Song, and Jie Li

583

A Physics-Regularized Degradation Model for Cooling System Health Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiao Liu and Mohammadmahdi Hajiha

607

xii

Contents

Volume 2 Part II

Simulation and Optimization of Energy Systems . . . . . . . . . . .

625

A Stackelberg Game-Theoretic Model of Fee-and-Rebate Pricing in a Load-Reduction Emergency Demand Response Program . . . . . . . . . . . Sreerag Choorikkat, Yu-Ching Lee, and Hsin-Wei Hsu

627

Energy Simulation Optimization for Building Insulation Materials . . . . Salih Himmeto˘glu, Yılmaz Delice, Emel Kızılkaya Aydo˘gan, and Burak Uzal

653

Machine Learning for Building Energy Modeling . . . . . . . . . . . . . . . . . . . Debaditya Chakraborty and Hakan Ba¸sa˘gao˘glu

667

Machine Learning to Facilitate the Integration of Renewable Energies into the Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ahlem Aissa Berraies, Alexandros Tzanetos, and Maude Blondin

689

Energy Management of Smart Homes by Optimizing Energy Consumption Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Farid Momayezi, Kamyar Sabri-Laghaie, and Nader Ghaffarinasab

713

Applications of Machine Learning for Renewable Energy: Issues, Challenges, and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. A. Jabbar and Syed Saba Raoof

735

Multi-criteria Optimization for System Integration of Decentralized Off-Grid Hybrid Renewable Polygeneration . . . . . . . . . . . . . . . . . . . . . . . . Avishek Ray, Poulami Das, and Sudipta De

749

Optimal Risk Management of Electric Power Systems with CLOUD Simulation and Security Meter Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . Mehmet Sahinoglu

761

A Novel Mathematical Model for Infrastructure Planning of Dynamic Wireless Power Transfer Systems for Electric Vehicles . . . . . . Afshin Ghassemi, Laura Soares, Hao Wang, and Zhimin Xi

817

Application of Machine Learning for Energy-Efficient Buildings . . . . . . Indrasis Chakraborty, Aritra Dasgupta, Javier Rubio-Herrero, Sai Pushpak Nandanoori, Soumya Kundu, and Vikas Chandan

837

A Simulation-Based Optimization Framework Applied to Assess the Resilience of Energy Distribution Center . . . . . . . . . . . . . . . . . . . . . . . Elham Taghizadeh

859

Mathematical Modeling and Simulation Validation in Optimizing Multi-objective Energy Systems Performance . . . . . . . . . . . . . . . . . . . . . . . Mohammad Pazouki and Ali Bozorgi-Amiri

873

Contents

Machine Learning Methods for Estimating Energy Performance of Building Facade Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bahram Abediniangerabi and Mohsen Shahandashti Systematic Literature Review of Multi-criteria Decision-Making Applied to Energy Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vania Aparecida Rosario de Oliveira, Valerio Antonio Pamplona Salomon, Geraldo Cesar Rosario De Oliveira, Antonella Petrillo, and Sandra Miranda Neves Game Theory Modeling of Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . Ehsan Haghi Risk Assessment of Electric Power Generation Systems by Stochastic Simulation Using CLOUD Computing . . . . . . . . . . . . . . . . . . . . Mehmet Sahinoglu

xiii

895

927

941

961

Green Scheduling of a Complex Flexible Manufacturing Problem . . . . . 1021 Maryam Fetri and Seyed Habib A. Rahmati A Simulation-Based Framework for the Adequacy Assessment of Integrated Energy Systems Exposed to Climate Change . . . . . . . . . . . . . . 1045 Francesco Di Maio, Susanna Morelli, and Enrico Zio A Brief Guide on the Modeling of Green Vehicle Routing Problems . . . . 1081 Matheus Diógenes Andrade, Rafael Kendy Arakaki, and Fábio Luiz Usberti US Natural Gas Consumption Analysis via a Smart Time Series Approach Based on Multilayer Perceptron ANN Tuned by Meta-heuristic Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1101 Kianoush Nokhbeh Dehghan, Soulmaz Rahman Mohammadpour, and Seyed Habib A. Rahamti Simulation and Optimization of Energy Systems . . . . . . . . . . . . . . . . . . . . 1115 Mustafa F. Kaddoura Data Clustering Method for Probabilistic Power Flow in Microgrids . . . 1133 Seyed Farhad Zandrazavi, Alejandra Tabares Pozos, and John Fredy Franco Material Supply Network Optimization in the Energy and Utility Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1155 Haitao Li Application of Digital Twin Technology on Simulation and Optimization of Prime Movers in Energy Systems . . . . . . . . . . . . . . . . . . . 1175 Vili Panov and Samuel Cruz-Manzo

xiv

Contents

Digital Twin for Multi-criteria Decision-Making Framework to Accelerate Fuel Qualification for Accident Tolerant Fuel Concepts . . . . 1217 Kazuma Kobayashi, Brandon Bloss, Alexander Foutch, Brenden Kelly, Ayodeji Alajo, Carlos H. C. Giraldo, Dinesh Kumar, and Syed Alam Model Predictive Control and Distributed Optimization in Smart Grid Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1239 Philipp Braun, Lars Grüne, Christopher M. Kellett, and Karl Worthmann High-Performance Solar Cells by Machine Learning and Pareto Optimality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1265 Giovanni Nastasi, Vittorio Romano, and Giuseppe Nicosia Drone Delivery Systems and Energy Management: A Review and Future Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1273 Mohammad Sadra Rajabi, Pedram Beigi, and Sina Aghakhani Applications of Machine Learning in the Planning of Electric Vehicle Charging Stations and Charging Infrastructure: A Review . . . . 1293 Bhagyashree Panda, Mohammad Sadra Rajabi, and Alimohammad Rajaee Surrogate Modeling-Driven Physics-Informed Multi-fidelity Kriging for the Prediction of Accident-Tolerant Fuel Properties . . . . . . . 1313 Kazuma Kobayashi, Shoaib Usman, Carlos Castano, Ayodeji Alajo, Dinesh Kumar, and Syed Alam Robust Design Optimization Method for Engineering System . . . . . . . . . 1325 Richa Verma, Dinesh Kumar, Kazuma Kobayashi, and Syed Alam Data-Driven Multi-scale Modeling and Robust Optimization of Composite Structure with Uncertainty Quantification . . . . . . . . . . . . . . . 1333 Kazuma Kobayashi, Shoaib Usman, Carlos Castano, Ayodeji Alajo, Dinesh Kumar, Susmita Naskar, and Syed Alam Data Analysis Applications in Optimizing the Smart Grid System . . . . . 1345 Nikolay Belyaev, Nikolay Korovkin, Vladimir Chudny, and Olga Sokolova Data Analysis Applications in the Optimal Integration of Energy Supply Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1377 Sai Tejaswini Thumpala Developing Energy Demand Forecasting Methods . . . . . . . . . . . . . . . . . . . 1393 Willian Y. Takano and Eduardo N. Asada Data-Driven Energy Waste Minimization at Energy Distribution Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1413 Babak Aslani

Contents

xv

Applications of Data Envelopment Analysis (DEA) for Optimizing Energy Consumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1435 Zahra Mohtashami, Majid Khedmati, and Kourosh Eshghi Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-objective Energy System Performance . . . . . . . . . . . . . 1477 Ali Esmaeel Nezhad and Pedro H. J. Nardelli Simulation Applications in Analyzing the Trade-Off Between Climate Change and Energy Consumption . . . . . . . . . . . . . . . . . . . . . . . . . 1509 Amin Vahidi Analysis of Energy Transition Pertaining to the Future Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1535 Engin Deniz and Melih Soner Çelikta¸s Data Mining Applications in Smart Grid System (SGS) . . . . . . . . . . . . . . 1557 Mohammad Taghi Dehghan Nezhad and Mohammad mahdi Sarbishegi

Volume 3 Part III

Intelligent Development of Energy Systems . . . . . . . . . . . . . . . 1575

The Role of Blockchain and Cryptocurrency in Smart Grid: Renewable Energy Trading, System Security and Privacy Preservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1577 Wenlin Han Energy Harvesting for Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . 1589 Shirin Momen, Javad Nikoukar, Arsalan Hekmati, Soheil Majidi, Zahra Zand, Mohammad Zand, and Mostafa Eidiani Dynamic Bayesian Network Based Approach for Modeling and Assessing Resilience of Smart Grid System . . . . . . . . . . . . . . . . . . . . . . . . . 1613 Niamat Ullah Ibne Hossain and Chiranjibi Shah Application of Machine Learning in Occupant and Indoor Environment Behavior Modeling: Sensors, Methods, and Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1633 Farzad Dadras Javan, Hamed Khatam Bolouri Sangjoeei, Behzad Najafi, Alireza Haghighat Mamaghani, and Fabio Rinaldi Software Engineering Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . . 1659 Marco Aiello, Laura Fiorini, and Ilche Georgievski An Intelligent Decision Support System for an Integrated Energy Aware Production-Distribution Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1689 Soulmaz Rahman Mohammadpour and Seyed Habib A. Rahmati

xvi

Contents

Design and Operational Strategies for Grid-Connected Smart Home . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1709 Manimuthu Arunmozhi, S. Senthilmurugan, and Viswanathan Ganesh The Role of Data Collection, Storage, and Processing in the Intelligent Energy Systems of Tomorrow . . . . . . . . . . . . . . . . . . . . . . . . . . . 1733 Anatoli Paul Ulmeanu, Adrian Valentin Boicea, and Adrian Vulpe-Grigora¸si Adaptive Decentralized Under-frequency Load Shedding in Smart Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1757 Mohammad Hosein Fazaeli Photovoltaic Rooftops in Smart Energy Systems . . . . . . . . . . . . . . . . . . . . 1767 F. J. Muñoz-Rodríguez, G. Jiménez-Castillo, and C. Rus-Casas Smart Grid Resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1795 Hamed Badihi Smart Energy Systems, Infrastructure Financing, and the Wider Economy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1821 Hai Hong Trinh Multi-step Ahead Power Demand Forecasting in Smart Grid . . . . . . . . . 1845 Ehsan Hajizadeh and Amin Hajizadeh Social Acceptance of Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . 1855 R. Ulusoy Yılmaz and U. Soyta¸s Edge Computing in Smart Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1875 Arthur Sousa de Sena, Mehar Ullah, and Pedro H. J. Nardelli Resilience of Smart Integrated Energy Systems . . . . . . . . . . . . . . . . . . . . . 1887 Davood Babazadeh, Payam Teimourzadeh Baboli, Michael Brand, Christoph Mayer, Christian Becker, and Sebastian Lehnhoff Smart Buildings in the IoT Era: Necessity, Challenges, and Opportunities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1915 Roohollah Heidary, Jubilee Prasad Rao, and Olivia J. Pinon Fischer Smart Grids in the IoT Era: Necessity, Challenges, and Opportunities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1937 Babak Aslani Governance of Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1955 Polinpapilinho F. Katina, Charles B. Keating, Enrico Zio, Marcelo Masera, and Adrian V. Gheorghe

Contents

xvii

A Hybrid Smart Neural Network Model for Short-Term Prediction of Energy Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1975 Kianoush Nokhbeh Dehghan, Seyed Habib A. Rahamti, and Soulmaz Rahman Mohammadpour Processing Smart Meter Data Using IoT, Edge Computing, and Big Data Analytics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1987 Mehar Ullah, Annika Wolff, and Pedro H. J. Nardelli Energy-Smart Transportation Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2003 Saba Sabet and Bilal Farooq Energy Efficient Mission Control of Unmanned Intelligent Swarm Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2025 Banu Kabakulak Building the Remote Surveying System of Energy Consumption in Maritime Transportation Using Internet of Things (IoT) Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2045 Tien Anh Tran AI for Electricity Market Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2059 Krishna Sathvik Mantripragada and Michel Fathi Predicting US Energy Consumption Utilizing Artificial Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2075 Mohammad Pasandidehpoor, João Mendes-Moreira, Soulmaz Rahman Mohammadpour, and Ricardo Teixeira Sousa A Survey of Cyber-physical Systems Applications (2017–2022) . . . . . . . . 2089 Nastaran Jadidi and Mohsen Varmazyar Using Artificial Intelligence for Nuclear Nonproliferation and Commercial Nuclear Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2119 Jordan Fox, James Eagan, Ayodeji Alajo, and Syed Alam Machine Learning and Artificial Intelligence-Driven Multi-scale Modeling for High Burnup Accident-Tolerant Fuels for Light Water-Based SMR Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2131 Shamim Hassan, Abid Hossain Khan, Richa Verma, Dinesh Kumar, Kazuma Kobayashi, Shoaib Usman, and Syed Alam Blockchain and Smart Grids: Opportunities, Open Issues, and Future Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2155 Seyed Mahdi Bohloul and Anjee Gorkhali A Blockchain-Based Smart Grid to Build Resilience Through Zero-Trust Cybersecurity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2177 Ava Hajian and Hsia-Ching Chang

xviii

Contents

Blockchain and Open Energy Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2197 Amirhossein Souhankar, Reza Hafezi, and Amir Nazemi Ashni Leveraging Industry 4.0: Deep Learning, Surrogate Model, and Transfer Learning with Uncertainty Quantification Incorporated into Digital Twin for Nuclear System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2217 M. Rahman, Abid Hossain Khan, Sayeed Anowar, Md Al-Imran, Richa Verma, Dinesh Kumar, Kazuma Kobayashi, and Syed Alam Using AI and Classical Controllers for Improving the Renewable Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2237 Nima Vaezi and Parastoo Poursoltani Uncertainty Quantification and Sensitivity Analysis for Digital Twin Enabling Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2265 Kazuma Kobayashi, Dinesh Kumar, Matthew Bonney, Souvik Chakraborty, Kyle Paaren, Shoaib Usman, and Syed Alam Wind Turbine Anomaly Detection Based on SCADA Data . . . . . . . . . . . . 2279 Francisco Bilendo, Hamed Badihi, and Ningyun Lu Future Energy System Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2303 Luca Galbusera Hybrid Threats to the European Union’s Energy Sector: An Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2329 Marina Alonso Villota, Etienne Willkomm, and Stefan Pickl Green Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2353 M. Enhessari and A. Salehabadi Integration of Renewable Energy Sources . . . . . . . . . . . . . . . . . . . . . . . . . . 2375 Mostafa Eidiani Integration of Renewable Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 2401 Gamze Mersin and Melih Soner Çelikta¸s

Volume 4 Part IV Sustainable Development of Energy Systems . . . . . . . . . . . . . 2425 Planning to Incorporate Energy Conservation Practices, Renewable Energy Production Systems, and Eco-friendly Building Design Practices to Support Sustainability in US Public Schools . . . . . . . . . . . . . 2427 Scott Warren, Scott Moran, and Kristen McGuffin

Contents

xix

An Integrated AI-Multiple Criteria Decision-Making Framework to Improve Sustainable Energy Planning in Manufacturing Systems: A Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2453 Aida Esmaeilidouki, Bryn J. Crawford, Amir Ardestani-Jaafari, and Abbas S. Milani Novel Integrated Membrane Auto-Thermal Reactors (NIMATRs) for Energy Efficiency and Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . 2473 S. S. E. H. Elnashaie and Elham El Zanati Environment Trade-Off in Using Renewable Energy in Oil-Rich Countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2505 Madjid Abbaspour and Fereshteh Abbasizade Data Analytics Applications in the Energy Systems Concerning Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2531 Fariba Bagherzadeh, Hume Winzar, and Masud Behnia Challenges of Smart Grid Technology Deployment in Developing Countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2551 Amir Ghorbani and Kiarash Fartash Cost-Sustainability Trade-Off Solutions for the Optimal Planning of Local Integrated Energy Systems from Nanogrids to Communities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2573 Marialaura Di Somma, Giorgio Graditi, and Bing Yan A Detailed Analysis of the Barriers of Using Renewable Energies and Their Roles in Sustainable Development in Iran . . . . . . . . . . . . . . . . . 2607 Mohammad Ghiasi, Moslem Dehghani, Taher Niknam, Pierluigi Siano, and Hassan Haes Alhelou Proposing a New Hedging Strategy Based on Considering the Efficiency of Energy Markets in Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2631 Ali Fereydooni and Ehsan Hajizadeh Analyzing the Solar Energy System Investment in Turkey . . . . . . . . . . . . 2653 ˙ Babek Erdebilli, Rabia Nur Can, and Ibrahim Yilmaz The Influence of Circular Economy in Renewable Energy Systems: EoL Solar Panel Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2685 Seyedmohammad Mousavian, Sajjad Mahmoudi, and Masud Behnia Integration Strategies of Renewable Energy Sources in a Conventional Community . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2711 Viswanathan Ganesh, S. Senthilmurugan, Rathinam Ananthanarayanan, Shiva Srenivasan Srinivasan, and N. R. S. Lakshanasri

xx

Contents

IEC-61850 Performance Evaluation in a 5G Cellular Network: UDP and TCP Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2729 Iurii Demidov, Dick Carrillo Melgarejo, Antti Pinomaa, Liana Ault, Jari Jolkkonen, and Kirsi Leppa Analysis of the Renewable Energy Generation Capability for Attending a National Renovation Fleet Through Ethanol-Cell Electric Vehicles in a South American Market . . . . . . . . . . . . . . . . . . . . . . 2763 Paulo Nocera Alves Junior, Isotilia Costa Melo, Fernando Toshio Okamura, and José Cesar Cruz Júnior Practical Application of the Decision-Making Grid (DMG) for Supporting Maintenance Strategy Decisions in a Small Hydroelectric Power Plant (SHPP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2787 Fernando Toshio Okamura, Paulo Nocera Alves Junior, José César Cruz Júnior, and Isotilia Costa Melo A Good Practice in Urban Energetics in a Hungarian Small Town, Kaposvar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2809 Gyöngyi Bánkuti and Zsuzsanna Zanatyné Uitz Digital Twin and Artificial Intelligence Incorporated with Surrogate Modeling for Hybrid and Sustainable Energy Systems . . . . . . . . . . . . . . . 2837 Abid Hossain Khan, Salauddin Omar, Nadia Mushtary, Richa Verma, Dinesh Kumar, and Syed Alam Evaluation of Energy Efficiency for Smart Manufacturing: Applications and Future Scopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2861 Chandan Kumar Jha The Role of Big Data in Building Renewable Energy Systems: The Case of Emerging Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2867 Howard Jean-Denis The Relation Among Ownership, Environmental, Social, Governance (ESG), and Corporate Social Responsibility (CSR) in China . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2877 Yichi Zhang and Kabir Hassan Mohammad Energy Semantic Data Management and Utilization in Smart Grid Networks with Focus on Circular Economy . . . . . . . . . . . . . . . . . . . . . . . . 2899 Mohammad Yaser Mofatteh, Amir Pirayesh, and Omid Fatahi Valilai Analyzing the Effect of Solar, Wind, Nuclear, and Total Renewable Energy Intensities on Economic Growth in BRICS . . . . . . . . . . . . . . . . . . 2923 Elizabeth Varsha Paul and Malay Kumar Patra

Contents

xxi

Supply Chain Resilience Along with Assessment of Sustainable Development: Experimental Data for the Integration of Supply Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2951 J. Kiarash Sadeghi R. and Saroj Karki The Food-Energy-Water Nexus in Sustainable Energy Systems Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2967 Marcello Di Martino, R. Cory Allen, and Efstratios N. Pistikopoulos Hydrogen-Based Dense Energy Carriers in Energy Transition Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2987 Rahul Kakodkar, Swaminathan Sundar, and Efstratios N. Pistikopoulos A Q-Learning-Based Demand Response Algorithm for Industrial Processes with Operational Flexibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3009 Farzaneh Karami, Manu Lahariya, and Guillaume Crevecoeur Indicators Framework for Sustainability and Circular Economy Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3027 Noushin Bagheri and Fouad Ben Abdelaziz Optimal Balancing of Wind Parks with Virtual Power Plants in the Market Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3047 Vadim Omelˇcenko and Valery Manokhin Recent Trends in Sustainable Supply-Chain Optimization . . . . . . . . . . . . 3095 Panagiotis Karakostas and Angelo Sifaleras Energy Technology RD&D Budgets, Environmental Sustainability, and Energy Transition: A Review of Emerging Trends, Policies, and Pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3119 Hai Hong Trinh Green Synthesis of Carbon Nanomaterials . . . . . . . . . . . . . . . . . . . . . . . . . 3143 Adil Wazeer, Apurba Das, Arijit Sinha, and Amit Karmakar Graphene and Graphene-Based Sustainable Composites . . . . . . . . . . . . . 3161 Adil Wazeer, Apurba Das, Arijit Sinha, and Amit Karmakar Social Innovation Approaches for Sustainable Living: A Review on Research Methods and Social Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3175 Chandan Kumar Jha and Amit Sachan Internet of Things Value Creation for Sustainable Energy . . . . . . . . . . . . 3181 Sara Memarian Esfahani and Hossein Mohit Big Data Analytics in Smart Energy Systems and Networks: A Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3201 Morteza Ghasemi and Mohammad Sadra Rajabi

xxii

Contents

Sustainability in a Circular Economy Context: Investigation of the Market-Based Instruments Potential for Energy Efficiency Improvement (Case Study: Iran) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3217 Fereshteh Abbasizade and Yousef Bahmani Wireless Charging of Electric Vehicles Through Pavements . . . . . . . . . . . 3235 Bhagyashree Panda, Faeze Momeni Rad, and Mohammad Sadra Rajabi Cost-Environment Trade-Off in Using Renewable Energy in an Energy System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3261 Ehsan Haghi and Grace Anne Thompson Data Analytics Applications in Reducing the Emission Footprint of an Energy System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3281 Vasudev Trivedi and Michel Fathi Logistics Processes Optimization with Regard to Sustainability Concerns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3295 Ardavan Babaei, Majid Khedmati, and Mohammad Reza Akbari Jokar Data-Driven Techniques for Optimizing the Renewable Energy Systems Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3317 Parastou Fahim and Nima Vaezi Applying Optimization Techniques to Develop a Renewable Energy Supply Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3339 Mostafa Eidiani Sustainable Process Intensification for Biomass Valorization . . . . . . . . . . 3355 Jianping Li Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3375

About the Editors

Michel Fathi received the B.S. and M.S. degrees from Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran, in 2006 and 2008, respectively, and the Ph.D. degree from the Iran University of Science and Technology, Tehran, Iran, in 2013. He was a Visiting Scholar at the University of Florida, USA, National Tsing Hua University, Taiwan, and Tecnológico de Monterrey, Mexico. He is currently an Assistant Professor at the University of North Texas, USA. He has authored or co-authored articles in journals such as Technometrics, IEEE Transactions on Automation Science and Engineering, and IEEE Transactions on Industrial Informatics. His research interests include Operations Research, Management Science, Data Science, AI in Business, Cybersecurity and Information Systems, Energy Systems, Healthcare, and Social Goods. Dr. Fathi has received three Postdoctoral Fellowships at Ecole Centrale Paris, France, Ghent University, Belgium, and Mississippi State University, USA. He is the Corresponding Editor of the textbooks Large Scale Optimization in Supply Chains and Smart Manufacturing: Theory and Applications and Optimization in Large Scale Problems: Industry 4.0 and Society 5.0 Applications. Dr. Fathi is a member of the Institute for Operations Research and the Management Sciences, Production and Operations Management Society, and Decision Sciences Institute and serves as an Associate Editor for journals such as AI in Business, Energy Systems, and Operations Research Forum. He is also currently editing the Handbook of Smart Energy Systems.

xxiii

xxiv

About the Editors

Enrico Zio received the M.Sc. degree in Nuclear Engineering from Politecnico di Milano in 1991 and in Mechanical Engineering from UCLA in 1995, and the Ph.D. degree in Nuclear Engineering from Politecnico di Milano and in Probabilistic Risk Assessment at MIT in 1996 and 1998, respectively. He is currently a Full Professor at the Centre for research on Risk and Crises (CRC) of Ecole de Mines, ParisTech, PSL University, France, Full Professor and President of the Alumni Association at Politecnico di Milano, Italy, CoDirector of the Center for REliability and Safety of Critical Infrastructures (CRESCI) and the Sino-French Laboratory of Risk Science and Engineering (RISE), at Beihang University, Beijing, China, and Adjoint and Guest Professor in several prestigious universities around the world. He is IEEE and Sigma Xi Distinguished Lecturer. In 2020, he has been awarded the prestigious Humboldt Research Award from the Alexander von Humboldt Foundation in Germany (https://www.humboldtfoundation.de/web/home.html), one of the world’s most prestigious research awards across all scientific disciplines. In 2021, he has been appointed as Member of the Board Committee of the International Joint Research Center for Resilient Infrastructure (ICRI) 2021–2026; 4TU.Resilience Ambassador by the 4TU Centre for Resilience Engineering and its backbone – the 4TU-programme DeSIRE (Designing Systems for informed Resilience Engineering), a strategic capacity building research program of the four Dutch Technical Universities; and Fellow of the Prognostics and Health Management Society. He is a world recognized scientist in the area of reliability-centered, condition-based, and predictive maintenance. His research focuses on the modeling of the failurerepair-maintenance behavior of components and complex systems, for the analysis of their reliability, maintainability, prognostics, safety, vulnerability, resilience, and security characteristics, and on the development and use of Monte Carlo simulation methods, artificial intelligence techniques, and optimization heuristics. He is author and co-author of 7 books and more than 500 papers on international journals, Chairman and Co-Chairman of several international Conferences, Associate Editor of several international journals, and referee of more than 20 international journals.

About the Editors

xxv

Panos M. Pardalos was born in Drosato (Mezilo) Argitheas in 1954 and graduated from Athens University (Department of Mathematics). He received his Ph.D. (Computer and Information Sciences) from the University of Minnesota. He is a Distinguished Emeritus Professor in the Department of Industrial and Systems Engineering at the University of Florida, and an affiliated faculty of Biomedical Engineering and Computer Science & Information & Engineering departments. Prof. Pardalos is a world-renowned leader in Global Optimization, Mathematical Modeling, Energy Systems, Financial applications, and Data Sciences. He is a Fellow of AAAS, AAIA, AIMBE, EUROPT, and INFORMS and was awarded the 2013 Constantin Caratheodory Prize of the International Society of Global Optimization. In addition, Panos Pardalos has been awarded the 2013 EURO Gold Medal prize bestowed by the Association for European Operational Research Societies. This medal is the preeminent European award given to Operations Research (OR) professionals for “scientific contributions that stand the test of time.” Prof. Pardalos has been awarded a prestigious Humboldt Research Award (2018–2019). The Humboldt Research Award is granted in recognition of a researcher’s entire achievements to date – fundamental discoveries, new theories, insights that have had significant impact on their discipline. Prof. Pardalos is also a Member of several Academies of Sciences, and he holds several honorary Ph.D. degrees and affiliations. He is the Founding Editor of Optimization Letters and Energy Systems, and Co-Founder of the International Journal of Global Optimization, Computational Management Science, and Springer Nature Operations Research Forum. He has published over 600 journal papers and edited/authored over 200 books. He is one of the most cited authors and has graduated 71 Ph.D. students so far. Details can be found at www.ise.ufl.edu/pardalos. Prof. Pardalos has lectured and given invited keynote addresses worldwide in countries including Austria, Australia, Azerbaijan, Belgium, Brazil, Canada, Chile, China, Cyprus, Czech Republic, Denmark, Egypt, England, France, Finland, Germany, Greece, Holland, Hong Kong, Hungary, Iceland,

xxvi

About the Editors

Ireland, Italy, Japan, Lithuania, Mexico, Mongolia, Montenegro, New Zealand, Norway, Peru, Portugal, Russia, South Korea, Singapore, Serbia, South Africa, Spain, Sweden, Switzerland, Taiwan, Turkey, Ukraine, United Arab Emirates, and the USA.

Contributors

Fereshteh Abbasizade Sharif Energy, Water and Environment Research Institute (SEWERI), Sharif University of Technology (SUT), Tehran, Iran Madjid Abbaspour School of Mechanical Engineering, Sharif University of Technology (SUT), Tehran, Iran Sharif Energy Water Environment Institute (SEWEI), Sharif University of Technology, Tehran, Iran Mohamed Abdelghany Department of Electrical and Computer Engineering, Concordia University, Montreal, QC, Canada Bahram Abediniangerabi Department of Engineering and Technology, Texas A&M University–Commerce, Commerce, TX, USA Sina Aghakhani Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran Marco Aiello Service Computing Department, IAAS, University of Stuttgart, Stuttgart, Germany Mohammad Reza Akbari Jokar Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran Ayodeji Alajo Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Syed Alam Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Thomas A. Alexopoulos University of Peloponnese, Tripoli, Greece International Centre for Economic Analysis, Waterloo, Canada Masoud Alilou Department of Electrical Engineering, Urmia University, Urmia, Iran Md Al-Imran Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh xxvii

xxviii

Contributors

R. Cory Allen Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Energy Institute, Texas A&M University, College Station, TX, USA Marina Alonso Villota Universität der Bundeswehr München, Germany

Neubiberg,

Paulo Nocera Alves Junior Escuela de Ingeniería de Coquimbo (EIC), Universidad Católica del Norte (UCN), Coquimbo, Chile Amit Sachan IIM, Ranchi, India Rathinam Ananthanarayanan SRM Institute of Science and Technology, Kattankulathur, Chennai, India Matheus Diógenes Andrade Institute of Computing, University of Campinas, Campinas, Brazil Sayeed Anowar Department of Industrial and Production Engineering, Jashore University of Science and Technology, Jashore, Bangladesh C. Anuradha Department of Electrical and Electronics Engineering, SRM Institute of Science and Technology, Chennai, India Rafael Kendy Arakaki Institute of Computing, University of Campinas, Campinas, Brazil Amir Ardestani-Jaafari Faculty of Management, University of British Columbia, Kelowna, BC, Canada Manimuthu Arunmozhi Energy Research Institute (ERIAN), Nanyang Technological University, Singapore, Singapore Eduardo N. Asada São Carlos School of Engineering, University of São Paulo, São Carlos, Brazil Babak Aslani Department of Systems Engineering and Operations Research, George Mason University, Fairfax, VA, USA Liana Ault Nokia, London, UK Adel Nazemi Babadi KNT University, Tehran, Iran Ardavan Babaei Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran Davood Babazadeh Institute of Electrical Power and Energy Technology, Hamburg University of Technology, Hamburg, Germany Hamed Badihi College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, China Noushin Bagheri NEOMA Business School, Rouen, France

Contributors

xxix

Fariba Bagherzadeh Business School, Macquarie University, Sydney, NSW, Australia Yousef Bahmani Energy and Environment Researcher, Tehran, Iran Lihui Bai Department of Industrial Engineering, University of Louisville, Louisville, KY, USA Gyöngyi Bánkuti Institute of Mathematics and Basic Science, MATE Hungarian University of Agriculture and Life Science, Gödöll˝o, Kaposvár, Hungary Hakan Ba¸sa˘gao˘glu Edwards Aquifer Authority, San Antonio, TX, USA Christian Becker Institute of Electrical Power and Energy Technology, Hamburg University of Technology, Hamburg, Germany Masud Behnia Business School, Macquarie University, Sydney, NSW, Australia Macquarie Graduate School of Management, Macquarie University, North Ryde, NSW, Australia Pedram Beigi Department of Civil Engineering, Sharif University of Technology, Tehran, Iran Nikolay Belyaev Federal State Budgetary Organization “Russian Energy Agency” (REA) by the Ministry of Energy of the Russian Federation, Moscow, Russia Fouad Ben Abdelaziz NEOMA Business School, Rouen, France Ahlem Aissa Berraies Faculty of Engineering, Multiobjective Optimization REsearch Lab (MORE Lab), Department of Electrical Engineering & Computer Engineering, Université de Sherbrooke, Sherbrooke, QC, Canada Ravi Varma Kumar Bevara University of North Texas, Denton, TX, USA Francisco Bilendo College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, China Maude Blondin Faculty of Engineering, Multiobjective Optimization REsearch Lab (MORE Lab), Department of Electrical Engineering & Computer Engineering, Université de Sherbrooke, Sherbrooke, QC, Canada Brandon Bloss Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Seyed Mahdi Bohloul Old Dominion University, Norfolk, VA, USA Adrian Valentin Boicea Department of Electrical Power Systems, Polytechnic University of Bucharest, Bucharest, Romania Matthew Bonney Department of Mechanical Engineering, University of Sheffield, Sheffield, UK Chiara Bordin The Arctic University of Norway, Tromsø, Norway

xxx

Contributors

Ali Bozorgi-Amiri School of Industrial Engineering, College of Engineering, University of Tehran, Tehran, Iran Michael Brand Energy Division, OFFIS – Institute for Information Technology, Oldenburg, Germany Philipp Braun School of Engineering, Australian National University, Canberra, Australia Rabia Nur Can Department of Industrial Engineering, TOBB University of Economics and Technology, Ankara, Turkey Carlos Castano Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Melih Soner Çelikta¸s Solar Energy Institute, Ege University Bornova, Izmir, Turkey Debaditya Chakraborty University of Texas at San Antonio, San Antonio, TX, USA Indrasis Chakraborty Center for Applied Scientific Computing (CASC), Lawrence Livermore National Laboratory, Livermore, CA, USA Souvik Chakraborty Department of Applied Mechanics, Indian Institute of Technology Delhi, New Delhi, India Vikas Chandan Pacific Northwest National Laboratory, Richland, WA, USA Hsia-Ching Chang Department of Information Science, College of Information, University of North Texas, Denton, TX, USA Fernando Charrua-Santos C-MAST, University of Beira Interior Covilha, Portugal Jianbing Chen State Key Laboratory of Disaster Reduction in Civil Engineering & College of Civil Engineering, Tongji University, Shanghai, P.R. China Heejin Cho Department of Mechanical Engineering, Mississippi State University, Mississippi State, MS, USA Sreerag Choorikkat Department of Industrial Engineering and Engineering Management, National Tsing Hua University, Hsinchu, Taiwan Natasha J. Chrisandina Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Vladimir Chudny Peter the Great St. Petersburg Polytechnic University, St. Petersburg, Russia Isotilia Costa Melo Facultad de Ingeniería y Ciencias (FIC), Universidad Adolfo Ibáñez (UAI), Viña del Mar, Chile

Contributors

xxxi

Bryn J. Crawford School of Engineering, University of British Columbia, Kelowna, BC, Canada Guillaume Crevecoeur Department of Electromechanical, Systems and Metal Engineering, Ghent University, Ghent, Belgium José Cesar Cruz Júnior Federal University of São Carlos (UFSCar), Sorocaba, Brazil Samuel Cruz-Manzo School of Engineering, University of Lincoln, Lincoln, UK Farzad Dadras Javan Dipartimento di Energia, Politecnico di Milano, Milan, Italy Tahereh Daemi Department of Electrical Engineering, Yazd Branch, Islamic Azad University, Yazd, Iran Apurba Das Aerospace Engineering and Applied Mechanics Department, IIESTShibpur, Howrah, India Poulami Das Department of Computer Engineering, K.C. College of Engineering and Management Studies and Research, Mumbai, India Aritra Dasgupta New Jersey Institute of Technology, Newark, NJ, USA Sudipta De Department of Mechanical Engineering, Jadavpur University, Kolkata, India Geraldo Cesar Rosario De Oliveira Sao Paulo State University, Sao Paulo, Brazil Vania Aparecida Rosario de Oliveira Sao Paulo State University, Sao Paulo, Brazil Moslem Dehghani Department of Electrical and Electronic Engineering, Shiraz University of Technology, Shiraz, Iran Mohammad Taghi Dehghan Nezhad Sharif University, Tehran, Iran Yılmaz Delice Department of Management and Organization, Develi Vocational College, Kayseri University, Kayseri, Turkey Iurii Demidov School of Energy Systems, LUT University, Lappeenranta, Finland Engin Deniz Hanwha Q CELLS GmbH, Berlin, Germany Beuth University of Applied Sciences Berlin (Beuth Hochschule für Technik Berlin), Berlin, Germany Francesco Di Maio Department of Energy, Politecnico di Milano, Milano, Italy Marcello Di Martino Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Energy Institute, Texas A&M University, College Station, TX, USA

xxxii

Contributors

Marialaura Di Somma Department of Energy Technologies and Renewable Sources, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, ENEA, Rome, Italy James Eagan Department of Nuclear Engineering and Radiation Science, University of Science and Technology, Rolla, MO, USA Mostafa Eidiani Khorasan Institute of Higher Education, Mashhad, Iran Mahmoud M. El-Halwagi Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Energy Institute, Texas A&M University, College Station, TX, USA TEES Gas and Fuels Research Center, Texas A&M University, College Station, TX, USA S. S. E. H. Elnashaie Chemical Engineering Department, University of British Columbia (UBC), Vancouver, Canada Elham El Zanati Chemical Engineering and Pilot Plat Department (CEPPD), National Research Centre (NRC), Cairo, Egypt M. Enhessari Fachbereich Biologie, Chemie, Pharmazie, Institut für Chemie und Biochemie – Anorganische Chemie, Freie Universität Berlin, Berlin, Germany Babek Erdebilli Department of Industrial Engineering, Ankara Yıldırım Beyazıt University, Ankara, Turkey Kourosh Eshghi Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran Ali Esmaeel Nezhad Department of Electrical Engineering, School of Energy Systems, LUT University, Lappeenranta, Finland Aida Esmaeilidouki School of Engineering, University of British Columbia, Kelowna, BC, Canada Antonio Espirito-Santo C-MAST, University of Beira Interior, Covilha, Portugal Parastou Fahim Ferdowsi University of Mashhad, Mashhad, Iran Odilon Faivre Enedis, Paris, France Bilal Farooq Laboratory of Innovations in Transportation (LiTrans), Ryerson University, Toronto, ON, Canada Kiarash Fartash Institute for Science and Technology Studies, Shahid Beheshti University, Tehran, Iran Omid Fatahi Valilai School of Business, Social & Decision Sciences, Jacobs University, Bremen, Germany

Contributors

xxxiii

Michel Fathi Information Technology and Decision Sciences, University of North Texas, Denton, TX, USA Mohammad Hosein Fazaeli Amirkabir University of Technology, Tehran, Iran Ali Fereydooni Department of Industrial Engineering and Management Systems, Amirkabir University of Technology (Tehran Polytechnics), Tehran, Iran Maryam Fetri Department of Industrial Engineering, Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran Laura Fiorini Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence, University of Groningen, Groningen, The Netherlands Alexander Foutch Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Jordan Fox Department of Nuclear Engineering and Radiation Science, University of Science and Technology, Rolla, MO, USA John Fredy Franco Department of Electrical Engineering, São Paulo State University, São Paulo, Brazil School of Energy Engineering, São Paulo State University, Rosana, Brazil Luca Galbusera European Commission, Joint Research Centre (JRC), Ispra, Italy Viswanathan Ganesh Department of Electrical Engineering, Chalmers University of Technology, Gothenburg, Sweden Department of Sustainable Electric Power Engineering and Electromobility, Chalmers University of Technology, Gothenburg, Sweden Ilche Georgievski Service Computing Department, IAAS, University of Stuttgart, Stuttgart, Germany Nader Ghaffarinasab Department of Industrial Engineering, University of Tabriz, Tabriz, Iran Muhammad Reza Ghahri Sharif University of Technology, Tehran, Iran Hani Gharavi Innovation & Planning Department, Current Network Team, Eirgrid PLC, Dublin, Ireland Reza Gharoie Ahangar Department of Information Technology and Decision Science, G. Brint Ryan College of Business, University of North Texas, Denton, TX, USA Morteza Ghasemi School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran

xxxiv

Contributors

Afshin Ghassemi Rutgers University, Piscataway, NJ, USA Adrian V. Gheorghe Old Dominion University, Norfolk, VA, USA Mohammad Ghiasi Department of Electrical and Electronic Engineering, Shiraz University of Technology, Shiraz, Iran Amir Ghorbani Institute for Science and Technology Studies, Shahid Beheshti University, Tehran, Iran Carlos H. C. Giraldo Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Anjee Gorkhali Susquehanna University, Selinsgrove, PA, USA Giorgio Graditi Department of Energy Technologies and Renewable Sources, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, ENEA, Rome, Italy Lars Grüne Chair of Applied Mathematics, University of Bayreuth, Bayreuth, Germany Hassan Haes Alhelou Department of Electrical Power Engineering, Tishreen University, Lattakia, Syria School of Electrical and Electronic Engineering, University College Dublin, Dublin, Ireland Reza Hafezi Science & Technology Futures Studies Department, National Research Institute for Science Policy (NRISP), Tehran, Iran Ehsan Haghi Community Energy Specialist, Musqueam Indian Band, Vancouver, BC, USA Musqueam Indian Band, Vancouver, BC, Canada Alireza Haghighat Mamaghani Department of Chemical Engineering, University of Waterloo, Waterloo, ON, Canada Ava Hajian Department of Information Science, College of Information, University of North Texas, Denton, TX, USA Mohammadmahdi Hajiha Department of Industrial Engineering, University of Arkansas, Fayetteville, AR, USA Amin Hajizadeh AAU Energy, Aalborg University, Esbjerg, Denmark Ehsan Hajizadeh Department of Industrial Engineering and Management Systems, Amirkabir University of Technology (Tehran Polytechnics), Tehran, Iran Wenlin Han Department of Computer Science, California State University, Fullerton, Fullerton, CA, USA

Contributors

xxxv

Shamim Hassan Institute of Nuclear Power Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh Roohollah Heidary Global Technology Connection Inc., Atlanta, GA, USA Arsalan Hekmati Revterra Co., Houston, TX, USA Salih Himmeto˘glu Department of Industrial Engineering, Engineering Faculty, Erciyes University, Kayseri, Turkey Ka Wai Christopher Hor Department of International Relations, Al-Farabi Kazakh National University, Almaty, Kazakhstan Niamat Ullah Ibne Hossain Department of Engineering Management College of Engineering and Computer Science, Arkansas State University, State University, USA Robert J Howlett KES International, Shoreham-by-sea, UK Hsin-Wei Hsu Department of Industrial and Systems Engineering, Chung Yuan Christian University, Taoyuan, Taiwan Hafiz Majid Hussain Lappeenrannan–Lahden teknillinen yliopisto LUT, Lappeenranta, Finland Eleftherios Iakovou Energy Institute, Texas A&M University, College Station, TX, USA Department of Engineering Technology and Industrial Distribution, Texas A&M University, College Station, TX, USA J. Mike Walker ’66 Department of Mechanical Engineering, Texas A&M University, College Station, TX, USA Mosbacher Institute of Trade, Economics and Public Policy, Bush School of Government and Public Service, Texas A&M University, College Station, TX, USA M. A. Jabbar Vardhaman College of Engineering, Hyderabad, India Nastaran Jadidi Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran Jafar Jafari-Asl Department of Civil Engineering, Faculty of Engineering, University of Sistan and Baluchestan, Zahedan, Iran Howard Jean-Denis Pepperdine University, Calabasas, CA, USA Chandan Kumar Jha Indian Institute of Management, Ranchi, Jharkhand, India Soil SoBD, Gurgaon, Gurgaon, India Indian Institute of Management, Ranchi/SOIL School of Business Design, Gurugram, Haryana, India

xxxvi

Contributors

G. Jiménez-Castillo Center for Advanced Studies in Earth Sciences, Energy and Environment (CEACTEMA), University of Jaen, Jaen, Spain Jari Jolkkonen Nokia, Espoo, Finland ˙ ˙ Banu Kabakulak Istanbul Bilgi University, Istanbul, Turkey Mustafa F. Kaddoura Department of Mechanical Engineering, University of Minnesota, Minneapolis, MN, USA Rahul Kakodkar Texas A&M Energy Institute, Texas AM University, College Station, TX, USA Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Panagiotis Karakostas Department of Applied Informatics, School of Information Sciences, University of Macedonia, Thessaloniki, Greece Farzaneh Karami Department of Electromechanical, Systems and Metal Engineering, Ghent University, Ghent, Belgium Department of Computer Science, KU Leuven, Ghent, Belgium Saroj Karki School of Mechanical, Industrial and Manufacturing Engineering, Oregon State University, Corvallis, OR, USA Amit Karmakar Mechanical Engineering Department, Jadavpur University, Kolkata, India Polinpapilinho F. Katina University of South Carolina Upstate, Spartanburg, SC, USA Charles B. Keating Old Dominion University, Norfolk, VA, USA Christopher M. Kellett School of Engineering, Australian National University, Canberra, Australia Brenden Kelly Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Abid Hossain Khan Institute of Nuclear Power Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh Hamed Khatam Bolouri Sangjoeei Dipartimento di Energia, Politecnico di Milano, Milan, Italy Majid Khedmati Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran Dongsu Kim Department of Architectural Engineering, Hanbat National University, Daejeon, South Korea

Contributors

xxxvii

Omar Kittaneh NSMTU- College of Engineering, Effat University, Jeddah, Saudi Arabia Kazuma Kobayashi Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Nikolay Korovkin Peter the Great St. Petersburg Polytechnic University, St. Petersburg, Russia Olga Kosheleva Department of Teacher Education, University of Texas at El Paso, El Paso, TX, USA Vladik Kreinovich Department of Computer Science, University of Texas at El Paso, El Paso, TX, USA Dinesh Kumar Department of Mechanical Engineering, University of Bristol, Bristol, UK Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Soumya Kundu Pacific Northwest National Laboratory, Richland, WA, USA Emel Kızılkaya Aydo˘gan Department of Industrial Engineering, Engineering Faculty, Erciyes University, Kayseri, Turkey Manu Lahariya IDLab – imec, Ghent University, Ghent, Belgium N. R. S Lakshanasri SRM Institute of Science and Technology, Kattankulathur, Chennai, India Oscar D. Lara Montaño Departamento de Ingeniería Química, División de Ciencias Naturales y Exactas, Campus Guanajuato, Universidad de Guanajuato, Guanajuato, Mexico Yu-Ching Lee Department of Industrial Engineering and Engineering Management, National Tsing Hua University, Hsinchu, Taiwan Sebastian Lehnhoff Energy Division, OFFIS – Institute for Information Technology, Oldenburg, Germany Madis Leinakse Tallinn University of Technology, Tallinn, Estonia Kirsi Leppa Nokia, Espoo, Finland Haitao Li Supply Chain and Analytics Department, College of Business Administration, University of Missouri, St. Louis, MO, USA Jianping Li Department of Chemical and Biological Engineering and DOE Great Lakes Bioenergy Research Center, University of Wisconsin-Madison, Madison, WI, USA Jie Li State Key Laboratory of Disaster Reduction in Civil Engineering & College of Civil Engineering, Tongji University, Shanghai, P.R. China

xxxviii

Contributors

Meng Li College of Big Data and Internet, Shenzhen Technology University, Shenzhen, Guangdong, China Xiao Liu Department of Industrial Engineering, University of Arkansas, Fayetteville, AR, USA Ningyun Lu College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, China Pedro J. Mago Department of Mechanical and Aerospace Engineering, West Virginia University, Morgantown, WV, USA Sajjad Mahmoudi Centre for Sustainable Materials Research and Technology, SMaRT@UNSW, School of Materials Science and Engineering, University of New South Wales, Sydney, NSW, Australia Soheil Majidi Research and Development Department, BLUE&P Group, Tehran, Iran Valery Manokhin Department of Computer Science, Royal Holloway, University of London, London, UK Krishna Sathvik Mantripragada University of North Texas, Information Technology and Computer Science, Denton, TX, USA Marcelo Masera European Commission – Joint Research Centre, The Netherlands

Petten,

Christoph Mayer Energy Division, OFFIS – Institute for Information Technology, Oldenburg, Germany Kristen McGuffin University of North Texas, Denton, TX, USA Dick Carrillo Melgarejo School of Energy Systems, LUT University, Lappeenranta, Finland Sara Memarian Esfahani Department of Information Technology and Decision Sciences, G. Brint Ryan College of Business, University of North Texas, Denton, TX, USA João Mendes-Moreira Institute for Systems and Computer Engineering (INESC TEC), Technology and Science, Faculty of Engineering, University of Porto, Porto, Portugal Jérémie Merigeault Enedis, Paris, France ˙ Gamze Mersin Enerjisa, Ata¸sehir/Istanbul, Turkey Solar Energy Institute, Ege University, Bornova, Turkey Abbas S. Milani School of Engineering, University of British Columbia, Kelowna, BC, Canada

Contributors

xxxix

Sambeet Mishra Tallinn University of Technology, Tallinn, Estonia Danish Technical University, Kongens Lyngby, Denmark Mohammad Yaser Mofatteh School of Business, Social & Decision Sciences, Jacobs University, Bremen, Germany Kabir Hassan Mohammad University of New Orleans, New Orleans, LA, USA Hossein Mohit Department of Information Technology and Decision Sciences, G. Brint Ryan College of Business, University of North Texas, Denton, TX, USA Zahra Mohtashami Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran Farid Momayezi Faculty of Industrial Engineering, Urmia University of Technology, Urmia, Iran Shirin Momen Department of Electrical Engineering, Islamic Azad University, Saveh, Iran Scott Moran University of North Texas, Denton, TX, USA Susanna Morelli Department of Energy, Politecnico di Milano, Milano, Italy Seyedmohammad Mousavian

Sydney, Australia

Nadia Mushtary Institute of Nuclear Power Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh F. J. Muñoz-Rodríguez Center for Advanced Studies in Earth Sciences, Energy and Environment (CEACTEMA), University of Jaen, Jaen, Spain Behzad Najafi Dipartimento di Energia, Politecnico di Milano, Milan, Italy Sai Pushpak Nandanoori Pacific Northwest National Laboratory, Richland, WA, USA Arun Narayanan Lappeenrannan–Lahden teknillinen yliopisto LUT, Lappeenranta, Finland Pedro H. J. Nardelli Lappeenrannan–Lahden teknillinen yliopisto LUT, Lappeenranta, Finland Department of Electrical Engineering, School of Energy Systems, LUT University Lappeenranta, Finland Lappeenranta–Lahti University of Technology, Lappeenranta, Finland Morteza Azimi Nasab CTIF Global Capsule, Department of Business Development and Technology, Denmark and Renewable Energy Lab (REL), Aarhus University, Herning, Denmark

xl

Contributors

Susmita Naskar Faculty of Engineering and Physical Sciences, University of Southampton, Southampton, UK Giovanni Nastasi Department of Mathematics and Computer Science, University of Catania, Catania, Italy Amir Nazemi Ashni Science & Technology Futures Studies Department, National Research Institute for Science Policy (NRISP), Tehran, Iran Sandra Miranda Neves Federal University of Itajubá, Itajubá, Brazil Giuseppe Nicosia Department of Biomedical and Biotechnological Sciences, University of Catania, Catania, Italy Taher Niknam Department of Electrical and Electronic Engineering, Shiraz University of Technology, Shiraz, Iran Javad Nikoukar Department of Electrical Engineering, Islamic Azad University, Saveh, Iran Mohammad Niyazi Smart MPower Company, Limerick, Ireland Kianoush Nokhbeh Dehghan Department of Industrial Engineering, Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran Sima Ohadi Department of Civil Engineering, Faculty of Engineering, University of Sistan and Baluchestan, Zahedan, Iran Fernando Toshio Okamura São Carlos School of Engineering (EESC), University of São Paulo (USP), São Carlos, Brazil Salauddin Omar Department of Mechanical and Production Engineering, Ahsanullah University of Science and Technology, Dhaka, Bangladesh Vadim Omelˇcenko Institute of Information Theory and Automation, Academy of Sciences of the Czech Republic, Praha, Czech Republic Gerardo J. Osorio C-MAST, University of Beira Interior, Covilha, Portugal REMIT, Portucalense University Infante D. Henrique, Porto, Portugal Kyle Paaren Fuel Development, Performance, and Qualification, Idaho National Laboratory, Idaho Falls, ID, USA Ivo Palu Tallinn University of Technology, Tallinn, Estonia Bhagyashree Panda Department of Civil and Environmental Engineering, The George Washington University, Washington, DC, USA Vili Panov School of Engineering, University of Lincoln, Lincoln, UK Siemens Energy Industrial Turbomachinery, Siemens Energy, Lincoln, UK

Contributors

xli

Mohammad Pasandidehpoor Institute for Systems and Computer Engineering (INESC TEC), Technology and Science, Faculty of Engineering, University of Porto, Porto, Portugal Malay Kumar Patra Indian Institute of Management, Tiruchirappalli, TN, India Elizabeth Varsha Paul Department of Humanities and Social Sciences, National Institute of Technology, Tiruchirappalli, TN, India Mohammad Pazouki School of Industrial Engineering, College of Engineering, University of Tehran, Tehran, Iran Antonella Petrillo Department of Engineering, University of Naples “Parthenope”, Napoli, Italy Stefan Pickl Bundeswehr University Munich, Neubiberg, Germany Antti Pinomaa School of Energy Systems, LUT University, Lappeenranta, Finland Olivia J. Pinon Fischer Aerospace Systems Design Laboratory, Georgia Institute of Technology, Atlanta, GA, USA Amir Pirayesh Centre of Excellence in Supply Chain and Transportation (CESIT), KEDGE Business School, Talence, France Efstratios N. Pistikopoulos Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Energy Institute, Texas A&M University, College Station, TX, USA Parastoo Poursoltani Ferdowsi University of Mashhad, Mashhad, Iran Jubilee Prasad Rao Global Technology Connection Inc., Atlanta, GA, USA Faeze Momeni Rad Department of Civil and Environmental Engineering, University of Alberta, Edmonton, AB, Canada Seyed Habib A. Rahamti Department of Industrial Engineering, Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran M. Rahman Department of Mechanical Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh Soulmaz Rahman Mohammadpour Department of Industrial Engineering, Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran Seyed Habib A. Rahmati Department of Industrial Engineering, Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran

xlii

Contributors

Mohammad Sadra Rajabi School of Civil Engineering, University of Tehran, Tehran, Iran School of Civil Engineering, Faculty of Engineering, University of Tehran, Tehran, Iran Alimohammad Rajaee Department of Maritime Engineering, Amirkabir University of Technology, Tehran, Iran P. S. Ramapraba Panimalar Institute of Technology, Chennai, India Syed Saba Raoof Vardhaman College of Engineering, Hyderabad, India Avishek Ray Department of Energy Science and Engineering, Indian Institute of Technology Bombay, Mumbai, India Fabio Rinaldi Dipartimento di Energia, Politecnico di Milano, Milan, Italy Vittorio Romano Department of Mathematics and Computer Science, University of Catania, Catania, Italy Arnab Roy Procter and Gamble Co., Cincinnati, Ohio, USA Javier Rubio-Herrero University of North Texas, Denton, TX, USA C. Rus-Casas Center for Advanced Studies in Earth Sciences, Energy and Environment (CEACTEMA), University of Jaen, Jaen, Spain Saba Sabet Laboratory of Innovations in Transportation (LiTrans), Ryerson University, Toronto, ON, Canada Kamyar Sabri-Laghaie Faculty of Industrial Engineering, Urmia University of Technology, Urmia, Iran J. Kiarash Sadeghi R. Department of Marketing and Supply Chain Management, Willie A. Deese College of Business and Economics, North Carolina A&T State University, Greensboro, NC, USA Naser Safaeian Hamzehkolaei Department of Civil Engineering, Bozorgmehr University of Qaenat, Qaen, Iran Mehmet Sahinoglu Computer Science Department, Troy University, Troy, AL, USA A. Salehabadi Priority Research Centre for Frontier Energy Technologies & Utilization, Discipline of Chemical Engineering, School of Engineering, Faculty of Engineering and Built Environment, The University of Newcastle, NSW, Australia Mohammad Reza Salehizadeh Department of Electrical Engineering, Marvdasht Branch, Islamic Azad University, Marvdasht, Iran Valerio Antonio Pamplona Salomon Sao Paulo State University, Sao Paulo, Brazil

Contributors

xliii

Mohammad mahdi Sarbishegi Tehran University, Tehran, Iran Arthur Sousa de Sena Lappeenranta–Lahti University of Technology, Lappeenranta, Finland S. Senthilmurugan SRM Institute of Science and Technology, Kattankulathur, Chennai, India Miadreza Shafie-khah School of Technology and Innovations, University of Vaasa, Vaasa, Finland Chiranjibi Shah Department of Electrical and Computer Engineering, Mississippi State University, Starkville, MS, USA Mohsen Shahandashti Department of Civil Engineering, The University of Texas at Arlington, Arlington, TX, USA Hossein Shayeghi Energy Management Research Centre, University of Mohaghegh Ardabili, Ardabil, Iran Volodymyr Sherstjuk Kherson National Technical University, Kherson, Ukraine Pierluigi Siano Department of Management and Innovation Systems, University of Salerno, Salerno, Italy Angelo Sifaleras Department of Applied Informatics, School of Information Sciences, University of Macedonia, Thessaloniki, Greece Ana R. C. Silva C-MAST, University of Beira Interior, Covilha, Portugal ICOVI – EdGeWiSe – Infras. e Concessoes da Covilha, Tortosendo, Portugal Ricardo M. Silva C-MAST, University of Beira Interior, Covilha, Portugal Federal University of Paraiba, João Pessoa, Brazil Arijit Sinha Department of Metallurgical Engineering, Kazi Nazrul University, Asansol, India Laura Soares Rutgers University, Piscataway, NJ, USA Nassim Sohaee Information Technology and Decision Science, University of North Texas, Denton, TX, USA Olga Sokolova Peter the Great St. Petersburg Polytechnic University, St. Petersburg, Russia Yupeng Song College of Civil Engineering, Nanjing Tech University, Nanjing, P.R. China Amirhossein Souhankar Knowledge-Base Economy Group, Technology Studies Institute, Tehran, Iran

xliv

Contributors

Ricardo Teixeira Sousa Institute for Systems and Computer Engineering (INESC TEC), Technology and Science, Faculty of Engineering, University of Porto, Porto, Portugal U. Soyta¸s Dept. of Technology, Management, and Economics, Technical University of Denmark, Lyngby, Denmark Shiva Srenivasan Srinivasan SRM Institute of Science and Technology, Kattankulathur, Chennai, India Swaminathan Sundar Texas A&M Energy Institute, Texas AM University, College Station, TX, USA Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Alejandra Tabares Pozos Department of Industrial Engineering, Los Andes University, Bogotá, Colombia Elham Taghizadeh Department of Industrial and System Engineering, Wayne State University, Detroit, MI, USA Sofiène Tahar Department of Electrical and Computer Engineering, Concordia University, Montreal, QC, Canada Willian Y. Takano São Carlos School of Engineering, University of São Paulo, São Carlos, Brazil Payam Teimourzadeh Baboli Energy Division, OFFIS – Institute for Information Technology, Oldenburg, Germany Grace Anne Thompson Department of Chemical Engineering, University of Waterloo, Waterloo, ON, Canada Sai Tejaswini Thumpala University of North Texas, Denton, TX, USA Behrouz Tousi Department of Electrical Engineering, Urmia University, Urmia, Iran Tien Anh Tran Faculty of Marine Engineering, Vietnam Maritime University, Haiphong City, Vietnam Marine Research Institute, Vietnam Maritime University, Haiphong City, Vietnam Hai Hong Trinh Department of Property, School of Economics and Finance, Massey Business School, Massey University, Palmerston North, New Zealand School of Economics and Finance, Massey Business School, Massey University, Palmerston North, New Zealand Vasudev Trivedi Information Technology and Decision Sciences, University of North Texas, Denton, TX, USA Marios Tsioufis Department of Economics, University of Peloponnese, Tripoli, Greece

Contributors

xlv

Alexandros Tzanetos Faculty of Engineering, Multiobjective Optimization REsearch Lab (MORE Lab), Department of Electrical Engineering & Computer Engineering, Université de Sherbrooke, Sherbrooke, QC, Canada Mehar Ullah Lappeenranta–Lahti University of Technology, Finland

Lappeenranta,

Anatoli Paul Ulmeanu Department of Power Generation and Use, Polytechnic University of Bucharest, Bucharest, Romania R. Ulusoy Yılmaz Middle East Technical University, Ankara, Turkey Fábio Luiz Usberti Institute of Computing, University of Campinas, Campinas, Brazil Shoaib Usman Department of Nuclear Engineering and Radiation Science, Missouri University of Science and Technology, Rolla, MO, USA Burak Uzal Department of Civil Engineering, Engineering Faculty, Abdullah Gül University, Kayseri, Turkey Nima Vaezi Ferdowsi University of Mashhad, Mashhad, Iran Amin Vahidi Industrial Engineering, Shahid Beheshti University, Facility of Mechanics and Energy, Tehran, Iran Mohsen Varmazyar Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran Shivam Vedant Energy Institute, Texas A&M University, College Station, TX, USA Department of Multidisciplinary Engineering, Texas A&M University, College Station, TX, USA Richa Verma Department of Electrical Engineering, Indian Institute of Technology, New Delhi, India Department of Electrical Engineering, Indian Institute of Technology Delhi, Delhi, India S. Vijayalakshmi Department of Electrical and Electronics Engineering, SRM Institute of Science and Technology, Chennai, India Adrian Vulpe-Grigora¸si Department of Applied Electronics and Information Engineering, Polytechnic University of Bucharest, Bucharest, Romania Hao Wang Rutgers University, Piscataway, NJ, USA Scott Warren University of North Texas, Denton, TX, USA Adil Wazeer School of Laser Science and Engineering, Jadavpur University, Kolkata, West Bengal, India Fushuan Wen Tallinn University of Technology, Tallinn, Estonia

xlvi

Contributors

Etienne Willkomm Joint Research Centre – European Commission, Ispra, Italy Hume Winzar Business School, Macquarie University, Sydney, NSW, Australia Annika Wolff Lappeenranta–Lahti University of Technology, Finland

Lappeenranta,

Karl Worthmann Optimization-based Control Group, Technische Universität Ilmenau, Ilmenau, Germany Zhimin Xi Rutgers University, Piscataway, NJ, USA Bing Yan Department of Electrical and Microelectronic Engineering, Rochester Institute of Technology, Rochester, NY, USA ˙ Ibrahim Yilmaz Department of Industrial Engineering, Ankara Yıldırım Beyazıt University, Ankara, Turkey Zsuzsanna Zanatyné Uitz Association of Hungarian District Heating Enterprises, Budapest, Hungary District Heating,Kaposvári Vagyonkezel˝o Zrt Kaposvár, Hungary Mohammad Zand CTIF Global Capsule, Department of Business Development and Technology, Denmark and Renewable Energy Lab (REL), Aarhus University, Herning, Denmark Zahra Zand Razi University, Kermanshah, Iran Seyed Farhad Zandrazavi Department of Electrical Engineering, São Paulo State University, São Paulo, Brazil Yichi Zhang Finance Department, Shaanxi Academy of Social Sciences, Xi’an, China Maryna Zharikova Bundeswehr University Munich, Neubiberg, Germany Kherson National Technical University, Kherson, Ukraine Jun Zheng School of Mathematics, Southwest Jiaotong University, Chengdu, Sichuan, China Department of Electrical Engineering, Polytechnique Montreal, Montreal, QC, Canada Guchuan Zhu Department of Electrical Engineering, Polytechnique Montreal, Montreal, QC, Canada Enrico Zio Department of Energy, Politecnico di Milano, Milano, Italy MINES ParisTech/PSL Université Paris, Centre de Recherche sur les Risques et les Crises (CRC), Sophia Antipolis, France Politecnico di Milano, Milan, Italy CentraleSupélec, Paris, France

Part I Reliability Enhancement of Energy Systems

The Need for Self-Sufficiency and Integrated Water and Energy Management The Case Study in the Water Supply System in a Small Mountain Town Ana R. C. Silva, Ricardo M. Silva, Gerardo J. Osorio, Fernando Charrua-Santos, and Antonio Espirito-Santo Contents 1 Highlights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 The Nexus Water-Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Energy and Water Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 The Development and Installation of the System in the Small Mountain Town of Covilha, Portugal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 The 4Es and the Integrated Energy and Water System Installed in Covilha . . . . . . . . . . . . 6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 4 8 8 9 10 14 17 18 19 19

A. R. C. Silva C-MAST, University of Beira Interior, Covilha, Portugal ICOVI – EdGeWiSe – Infras. e Concessoes da Covilha, Tortosendo, Portugal e-mail: [email protected] R. M. Silva C-MAST, University of Beira Interior, Covilha, Portugal Federal University of Paraiba, João Pessoa, Brazil G. J. Osorio C-MAST, University of Beira Interior, Covilha, Portugal REMIT, Portucalense University Infante D. Henrique, Porto, Portugal e-mail: [email protected] F. Charrua-Santos · A. Espirito-Santo C-MAST, University of Beira Interior, Covilha, Portugal © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_6

3

4

A. R. Silva et al.

Abstract

“Smart energy” concept focuses on (i) being available and accessible in quantity and quality, (ii) being sustainable, that is, economically viable, socially fair, and environmentally correct, and (iii) the 4Es: Energy Amount, being available for consumption; Economic Energy, having viable cost and price; Equity Energy, serving the population with social justice; and Ecologic Energy, not deteriorating nature. Obeying 4Es is demanding, with major stakeholders interacting in the production and consumption policies, and so, the balance between energy and water security within the 4E’s is quite complex, regarding the minimization damage. The main goal of this chapter is to present the results of the EdGeWiSe Project – Energy and Water Systems Integration and Management, which research area focuses on contributing to the integration of water and energy systems in a single and highly efficient system, following the dynamics of the 4Es, specifically, presenting the case study of the development and implementation of a real scale pilot in urban context, which took place at Alexandre Aibeo Park, in the mountain city of Covilha, Portugal, considering an integrating water-energy system. Keywords

Integrated Systems · Self-Sufficiency · Smart Energy · Water-Energy Nexus · Water Systems

1 • • • • • •

2

Highlights Increased energy and climate crisis demand actions from stakeholders. Water needs energy and energy needs water paradigm. Opportunity to rethink the water cycle and ensure a circular management. Both sectors require solutions toward global and sustainable efficiency. Decentralized endogenous energy production ensures higher safety. Water and energy integrated smart systems for a sustainable future.

Introduction

The current economic and population growth is promoting an energy crisis and intensifying the climate change, and so, managing the interdependence between energy and water in an efficient, effective, and sustainable manner is becoming increasingly important (Halkos and Gkampoura 2021; Owusu and AsumaduSarkodie 2016). “Smart Energy” is a concept where all the elements are efficiently integrated, counting on the users’ active engagement, which some call “collective intelligence.” In fact, the balance between production and consumption is difficult regarding the ongoing population global growth that generates an increase in the consumption of goods, services, and food, which in turn, drives the growth of demand for energy

The Need for Self-Sufficiency and Integrated Water and Energy Management

5

under a snowball effect (Thornbush and Golubchikov 2021; Savelli and Morstyn 2021). The International Energy Agency (IEA) estimates that even with all energy efficiency programs in place, energy production and consumption will increase from 14,000 Mtoe (million tons of oil equivalent) in 2020 to 18,000 Mtoe in 2040 (IEA (International Energy Agency) 2018). Thus, guaranteeing energy for all is a priority of any public policy, at a global and local level. Water consumption follows the same trend. Despite being mistakenly considered a renewable and abundant resource, it is not always available for human consumption and, therefore, its availability can directly affect human-geographic-energy systems. Hence, water consumption is vital for life with direct association to the energy sector as it becomes indispensable in extracting, processing, storing, and transporting processes, energy is also needed in water processes such as collection, treatment, and supply (Djehdian et al. 2019; Fernandez Garcia et al. 2018; Harding 1888; Wang et al. 2019). Given that the confidence and quality of energy and water states are a global concern that affects the daily lives of modern populations, expecting a constant increase in the quality of life standard, makes the management of these resources – both in rural and urban areas – to have a profound impact on the local economy. Consequently, the exponential increase in the demand and exploitation of energy and water has negative repercussions, both in terms of climatic changes, pollution, and water-energy insecurity, as well as an increase in exploration pressure, mainly on freshwater sources, with the agricultural, industrial, and energy sectors most affected (Silva et al. 2017; Phillis et al. 2018; Pukšec et al. 2018; Su et al. 2018; Stang et al. 2018). The report produced by the World Energy Council in 2010, Water for Energy, assesses the scale of the problem, and reveals the necessary stages of the process to ensure water-energy nexus sustainability and that water and energy are available to the society worldwide, considering the needs, technology, and renewable resources (Gadonneix et al. 2010). Currently, Portugal is focused on achieving the following objectives associated with the water-energy nexus (Oliveira et al. 2017; Fernandes et al. 2017): to reduce at least 50% the energy consumed in water systems; improve the efficiency of operational activities; transform energy-dependent facilities; use more renewable energy sources in the urban water cycle, without changing the reliability and efficiency of processes, and increase the integration of water services with other sectors, especially with energy. Whether for economic, political, or social reasons, both sectors are evolving rapidly, requiring the implementation of solutions toward global efficiency and longterm sustainable exploitation. There is still limited research on their interdependence and urgent action is needed. In this sense, it is necessary to understand the global public policies to articulate the water and energy sectors and create beneficial synergies to increase resource efficiency and sustain future generations. There is a huge potential for growth in the theme of water-energy nexus, involving management entities that seek to reduce costs, environmental care, and integration with society through integrated water and energy solutions, using smart

6

A. R. Silva et al.

grids and technology companies that have evolved into a new phase of development, answering the problems and restrictions encountered. In this sense, and seeking that the planet is close to the environmental nonreturn, both sectors (water and energy) must join efforts for a new era of integration and cooperation in search of new integrated management strategies and improved efficiency in exploration resources toward a more sustainable future (Silva et al. 2018a; Kabayo et al. 2019; Mamade et al. 2017; Silva 2019). Considering, for example, a water distribution system in a mountain town, and therefore with gravitational potential hydro energy complemented with photovoltaic and/or wind systems, the energy can be recovered and used when necessary to meet the local requirements. However, the population’s need for water according to the day and night consumption pattern should be taken into consideration, as well as the available energy storage capacity that may be part of a set of water reservoirs and/or a battery system (Silva 2019). Integrated water and energy management systems are also directly related to the concept of smart cities, whether in urban or rural environments, and consequently associated with the concept of Living Lab (LL), an open and innovation space oriented toward the user, allowing active participation and contribution to the scientific development and innovation process, allowing as well to show the validity of innovative services and business models to citizens (Bevilacqua and Pizzimenti 2016; Bulkeley et al. 2016; Hossain et al. 2019; Sestini et al. 2009). Thus, this work is a result of the EdGeWiSe Project, which focuses on the integrated management of energy and water infrastructures in urban context, specifically, in the case study of the integrated system installed in the small mountain town of Covilha, in Portugal, addressing the decentralized energy micro production and consumption management (Silva et al. 2018a; Cherif et al. 2018a). The EdGeWiSe Project has successfully achieved the objectives: (i) promoting the efficiency based on data collected by wireless and low-power sensor networks, powered by microbial fuel cells and vibration; (ii) using solar and wind for micro production with battery storage; (iii) encouraging the intelligent use of renewable sources by integrating water and energy; (iv) exploring the impact of micro-hydro technology on river systems; (v) stimulating teaching and learning at a higher level with a partnership with the University of Beira Interior, and (vi) promoting environmental education and knowledge among the local population. In this sense, the following premises are evident: (i) the clear and important connection between energy and water systems; (ii) the preservation of both resources to meet lasting sustainability; (iii) under the concept of Smart Cities, every city should be governed as an integrated environment, where all systems collaborate to reach an optimum point of operation, where a set of houses/villages/cities should be provided with renewable energy and water for all. In short, there is an evident gap in literature on the development and implementation of integrated energy and water systems. On one side, there is the current and very present climate change and energy crisis, the populational growth, the increasing demand for water and energy and their very known scarcity and insecurity around the globe.

The Need for Self-Sufficiency and Integrated Water and Energy Management

7

On the other side, there is the urgent need to: develop new technologies on both sectors, increase the use of renewable energy, reduce energy and water consumption, convert energy consumption places into production ones, among others. However, the lack of integrated policies and a few more incentives or programs (which are still not enough) to actual proceed with the integration that both sectors require, is delaying the global communities’ progression and consequent mitigation measures on climate change and insecurity. The matter on integrated and closed energy and water systems is exposed in this work, with the development of a real scale pilot, and consequent experimental validation in real and urban context, installed in a pipeline, located in “Jardim Botanico de Montanha” (The Mountain Botanical Garden), also called “Parque Alexandre Aibeo,” in the Water Storage System (WSS) of Covilha, Portugal, from the perspective of “4Es” of energy: Energy amount (it exists in quantity and it is available for consumption); Economic Energy (having viable cost and price); Equity Energy (serving the population with social justice), and Ecologic Energy (its production and consumption does not deteriorate nature). The pilot serves as a starting point to study and implement these premises, being eventually extrapolated to other sections of the city, country, or even the world, taking advantage of the gravitational potential of mountain areas. It presents itself as a proposal for an intelligent and integrated system for water supply networks, which will be prepared to operate alone within a specific set of infrastructures and electronic monitoring devices adapted to local conditions. It is intended, therefore, to disseminate the importance of the connection between water and energy decentralized systems to communities, so that their efficiency in terms of resource use/exploitation can be improved, calling for an improvement of national policies and minimization of negative impacts. This concept is an attractive solution, because through micro solar, wind and hydro production, it provides self-support to the water distribution system, supplying a potential gravitational off-grid energy system, taking advantage of the city’s slope. When renewable sources are idle, they supply the energy to engines that pump water to a higher level, which, when necessary, return to the lower level generating energy. The implementation of these systems takes place between two or more water reservoirs in different areas and altitudes and their water level must be managed throughout the day considering two factors: the need for water by consumers and the maximization of the system’s capacity to store energy. It is noteworthy, despite the pumping power being higher than the hydric production from the same flow, it cannot be considered as a loss, since the pumping occurs in idle hours or “lost” energy that is recovered, within the concept of energy harvesting. In this way, energy can then be used, when necessary, for example, to illuminate public streets and the infrastructures at night. This system is being developed since 2016, and a long methodology was used to build the concept, from the survey of the state-of-the-art, to the implementation and testing of the real pilot. It is then intended to continuously study the optimization problem in integrated and closed systems of energy and water through this system, a full-scale pilot

8

A. R. Silva et al.

installation in urban context. The viability study of integrated systems for energy micro production of this type, allows for the contribution of knowledge to the entire scientific community, as well as to a more sustainable future and to the improvement of energy efficiency of water supply and distribution systems. Also, it promotes the reduction of fossil fuel consumption and carbon emissions. The methodological path was focused on: (i) the model analysis for energy independence based on endogenous resources in mountain towns; (ii) the improvement and management of integrated systems of hydro, solar and wind energy in the scope of micro production, defining their point of optimization; (iii) the research and prepare the state-of-the-art; (iv) the implementation of the pilot in real and urban context; (v) in situ data analysis, using computer and advanced technology equipment; (vi) the definition of the optimal point of operation definition of the system; (vii) the exploration plan as a way to propose the dynamization of the investment made in the near future; and (viii) the conclusion of the results obtained in the vision of the 4Es.

3

The Nexus Water-Energy

3.1

Energy

In physics, energy corresponds to the ability to do work and it is an essential resource on the globe. The presence of energy occurs in any action that implies movement, temperature variation, or wave transmission. Renewable energy sources “are natural resources, capable of regenerating in a short period, and in a sustainable way ( . . . ), avoiding the greenhouse gases emissions (GHG) and reducing the price of electricity in the market, contributing to a greater economic and environmental sustainability” (APREN 2021). It is a priority objective of the World Energy Council (WEC) to contribute to the global energy strategies, promoting the sustainable energy supply and use. To this end, the concept of the Energy Trilemma was developed, which addresses the triple challenge: ensuring safety, affordable, and environmentally responsible energy (World Energy Trilemma 2014 2021). In (World Energy Scenarios 2013) indicates the potential role of renewable energies to reduce fossil fuels in the primary energy mix, the needed investments in the new infrastructures to meet the future electricity demand and regional differences for each scenario developed (World Energy Scenarios 2016). The Jazz scenario focuses on energy equity, with priority on individual access and accessibility and quality of energy through economic growth, whose main players are multinational companies, banks, venture capitalists, and price-conscious consumers. The basic way to produce electricity is through a primary source that prints kinetic energy (through fossil sources that generate heat), or renewable sources (such as wind, or moving water), turning a turbine/alternator assembly, which generates electricity, or through alternative sources such as solar radiation captured by photovoltaic plates.

The Need for Self-Sufficiency and Integrated Water and Energy Management

9

In 2019, renewable energy production was around 51%, of which 26% belongs to wind (APREN (Associação de Energias Renováveis) 2014; REN (Redes Energeticas Nacionais) 2020). In fact, Directive 28/2009/EC of 23 April 2009, introduced the obligation for EU member countries to submit a plan to use renewable energy. Even so, the energy sector made the largest contribution to GHG emissions in 2017 (72.6%) with energy production and processing (Relatório do Estado do Ambiente de 2019). Hence, the Energy Trilemma, it focuses on the following: “achieving high performance in all the three dimensions which implies a complex interwoven links between public and private bodies, governments and regulators, economic and social factors, national resources, environmental concerns and individual consumer behavior.” In 128 countries analyzed, Portugal is in 29th place in the overall equilibrium ranking of the Energy Trilemma. In the individual rankings, Portugal is in 54th in security, 42nd in equity, and 34th in sustainability (World Energy Trilemma Index 2019).

3.2

Water

Despite being considered a renewable and abundant resource, water is not always available for human and animal consumption in the quantity, quality, time, and place required. The water (freshwater) suitable for human use (in agriculture, industrial sector, and municipal) represents a very small percentage in relation to the total availability of water on the planet, more precisely, 0.0075%. In (Understanding the Energy-Water Nexus 2014), the authors reveal that approximately 2.5% of the planet’s water is sweet and about 0.3% of it is superficial, which 10% correspond to water suitable for human use, with all the remaining water in inaccessible places or stored in glaciers, polar ice caps, and in very deep subsoil layers. Also, it reveals that 19% of the water is used in the industrial sector and that according to the ECN (Energy research Center of the Netherlands) about 5% of this value is used for energy production in the conventional system based on fuels fossils. In (Relatório do Estado do Ambiente de 2019), the European legislation that has been developed for the water sector, reclaim as their main objective the protection of inland surface waters, transition waters, coastal waters and groundwater in order to: (i) avoid degradation, protect and improve the ecosystems state; (ii) promote the sustainable water consumption; (iii) strengthen and improve the aquatic environment by reducing or ceasing discharges, emissions and losses of priority substances; (iv) ensure the gradual reduction avoiding the worsening groundwater pollution; (v) contribute to mitigate the effects of floods and droughts; (vi) guarantee, in sufficient quantity, the good surface and underground water quality, aiming at a sustainable, balanced and equitable use of water; and (vii) protect the marine waters, by promoting the pollution prevention in the marine environment. Regarding how to supply water to the community, the supply cycle begins with the collection phase, in which the process of obtaining water needs to respect the

10

A. R. Silva et al.

system requirements and the needs of the population supplied by it. For instance, in Portugal, in 2019, there were 5896 abstractions of underground water and 295 abstractions of surface water. According to (Relatório anual dos serviços 2019), in the city of Covilha, Portugal, there are 28 abstractions of underground water and 1 of surface water. Additionally, in Portugal, in the same year, there were 269 WTP (Water Treatment Plants) (Relatório anual dos serviços 2019), where in Covilha there is one located. Water treatment represents the “set of unitary operations which purpose is to make the necessary corrections so that the physical, chemical and bacteriological characteristics of treated water are suitable for human consumption” (Abastecimento de água 2021). As for elevation, the energy is introduced into the flow so that water can be transported to higher areas (if necessary). In Portugal, in 2019, there were 2395 pumping stations (Relatório anual dos serviços 2019), where in Covilhã, there are located 11. After water treatment and elevation, the transport phase comes into action, transporting the water from the reservoirs or treatment stations through pipelines to the population. The storage phase is crucial, since it consists of reservoirs that balance fluctuations in water consumption and respond to emergency circumstances, such as firefighting. They also help to stabilize the network pressure from consequently pump operation. Hence, the distribution process consists of transporting water from reservoirs to the final consumer in required quantity, continuously and with good quality. In Portugal, there were about 112,950 kilometers of water pipes and 8829 reservoirs in 2019 (Relatório anual dos serviços 2019). In this context, regulation of the water sector is crucial for human survival, and so it has to: “protect the interests of users; promote the efficiency and innovation; ensure its stability, sustainability and robustness,” and which obligations focus on the universality, equity, accessibility, continuity, adaptability, and transparency (Marques 2011). In Portugal, the first public policies in the water supply sector and sanitation emerged at the end of the nineteenth century (Evolufão Histórica 2020). In April 2015, PENSAAR 2020 (New Strategy for the Water Supply and Wastewater Sanitation Sector, Dispatch No. 4385/2015, of 30 April 2015) appears. In this plan, the resilience and security of basic water and sanitation services are “recognized as a human right, or their support for green growth, require the provision of quality services in a professionalized way, considering the social, economic and financial sustainability” (PENSAAR 2020 – Uma Estrategia ao Servifo da Populafão 2015).

3.3

Energy and Water Systems

Over time, and especially in the last two decades, a considerable number of occurrences have arisen related to the increase in energy resources and the decrease in water resources, whether directly or not because of climate change. Such facts

The Need for Self-Sufficiency and Integrated Water and Energy Management

11

focus on the need to reformulate most of the environmental policies that involve the water-energy nexus. However, the responsible entities for decision making in this matter had come around the situation by solving individual problems related only to water or only to the energy sector (Silva et al. 2017; Al-Saidi and Elagib 2017; Cao et al. 2018). In 1996, the EPRI’s (Electric Power Research Institute) published a report entitled “Water and Wastewater Industries: Characteristics and Energy Management Opportunities” (Burton 1996), in which it describes how the electricity is used and managed more effectively in the treatment of drinking water and wastewater. Already at this time, in the USA, both the energy production industry and the water supply and sanitation companies recognized the irreversible link between water and energy (Pabi et al. 2013). Since its publication, the report has been cited as one of the earliest resources in the study of the water-energy nexus and continues to be an important reference today, although with updates in 2013. The “water-energy nexus” concept first appeared in 2002 (Malik 2002), revealing that India has reached a point where the demand for resources exceeds its availability, requiring emergency measures to population may be able to deal with uncertainties, unreliability, and shortages of water and energy. The article analyzes the connection nature between water and energy at the final consumer level, for instance, in agriculture and water supply sectors, as well as some local policies that may help to improve the balance between the demand and supply of resources in an integrated manner. In the analyzed literature, the nexus has been approached and developed according to the needs of the planet. The example in (Lamberton et al. 2010) is a reflection of it: in 2003 a massive heat wave swept through France, killing more than 15,000 people. The sudden drop in water levels and high temperatures led the country to take the decision to reformulate some policies regarding the water supply for the cooling of nuclear power plants. However more catastrophic events are described until nowadays. In Portugal, the research on “water-energy nexus” is a very recent subject, since the oldest reference found dates back to 2012 (Kenov and Ramos 2012). In 2014 alone, an article appeared on the technology for the intelligent measurement of consumption flows to increase water and energy efficiency, analyzing more than 50 domestic and service cases (Loureiro et al. 2014). Lately, both “water” and “energy” have been used in technical and holistic approaches regarding their interconnection and management, allowing manageable and affordable solutions to the development of future renewable sources technology and helping to minimize climate change impact. In a survey conducted on the Scopus platform in March 2020 with the terms “water-energy nexus” and “Portugal,” it was found that smart municipal water supply services and their respective energy use are the most researched. In 2014, there was an article on smart metering technology for consumption flow to increase water and energy efficiency, analyzing more than 50 cases at household and service level (Loureiro et al. 2014), warning of the lack of existing consumption data and how it impacts the efficiency of systems.

12

A. R. Silva et al.

In 2015 a proposal for an energy audit tool appears to describe the causes of inefficiency in water supply systems (Mamade et al. 2015) and in 2016, performance indicators emerge as an energy management methodology (Teixeira et al. 2016). The remaining 85% of articles in this category appear from 2017 onward, covering the following subjects: reviews on the potential recovery of hydroelectric energy in water networks (Pérez-Sánchez et al. 2017a; Coelho and Andrade-Campos 2018; Dadfaret al. 2019); mitigation strategies to reduce water and energy losses and improve efficiency (Ribeiro et al. 2017; Mamade et al. 2018; Luna et al. 2019; Loureiro et al. 2020); intelligent management of water and energy systems with monitoring (Espirito-Santo et al. 2017; Espirito-Santo et al. 2018; Ramos et al. 2020; Kuski et al. 2020; Giudicianni et al. 2020); analysis of the behavior of pumps as turbines (Capelo et al. 2017; Fernandes et al. 2019); integrated planning of energy and water distribution in islands (Segurado et al. 2018), and implementation of integrated water and energy production systems (among them, the practical case of this work) (Silva et al. 2018a; Cherif et al. 2018b; Silva et al. 2018b). The second most researched field, is the integration of end-use water and energy systems, namely: water consumption reduction, energy and emissions from water conservation measures (Matos et al. 2017; Matos et al. 2019; Pimentel-Rodrigues and Silva-Afonso 2019; Meireles and Sousa 2020; Rodrigues et al. 2020); low-cost water and energy consumption readings (Cunha et al. 2017); the sustainable nexus of water and energy in the optimization of a golf course through the use of renewable energy (Ramos et al. 2019), and the reuse of resources in the use phase of buildings (Pimentel-Rodrigues and Siva-Afonso 2019). The technology in the treatment of rainwater and wastewater was studied in third place (11.5% of the published articles), namely discussing the following subjects: optimization of drainage systems through retention ponds (Ramos et al. 2013a), and for energy recovery (Ramos et al. 2013b); energy performance indicators for wastewater treatment (Silva and Rosa 2015); increasing the self-sufficiency of a wastewater treatment plant with integrated implementation of anaerobic co-digestion and photovoltaic energy (Duarte et al. 2018); evaluation of the waterenergy nexus as a tool to achieve sustainability in water management, the example of rainwater harvesting (Marteleira and Niza 2018); energy collection from wastewater through a microbial fuel cell (Domingos Serra et al. 2018; Serra et al. 2020), and intelligent capacitive level sensors in the management of wastewater treatment processes (Serra et al. 2019). Other types of technology in renewable energies were also addressed: the optimization of a wind desalination system and pumped hydro-storage (Segurado et al. 2016); aspects of the flow regime in the maximization of energy production in hydroelectric plants (Kuriqi et al. 2019); desalination as a viable solution for regions with water scarcity (Azinheira et al. 2019), and an energy storage system for compressed air induced by transient flow (TI-CAES) in hydroelectric plants (Besharat et al. 2020). Additional issues, such as modeling irrigation networks to quantify energy recovery potential (Perez-Sanchez et al. 2016) and optimization strategies to improve their energy efficiency (Pérez-Sánchez et al. 2017b); the integrated waste management of the sewage sludge and olive oil production chain (Fragoso et al.

The Need for Self-Sufficiency and Integrated Water and Energy Management

13

2019); the water-energy nexus in typical industrial water circuits (Oliveira et al. 2019), and the modeling of industrial water circuits with a custom model (Oliveira and Iten 2020) were also addressed. Works with proposals for strategies and solutions for the sustainable use of water and energy in the future were also published. For instance, in 2013, (Braga et al. 2013) was highlighted the necessary changes in the use of energy, food, and services regarding the exploitation of water resources, providing recommendations to address the imbalance caused by the exploitation of water in the agricultural, industrial, and municipal sectors. In 2015, the sustainable efficiency of water-energy-food (WEF) systems (Haie 2016) was addressed. The work warns of the importance of a systematic case-bycase analysis, to improve the safety of the state of the water, not only in quantitative, but also in qualitative terms, since the pollution of the agricultural sector is the most relevant on the planet. Also in 2015, a proposal for an energy audit tool appears to describe the causes of inefficiency in WSS (Mamade et al. 2015) and in 2016, performance indicators has appeared as an energy management methodology (Teixeira et al. 2016). In 2018, climate projections and scale-down techniques in urban systems were addressed in (Smid and Costa 2018). The authors reveal that the adaptation to climate change has been an integral part of urban planning (in terms of energy infrastructure, water supply, wastewater treatment, transport systems, among others) in the short and medium term, but that nothing was specified in terms of long-term measures (50 to 100 years), due to several issues, and by the need of “more urgent” problems. In 2019, within the same subject, in (Markantonis et al. 2019), it is stated that the WEF nexus can help sustain economic growth in the Mediterranean area (and beyond), analyzing the current state of the region and providing strategies in development, especially from the point of view of efficient water use. Elements such as market economic instruments (incentives, subsidies, fees, etc.), integrated assessment approaches, new employment opportunities, management, and availability of information (data), innovative institutional configurations, and organized dialogue between stakeholders are considered essential in the study and necessary for local economic development. In the same year, the efficiency analysis of the water-energy-earth-food nexus (WEF nexus) was investigated in OECD countries (Ibrahim et al. 2019), where the efficiency of the WEF nexus in present and future generations is analyzed from the point of view of environmental viability, taking into account the availability of natural resources and innovation-oriented policies. As a main conclusion, the study reveals that the efficiency of the WEF nexus is not so much to do with the minimum use of associated resources, but with their adequate use, even revealing that, according to demographic development forecasts, it will be necessary to increase the use of water for the nexus to be efficient and sustainable. In 2020, transnational competition for water in the context of the water-energy nexus in the Mediterranean region has been studied, also evaluating the economic impacts of climate change and in the energy sector (Teotonio et al. 2020). Two

14

A. R. Silva et al.

different climate change scenarios were analyzed (one moderate and one intense) as well as two cross-border water scenarios between Portugal and Spain. The results clearly show that climate change has economic consequences that directly affect the availability of water resources, depending on the severity of water restrictions, and that the promotion of public policies that encourage water allocation is essential. The same document shows that economic and social costs might be minimized when the priority is given to the use of water in non-electrical production sectors. However, it is expected in the coming years, some new technologies development in the electricity production associated with the minimization of the water impact, such as the improved integration of decentralized photovoltaic and wind energy production.

4

The Development and Installation of the System in the Small Mountain Town of Covilha, Portugal

In the municipality of Covilha, the public supply system is characterized by the following: (i) in areas of high energy consumption and, (ii) medium volume of water transported for it; depending on the distance covered, the water losses represent higher energy losses, although there are areas with a slope, favoring the energy potential. The municipal WSS is focused on the quality of the water distribution to the population, which is vital for the development of the city and for the safety of public health. To carry out such supply, the follow-up of environmental regulation can affect the increase in operating costs. This entails opportunities for the optimized management of resources, that according to the report developed by EPRI in 2013 (Pabi et al. 2013), consists of a practical tool to: • Understand the water supply and sanitation sector and its associated challenges. • Understand the other operations and processes implemented in the treatment of water and the way electricity is used in different system configurations. • Identify and characterize opportunities to improve energy efficiency. • Assist in the development of energy management plans to reach such opportunities. To achieve the objectives, the EdGeWiSe team composed of University of Beira Interior (UBI), Portugal, University of Poitiers (UP), France, University of Tunes (UT), Tunisia, National Technical University of Athens (NTUA), Greece, University of Cyprus (UC), Cyprus, Malta College of Arts, Science and Technology (MCAST), Malta, ICOVI, Portugal, and Greenbirdie Group Solar (GBS), France, gathered information in literature, government entities, private research groups, and other sources, to characterize the public supply system on the number and type of infrastructure, associated processes, energy use, and respective consumption patterns.

The Need for Self-Sufficiency and Integrated Water and Energy Management

15

The EdGeWiSe team segmented each sector (energy and water) based on the following parameters: size, function, and key elements of the process to characterize the management of energy resources. Representative infrastructures were included in the study also, to analyze the application of technological solutions and optimized management. In detail: The environment: the search for solutions to improve the efficiency of the systems, and when the problem of water scarcity starts to be present in decision making, there is the opportunity for companies to rethink the water cycle, being able to adopt its circular management, introducing recirculation strategies according to the 5R’s approach: reduce, reuse, recycle, restore, and recover. The concept of integration: Under the concept of smart cities, the city is governed as an integrated environment, where all systems collaborate to reach an optimum point of operation. A good scenario would be a set of blocks (houses/villages/cities) provided with renewable energy (solar, wind, hydric), controlled by an integrated management center: a concept that was extrapolated to the development of the integrated management model installed by ICOVI, in Parque Alexandre Aibeo, presented in Fig. 1. In this sense, in rural areas and small mountain towns like Covilha, especially where the general electricity grid is not available, this solution becomes quite attractive. In addition to that, the electric energy produced from renewable photovoltaic and/or wind sources can also be stored in the water distribution system as potential gravitational energy, taking advantage of the city’s slope.

Fig. 1 Parque Alexandre Aibéo, in the city of Covilhã, Portugal – the park where the integrated water and energy system is installed

16

A. R. Silva et al.

The pilot configuration: After defining the scenarios, preliminary calculations of electric energy production through renewable sources were carried out (solar, micro-hydric, and wind), where several parameters were analyzed. In hydropower, parameters such as the flow, the useful height, and the type of equipment converting potential energy into electricity, and the respective efficiency are fundamental. In the case of solar, the average annual sunshine hours of Portugal were analyzed and thus, annual solar irradiation data were obtained along with ambient temperature. As for wind energy, the Atlas of Wind Potential of Portugal was used. Therefore, the operating conditions of the system were defined based on the day and night periods, as well as the definitive data to be collected, the necessary equipment, and the objectives of the pilot project: 1. To contribute to a more sustainable future. 2. To use the energy potential of water movement to meet the park’s electrical requirements. 3. To use the available sun and wind hours to pump water to the upper reservoir and store water potential. 4. To test self-production and self-consumption on closed integrated and decentralized systems. 5. To study the maximization of energy and efficiency of such systems. The following elements were chosen based on equipment efficiency and main facilities: (i) the installation of an independent pipeline with no connection to the population (between two reservoirs), with a small diameter of 63 mm, avoiding the infrastructures’ stability; (ii) a PAT (Pump as Turbine), a centrifugal pump that functions in normal and reverse mode, depending on the flow direction; (iii) a solar panel system; (iv) a wind turbine and; (v) a technical room to store machinery, instrumentation, and to monitor the equipment and collect data. The general concept is presented in Fig. 2. The pilot is contributing to the following objectives: • • • • • • •

Obey the European regulation associated with Water-Energy nexus. Reduce the ecological footprint of water services. Reduce the energy consumption in Portugal in water systems by at least 50%. Increase self-sufficiency in energy consumption and production in WSS. Improve the efficiency of operational activities. Transform places of consumption in production facilities. Increase energy recovery from endogenous and renewable resources, such as water, sun, and wind. • Reduce the energy consumption per m3 of water consumed (used, treated, and reused). • The integration and optimization of different uses of water (agricultural, domestic, industrial, among others).

The Need for Self-Sufficiency and Integrated Water and Energy Management

17

Fig. 2 The pilot’s final system configuration

5

The 4Es and the Integrated Energy and Water System Installed in Covilha

Energy amount: Decentralized systems grant a greater flexibility and safety, reaching the final consumer regardless of the location. In the case of Alexandre Aibeo Park, when micro solar and wind power are integrated with a Pump as Turbine (PAT), energy independence was made possible for the system, regarding the public illumination at night, and an artificial stream recirculation pump during the day. The diversity of energy sources always makes the system available: being open, but functioning as if it was a closed system, since at night, when water and energy consumption decreases, the electric wind system pumps the water to the upper reservoir making a gravitational reserve of water for day consumption. Ecologic: The system operates with clean energy without any type of GHG emissions during its useful life. In addition to using the natural gravitational slope, it uses clean energy, whether it is (i) photovoltaic during the day taking advantage of solar irradiation, (ii) wind power using the resource from the mountains during the day and night, and (iii) micro-hydric taking advantage of an independent pipeline from the WSS and generating energy when there is surplus water for consumption. Economic: It is self-sustainable, which already represents an important factor in favor of sustainability. It is also clear that it is not used with the propose of financial benefit, but to serve the population with drinking water and lighting. In addition, the design for the system’s energy enables the use of energy-efficient processes and equipment, with the integration of storage systems, together with the renewable energy production, consequently supplying the existing infrastructures. In this integrated management system, continuous enhancement programs are established, encouraging the progressive implementation of audits in the waterenergy sectors, bringing transparency to the public sector.

18

A. R. Silva et al.

Equity: The pilot itself already has a whole aspect of bringing social justice when it comes to water distribution, however with its implementation, visits focused on environmental education are possible, in addition to the development of higher human resources with the research on the pilot itself, including at the doctoral level. In this system, the EdGeWiSe pilot uses the smart-to-energy concept, which deals with energy monitoring and management, surveillance, and alert systems, with integration of intelligent systems, integrating facilities for people. Thus, there is an operational integration with other sectors, aiming at a global reduction/production and energy integration (irrigation, water reuse, waste treatment, consumption management integrated with the electricity supplier). The EdGeWiSe pilot was developed and implemented within the scope of the ERANETMED/0005/2014 R&D Project, by FCT (Portugal’s Foundation for Science and Technology). The project’s initial objective was to address the problem of the energy-water nexus for the optimization of cost-efficiency in the presence of technologies enabling smart grids, proposing a multi-agent scheduling model. In fact, the EdGeWiSe project faced several barriers to be overcome, as happening in all water-energy nexus projects: (i) European public policies for water and energy are not articulated; (ii) Lack of policies to create synergies and integration between the water sector and industrial sector; (iii) Different programs to promote renewable energy and energy efficiency between European countries; (iv) Absence of incentives in pricing and tariff policies in Europe (potential distortion and disincentive to efficient water and energy solutions); (v) Technologies not adapted to the water sector, developed for large installations and not for small systems and; (vi) Lack of innovative financing models (Serra 2012).

6

Conclusion

The planet cannot attain a sustainable development without first developing both sectors by themselves and satisfy the respective demand, with focus on high population growth, lack of local skills, and low efficiency management. Also, the transition toward a sustainable system requires developing integrated, coherent, and innovative policies dealing with agriculture, economy, environment and natural resources, health, nutrition, societal priorities, and culture. Furthermore, comprehensive sustainability requires scientific research as well as technological development with access to an adequate educational system, determining thereby a knowledge-based community. Immediate action is advised to tackle the environmental challenges and degradation, especially on water scarcity, mainly driven by consumption patterns. There is an enormous market potential for the water-energy nexus in mountain cities: management entities seeking additional profit (or cost reduction) through integrated solutions, together with smart grid development agents, and technology companies, covering the water and energy problems and restrictions.

The Need for Self-Sufficiency and Integrated Water and Energy Management

19

The integrated system of renewable energy installed in Covilha, came to counter all barriers, as it is a system based on the sustainability of the planet, which uses renewable energy, considering the potential of kinetic/gravitational energy as well. There is a potential for improvement in energy efficiency and water systems (including shareholders, employees, suppliers, communities, among others). Thus, the installation of such system is relevant by its ecological contribution and serves as a starting point to reinforce decentralization in the city, facing the impacts of climate change, building resilience, and bringing energy safety to the city.

7

Glossary

• Decentralization: the process by which the elements are distributed away from a central and authoritative group. • Endogenous resources: Local resources that are naturally created and available. • Micro production: Low-energy production activity, with the possibility of delivering electricity to the public electricity grid in a low-voltage level. • Nexus: A connection linking with two or more elements (e.g., water and energy integration). • Off-grid: Independent electricity or water grid which is not connected with the public utilities. • Smart grid: Electricity supply network with smart technology, with innovative technologies for monitoring, communicating, and interacting with the changes. • Trilemma: A condition in which a difficult challenge is composed of three opposed factors or non-matchable variables.

References Abastecimento de água. http://www.ersar.pt/pt/setor/caracterizacao/abastecimento-de-agua. Accessed 24 June 2021 M. Al-Saidi, N.A. Elagib, Towards understanding the integrative approach of the water, energy and food nexus. Sci. Total Environ. 574, 1131–1139 (2017). https://doi.org/10.1016/ j.scitotenv.2016.09.046 APREN – O que sao. https://www.apren.pt/pt/energias-renovaveis/o-que-sao. Accessed 24 June 2021 G. Azinheira, R. Segurado, M. Costa, Is renewable energy-powered desalination a viable solution for water stressed regions? A case study in Algarve, Portugal. Energies 12(24) (2019). https:// doi.org/10.3390/en12244651 M. Besharat, A. Dadfar, M.T. Viseu, B. Brunone, H.M. Ramos, Transient-flow induced compressed air energy storage (TI-CAES) system towards new energy concept. Water (Switzerland) 12(2) (2020). https://doi.org/10.3390/w12020601 C. Bevilacqua, P. Pizzimenti, Living lab and cities as smart specialisation strategies engine. Procedia. Soc. Behav. Sci. 223, 915–922 (2016). https://doi.org/10.1016/j.sbspro.2016.05.315 B. Braga et al., Water and energy, in Water and the Future of Humanity: Revisiting Water Security, (Springer International Publishing, Lisboa, 2013), pp. 159–184. https://doi.org/10.1007/978-3319-01457-9_7

20

A. R. Silva et al.

H. Bulkeley et al., Urban living labs: Governing urban sustainability transitions. Curr. Opin. Environ. Sustain. 22, 13–17 (2016). https://doi.org/10.1016/j.cosust.2017.02.003 F.L. Burton, Water and Wastewater Industries: Characteristics and Energy Management Opportunities (EPRI (Electric Power Research Institute, Palo Alto, 1996) T. Cao, S. Wang, B. Chen, The energy-water nexus in interregional economic trade from both consumption and production perspectives. Energy Procedia 152, 281–286 (2018). https:// doi.org/10.1016/j.egypro.2018.09.124 B. Capelo, M. Pérez-Sánchez, J.F.P. Fernandes, H.M. Ramos, P.A. López- Jiménez, P.J.C. Branco, Electrical behaviour of the pump working as turbine in off grid operation. Appl. Energy 208, 302–311 (2017). https://doi.org/10.1016/j.apenergy.2017.10.039 H. Cherif, S. Tnani, J. Belhadj, A.R. Silva, Optimal sizing and technical evaluation of energy and water system based on micro-hydric solar and wind sources, in Proceedings – IEEE 16th International Conference on Industrial Informatics, INDIN 2018, (2018a), pp. 1018–1023. https://doi.org/10.1109/INDIN.2018.8472076 H. Cherif, S. Tnani, J. Belhadj, A.R. Silva, Optimal sizing and technical evaluation of energy and water system based on micro-hydric solar and wind sources, in Proceedings – IEEE 16th International Conference on Industrial Informatics, INDIN2018, (2018b), pp. 1018–1023. https://doi.org/10.1109/INDIN.2018.8472076 B. Coelho, A. Andrade-Campos, Energy recovery in water networks: Numerical decision support tool for optimal site and selection of micro turbines. J. Water Res. Plan. Manag. 144(3) (2018). https://doi.org/10.1061/(ASCE)WR.1943-5452.0000894 E.S. Cunha, F. Pereira, A. Briga-Sa, S. Pereira, From water to energy: Low cost water & energy consumptions readings. Procedia Comput. Sci. 121, 960–967 (2017). https://doi.org/10.1016/ j.procs.2017.11.124 A. Dadfar, M. Besharat, H.M. Ramos, Storage ponds application for flood control, hydropower generation and water supply. Int. Rev. Civ. Eng. 10(4), 219–226 (2019). https://doi.org/ 10.15866/irece.v10i4.17133 L.A. Djehdian, C.M. Chini, L. Marston, M. Konar, A.S. Stillwell, Exposure of urban food-energywater (FEW) systems to water scarcity. Sustain. Cities Soc. 50 (2019). https://doi.org/10.1016/ j.scs.2019.101621 P.M. Domingos Serra, A. Esoirito-Santo, M. Magrinho, Energy harvesting from wastewater with a single-chamber air-cathode microbial fuel cell, in IECON 2018 – 44th Annual Conference of the IEEE Industrial Electronics Society, (Washington, DC, October 2018), pp. 3847–3851. doi: https://doi.org/10.1109/IECON.2018.8592827 P.T. Duarte, E.A. Duarte, J. Murta-Pina, Increasing self-sufficiency of a wastewater treatment plant with integrated implementation of anaerobic co-digestion and photovoltaics, 103–108 (2018). https://doi.org/10.1109/YEF-ECE.2018.8368947 P. Espirito-Santo S. Serra B. Albuquerque F.S. Ribeiro, J. Pascoa, Low-power smart sensing in energy and water systems integration, in 2017 IEEE International Workshop on Measurement and Networking (M&N), (Naples, Italy, September 2017), pp. 1–6. https://doi.org/10.1109/ IWMN.2017.8078408 A. Espirito-Santo, B.J.F. Ribeiro, C.G.M. Lima, S. Ambrosio, J. Bonifacio, Self-powered smart sensors in the management of water infrastructures, in 2018 IEEE 16th International Conference on Industrial Informatics (INDIN), (Porto, July 2018), pp. 1024–1029. https://doi.org/10.1109/ INDIN.2018.8471949 Evolufão Histórica. http://www.ersar.pt/pt/a-ersar/evolucao-historica. Accessed 29 April 2020 A.C. Fernandes, M.D. Guerra, R. Ribeiro, S. Rodrigues, Relatório do Estado do Ambiente de 2017 (Agênda Portuguesa do Ambiente, December 2017) J.F.P. Fernandes, M. Pérez-Sánchez, F.F. da Silva, P.A. López-Jiménez, H.M. Ramos, P.J.C. Branco, Optimal energy efficiency of isolated PAT systems by SEIG excitation tuning. Energy Convers. Manag. 183, 391405 (2019). https://doi.org/10.1016/j.enconman.2019.01.016 R. Fragoso, A.C. Henriques, J. Gominho, J.M. Ochando-Pulido, E. Duarte, Integrated management of sewage sludge and olive oil production chain waste: Improving conversion process into biomethane, 983–986 (2019)

The Need for Self-Sufficiency and Integrated Water and Energy Management

21

P. Gadonneix, F.B. de Castro, N.F. de Medeiros, R. Drouin, C.P. Jain, Y.D. Kim, J. Ferioli, M.J. Nadeau, A. Sambo, J. Teyssen, A.A. Naqi, G. Ward, Z. Guobao, C. Frei, Water for Energy (World Energy Council, London, 2010) I. Fernandez Garcia, A. Merida Garcia, J.A. Rodriguez Diaz, P.M. Barrios, E.C. Poyato, Waterenergy nexus in irrigated areas. Lessons from real case studies, in Water Scarcity and Sustainable Agriculture in Semiarid Environment, ed. by I. F. Garcia Tejero, V. H. B. T.-W. S. Duran Zuazo, (Academic Press, 2018), pp. 41–59. https://doi.org/10.1016/b978-0-12-8131640.00002-8 C. Giudicianni, M. Herrera, A. di Nardo, A. Carravetta, H.M. Ramos, K. Adeyeye, Zero-net energy management for the monitoring and control of dynamically-partitioned smart water systems. J. Clean. Prod. 252 (2020). https://doi.org/10.1016/j.jclepro.2019.119745 N. Haie, Sefficiency (sustainable efficiency) of water-energy-food entangled systems. Int. J. Water Resource Dev. 32(5), 721–737 (2016). https://doi.org/10.1080/07900627.2015.1070091 G.E. Halkos, E.-C. Gkampoura, Evaluating the effect of economic crisis on energy poverty in Europe. Renew. Sust. Energ. Rev. 144, 110981 (2021). https://doi.org/10.1016/ j.rser.2021.110981 C. Harding, Threatened scarcity of water. Nature 37(955), 955 (1888). https://doi.org/10.1038/ 037375a0 M. Hossain, S. Leminen, M. Westerlund, A systematic review of living lab literature. J. Clean. Prod. 213, 976–988 (2019). https://doi.org/10.1016/j.jclepro.2018.12.257 M.D. Ibrahim, D.C. Ferreira, S. Daneshvar, R.C. Marques, Transnational resource generativity: Efficiency analysis and target setting of water, energy, land, and food nexus for OECD countries. Sci. Total Environ. 697 (2019). https://doi.org/10.1016/j.scitotenv.2019.134017 IEA (International Energy Agency). World Energy Outlook 2018 (IEA, Paris, 2018). [Online]. Available: www.iea.org/weo/water/ J. Kabayo, P. Marques, R. Garcia, F. Freire, Life-cycle sustainability assessment of key electricity generation systems in Portugal. Energy (2019). https://doi.org/10.1016/j.energy.2019.03.166 K.N. Kenov, H.M. Ramos, Water and energy sustainable management in irrigation systems network. Int. J. Energ. Environ. 3(6), 833–860 (2012) A. Kuriqi, N. Pinheiro, A. Sordo-Ward, L. Garrote, Flow regime aspects in determining environmental flows and maximising energy production at run-of-river hydropower plants. Appl. Energ. 256 (2019). https://doi.org/10.1016/j.apenergy.2019.113980 L. Kuski, E. Maia, P. Moura, N. Caetano, C. Felgueiras, Development of a decentralized monitoring system of domestic water consumption. Energy Rep. 6, 856–861 (2020). https://doi.org/ 10.1016/j.egyr.2019.11.019 M. Lamberton, D. Newman, S. Eden, J. Gelt, The water-energy nexus, in Sharon Megdal, vol. 20, (The University of Arizona, 2010) D. Loureiro et al., Smart metering use cases to increase water and energy efficiency in water supply systems. Water Sci. Technol. Water Supply 14(5), 898–908 (2014). https://doi.org/10.2166/ ws.2014.049 D. Loureiro, C. Silva, M.A. Cardoso, A. Mamade, H. Alegre, M.J. Rosa, The development of a framework for assessing the energy efficiency in urban water systems and its demonstration in the Portuguese water sector. Water (Switzerland) 12(1) (2020). https://doi.org/10.3390/ w12010134 T. Luna, J. Ribau, D. Figueiredo, R. Alves, Improving energy efficiency in water supply systems with pump scheduling optimization. J. Clean. Prod. 213, 342–356 (2019). https://doi.org/ 10.1016/j.jclepro.2018.12.190 R.P.S. Malik, Water-energy nexus in resource-poor economies: The Indian experience. Int. J. Water Resource. Dev. 18(1), 47–58 (2002). https://doi.org/10.1080/07900620220121648 C. Mamade, A. Sousa, D. Marques, H.A. Loureiro, D. Covas, Energy auditing as a tool for outlining major inefficiencies: Results from a real water supply system. Procedia Eng. 119(1), 1098–1108 (2015). https://doi.org/10.1016/j.proeng.2015.08.944 D. Mamade, H.A. Loureiro, D. Covas, A comprehensive and well tested energy balance for water supply systems. Urban Water J. 14(8), 853–861 (2017). https://doi.org/10.1080/ 1573062X.2017.1279189

22

A. R. Silva et al.

A. Mamade, D. Loureiro, H. Alegre, D. Covas, Top-down and bottom-up approaches for waterenergy balance in Portuguese supply systems. Water (Switzerland) 10(5) (2018). https://doi.org/ 10.3390/w10050577 V. Markantonis et al., Can the implementation of the water-energy-food nexus support economic growth in the Mediterranean region? The current status and the way forward. Front. Environ. Sci. 7(July) (2019). https://doi.org/10.3389/fenvs.2019.00084 R.C. Marques, A regulação dos serviços de abastecimento de água e de saneamento de águas residuais: Uma perspetiva internacional (ERSAR (Entidade Reguladora dos Serviços de Água e Residuos); Instituto Superior Técnico de Lisboa, 2011) R. Marteleira, S. Niza, Does rainwater harvesting pay? Water-energy nexus assessment as a tool to achieve sustainability in water management. J. Water Clim. Change 9(3), 480–489 (2018). https://doi.org/10.2166/wcc.2017.003 C. Matos, A. Briga-Sá, I. Bentes, D. Faria, S. Pereira, In situ evaluation of water and energy consumptions at the end use level: The influence of flow reducers and temperature in baths. Sci. Total Environ. 586, 536–541 (2017). https://doi.org/10.1016/j.scitotenv.2017.02.008 C. Matos, I. Bentes, S. Pereira, D. Faria, A. Briga-Sa, Energy consumption, CO2 emissions and costs related to baths water consumption depending on the temperature and the use of flow reducing valves. Sci. Total Environ. 646, 280–289 (2019). https://doi.org/10.1016/ j.scitotenv.2018.07.290 I. Meireles, V. Sousa, Assessing water, energy and emissions reduction from water conservation measures in buildings: A methodological approach. Environ. Sci. Pollut. Res. 27(5), 4612–4629 (2020). https://doi.org/10.1007/s11356-019-06377-3 M.C. Oliveira, M. Iten, Modelling of industrial water circuits with a customized Modelica library. Appl. Therm. Eng. 169 (2020). https://doi.org/10.1016/j.applthermaleng.2019.114840 V.M. Oliveira, F. Navega, L. Cabajo, H. Brás, G. Motta, Sustainable Development Goals – National Report on the Implementation of the 2030 Agenda for Sustainable Development, Portugal (Ministry of Foreign Affairs, 2017) M.C. Oliveira, M. Iten, H.A. Matos, J. Michels, Water-energy nexus in typical industrial water circuits. Water (Switzerland) 11(4) (2019). https://doi.org/10.3390/w11040699 P.A. Owusu, S. Asumadu-Sarkodie, A review of renewable energy sources, sustainability issues and climate change mitigation. Cogent Eng. 3(1) (2016). https://doi.org/10.1080/ 23311916.2016.1167990 S. Pabi, A. Amarnath, R. Goldstein, L. Reekie, Electricity Use and Management in the Municipal Water Supply and Wastewater Industries (Water Research Foundation; EPRl (Electric Power Research Institute), November 2013). Accessed 24 June 2021. [Online]. Available: https://www.sciencetheearth.com/uploads/2/4/6/5/24658156/electricity_ use_and_management_in_the_municipal_water_supply_and_wastewater_industries.pdf PENSAAR 2020 – Uma Estrategia ao Servifo da Populafão: Servifos de Qualidade a um Prefo Sustentavel (APA (Agência Portuguesa do Ambiente); MAOTE (Ministério do Ambiente, Ordenamento do Território e Energia); AdP (Águas de Portugal), April 2015). [Online]. Available: https://apambiente.p1/_zdata/Politicas/Agua/PlaneamentoeGestao/PENSAAR2020/ PENSAAR2020_Relatorio_Vol1.pdf M. Perez-Sanchez, F.J. Sanchez-Romero, H.M. Ramos, P.A. Lopez-Jimenez, Modeling irrigation networks for the quantification of potential energy recovering: A case study. Water (Switzerland) 8(6) (2016). https://doi.org/10.3390/w8060234 M. Pérez-Sánchez, F.J. Sanchez-Romero, H.M. Ramos, P.A. López-Jiménez, Energy recovery in existing water networks: Towards greater sustainability. Water (Switzerland) 9(2) (2017a). https://doi.org/10.3390/w9020097 M. Pérez-Sánchez, F.J. Sánchez-Romero, H.M. Ramos, P.A. López-Jiménez, Optimization strategy for improving the energy efficiency of irrigation systems by micro hydropower: Practical application. Water (Switzerland) 9(10) (2017b). https://doi.org/10.3390/w9100799 Y.A. Phillis, N. Chairetis, E. Grigoroudis, F.D. Kanellos, V.S. Kouikoglou, Climate security assessment of countries. Clim. Chang. 148(1/2), 25–43 (2018). https://doi.org/10.1007/s10584018-2196-0

The Need for Self-Sufficiency and Integrated Water and Energy Management

23

C. Pimentel-Rodrigues, A. Silva-Afonso, Contributions of water-related building installations to urban strategies for mitigation and adaptation to face climate change. Appl. Sci. (Switzerland) 9(17) (2019). https://doi.org/10.3390/app9173575 C. Pimentel-Rodrigues, A. Siva-Afonso, Reuse of resources in the use phase of buildings. Solutions Water 225(1) (2019). https://doi.org/10.1088/1755-1315/225/1/012050 T. Pukšec, P. Leahy, A. Foley, N. Markovska, N. Dui´c, Sustainable development of energy, water and environment systems 2016. Renew. Sustain. Energy Rev. 82, 1685–1690 (2018). https:// doi.org/10.1016/j.rser.2017.10.057 H. Ramos, C. Teyssier, P. López-Jiménez, Optimization of retention ponds to improve the drainage system elasticity for water-energy nexus. Water Resour. Manag. 27(8), 2889 (2013a) H.M. Ramos, C. Teyssier, I. Samora, A.J. Schleiss, Energy recovery in SUDS towards smart water grids: A case study. Energy Policy 62, 463–472 (2013b). https://doi.org/10.1016/ j.enpol.2013.08.014 H.M. Ramos, M. Zilhao, P.A. López-Jiménez, M. Pérez-Sánchez, Sustainable water-energy nexus in the optimization of the BBC golf-course using renewable energies. Urban Water J. 16(3), 215–224 (2019). https://doi.org/10.1080/1573062X.2019.1648529 H.M. Ramos, A. McNabola, P.A. López-Jiménez, M. Pérez-Sánchez, Smart water management towards future water sustainable networks. Water (Switzerland) 12(1) (2020). https://doi.org/ 10.3390/w12010058 Relatório anual dos serviços de Águas e Residuos em Portugal, Volume 1: Caracterizaçaó do setor de águas e residuos (ERSAR (Entidade Reguladora dos Serviços de Água e Residuos), 2019) Relatório do Estado do Ambiente de 2019 (APA (Agência Portuguesa do Ambiente), June 2019). [Online]. Available: https://sniambgeoviewer.apambiente.pt/GeoDocs/geoportaldocs/ rea/REA2019/REA2019.pdf APREN (Associação de Energias Renováveis), Energia Eólica em Portugal 2014, September 2014). [Online]. Available: https://www.apren.pt/contents/documents/portugal-parqueseolicos-201412.pdf REN (Redes Energeticas Nacionais), Dados Técnicos, Technical Data 2019 (2020). [Online]. Available: http://www.centrodeinformacao.ren.pt/PT/InformacaoTecnica/Paginas/ DadosTecnicos.aspx R. Ribeiro, H. Alegre, D. Loureiro, A. Mamade, The role of communication in the deployment of water loss and energy management strategies: The experience of collaborative research with water utilities. Water Sci. Technol. Water Supply 17(4), 1035–1042 (2017). https://doi.org/ 10.2166/ws.2016.198 F. Rodrigues, A. Silva-Afonso, A. Pinto, J. Macedo, A.S. Santos, C. Pimentel-Rodrigues, Increasing water and energy efficiency in university buildings: A case study. Environ. Sci. Pollut. Res. 27(5), 4571–4581 (2020). https://doi.org/10.1007/s11356-019-04990-w I. Savelli, T. Morstyn, Better together: Harnessing social relationships in smart energy communities. Energ. Res. Soc. Sci. 78, 102125 (2021). https://doi.org/10.1016/j.erss.2021.102125 R. Segurado, J.F.A. Madeira, M. Costa, N. Duic, M.G. Carvalho, Optimization of a wind powered desalination and pumped hydro storage system. Appl. Energy 177, 487–499 (2016). https:// doi.org/10.1016/j.apenergy.2016.05.125 R. Segurado, M. Costa, N. Dui´c, Integrated planning of energy and water supply in Islands, in Renewable Energy Powered Desalination Handbook: Application and Thermodynamics, ed. by V.G.B.T.-R.E.P.D.H. Gude (Butterworth-Heinemann, 2018), pp. 331–374. doi: https://doi.org/ 10.1016/B978-0-12-815244-7.00009-X Serra, Nexus Agua – Energia, presented at the Parceria Portuguesa para a Agua no centro das decisoes da Estrategia, Europa 2020 para a Agua, November 16, 2012. [Online]. Available: http://www.ppa.pt/wp-content/uploads/2012/12/Alexandra-Serra.pdf P.M.D. Serra, A. Espirito-Santo, J. Bonifacio, F.S. Relvas, Capacitive Level Smart Sensors in the Management of Wastewater Treatment Processes. Presented at the 2019 IEEE International Symposium on Measurements and Networking, M and N 2019 – Proceedings, 2019. https:// doi.org/10.1109/IWMN.2019.8804993

24

A. R. Silva et al.

P.M.D. Serra, A. Espirito-Santo, M. Magrinho, A steady-state electrical model of a microbial fuel cell through multiple-cycle polarization curves. Renew. Sust. Energ. Rev. 117 (2020). https:// doi.org/10.1016/j.rser.2019.109439 F. Sestini, G. Tselentis, M. Kolodziejski, J. Babot, Living Labs for User-Driven Open Innovation: An Overview of the Living Labs Methodology, Activities and Achievements (European Commission, Directorate-General for the Information and Media, 2009). https://doi.org/10.2759/34481 A.R.C. Silva, Gestão Integrada de Infraestruturas em Contexto Urbano: Implementação de um projeto-piloto de Energias Renováveis na Rede de Distribuição de Água, A.E. Santo, Gestão Integrada de Infraestruturas em Contexto Urbano: Implementação de um projetopiloto de Energias Renováveis na Rede de Distribuição de Água, presented at the ICEUBI2019 (International Congress on Engineering) (Covilhã, Portugal, November 2019) C. Silva, M.J. Rosa, Energy performance indicators of wastewater treatment: A field study with 17 Portuguese plants. Water Sci. Technol. 72(4), 510–519 (2015). https://doi.org/10.2166/ wst.2015.189 A. Silva F. Espirito-Santo J.P. Santos, C. Fael, Water-Energy Nexus: Review of Literature in Management of Integrated Systems Challenges and Opportunities. Case Study in Urban Context (2017) R. Silva, F. Santos, A. Espirito-Santo, J.P. Marques, C.S. Fael, Development of the concept vs prototyping: implementation of a real scale water-energy integrated system, in 2018 IEEE 16th International Conference on Industrial Informatics (INDIN), Porto, (2018a), pp. 1006–1011. https://doi.org/10.1109/INDIN.2018.8471998 R. Silva, F. Santos, A. Espirito-Santo, J.P. Marques, C. Fael, Study of a water-energy integrated system: Challenges of prototyping, in Energy and Sustainability in Small Developing Economies, ES2DE 2018 – Proceedings, (2018b), pp. 63–68. https://doi.org/10.1109/ES2DE.2018.8494315 M. Smid, A.C. Costa, Climate projections and downscaling techniques: A discussion for impact studies in urban systems. Int. J. Urban Sci. 22(3), 277–307 (2018). https://doi.org/10.1080/ 12265934.2017.1409132 S. Stang, H. Wang, K.H. Gardner, W. Mo, Influences of water quality and climate on the waterenergy nexus: A spatial comparison of two water systems. J. Environ. Manag. 218, 613–621 (2018). https://doi.org/10.1016/j.jenvman.2018.04.095 Q. Su, H. Dai, Y. Lin, H. Chen, R. Karthikeyan, Modeling the carbon-energy-water nexus in a rapidly urbanizing catchment: A general equilibrium assessment. J. Environ. Manag. 225, 93– 103 (2018). https://doi.org/10.1016/j.jenvman.2018.07.071 M.R. Teixeira, P. Mendes, E. Murta, L.M. Nunes, Performance indicators matrix as a methodology for energy management in municipal water services. J. Clean. Prod. 125, 108–120 (2016). https:/ /doi.org/10.1016/j.jclepro.2016.03.016 C. Teotonio, M. Rodriguez, P. Roebeling, P. Fortes, Water competition through the “water-energy” nexus: Assessing the economic impacts of climate change in a Mediterranean context. Energy Econ. 85 (2020). https://doi.org/10.1016/j.eneco.2019.104539 M. Thornbush, O. Golubchikov, Smart energy cities: The evolution of the city-energysustainability nexus. Environ. Develop., 100626 (2021). https://doi.org/10.1016/ j.envdev.2021.100626 Understanding the Energy-Water Nexus (Caribbean Community Climate Change Centre, November 14, 2014). https://www.caribbeanclimate.bz/understandingtheenergy-waternexus/. Accessed 24 June 2021 C. Wang, R. Wang, E. Hertwich, Y. Liu, F. Tong, Water scarcity risks mitigated or aggravated by the inter-regional electricity transmission across China. Appl. Energy 238, 413–422 (2019). https://doi.org/10.1016/j.apenergy.2019.01.120 World Energy Scenarios: Composing Energy Futures to 2050 (World Energy Council, Londres, 2013). [Online]. Available: https://www.worldenergy.org/assets/downloads/WorldEnergy-Scenarios_Composing-energy-futures-to-2050_Full-report1.pdf

The Need for Self-Sufficiency and Integrated Water and Energy Management

25

World Energy Scenarios (2016). https://www.worldenergy.org/publications/entry/world-energyscenarios-2016-the-grand-transition World Energy Trilemma 2014: Time to Get Real – The Myths and Realities of Financing Energy Systems. World Energy Council. https://www.worldenergy.org/publications/entry/worldenergy-trilemma-2014-time-to-get-real-the-myths-and-realities-of-financing-energy-systems. Accessed 24 June 2021 World Energy Trilemma Index (World Energy Council, 2019). [Online]. Available: https:// www.worldenergy.org/assets/downloads/WETrilemma_2019_Full_Report_v4_pages.pdf

Rethinking Renewable Energy Development in the Republic of Kazakhstan from the Perspectives of International Relations Ka Wai Christopher Hor

Contents 1 2 3 4 5 6

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electricity Infrastructure and International Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regionalism and Energy Regionalism in Central Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Renewable Energy Development in Kazakhstan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Central Asia Power System and Extra-regional Powers . . . . . . . . . . . . . . . . . . . . . . . . Energy Future for Kazakhstan and Central Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1 Energy Security: National Versus Regional . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Involvement of Extra-regional Entities: Pros and Cons . . . . . . . . . . . . . . . . . . . . . . . . 7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28 30 31 33 35 38 38 39 41 42 43

Abstract

Energy transition represents a disruptive innovation to the conventional energy industry. Most countries in the world have embarked on the energy transition under the dual influence of the realization that hydrocarbon fuels are a limited resource and the growing awareness that they are adversely affecting the planet’s climate. As a petroleum exporting country, the Republic of Kazakhstan’s ambitious decarbonization goal of achieving 15% share of renewable energy in the domestic power generation matrix by 2030 and 50% by 2050 is in sync with global trends. The lesser-known fact is that the geopolitical ramifications due to the rise of renewable energy and electrification have opened a new front for competition between Russia, China, and the United States with regard to their

K. W. C. Hor () Department of International Relations, Al-Farabi Kazakh National University, Almaty, Kazakhstan © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_9

27

28

K. W. C. Hor

strategies in Central Asia, producing significant implications on Kazakhstan’s intra-regional and extra-regional relations. This chapter aims to explore the external factors that are influential to the republic’s energy transition pathways. The latest developments indicate that Kazakhstan and other Central Asian countries have shown a favorable response to a coordinated regional approach amid the Russia-China-US geopolitical triangle for a resilient, cooperative, impactful, and environment-friendly energy future. Keywords

Kazakhstan · Central Asia · Electrification · Energy regionalism · Energy transition · Geopolitics · Neofunctionalism · Renewable energy

1

Introduction

The world has gone through several energy transitions that have marked different industrial eras – from wood to coal, from oil to gas and nuclear. Now, the world is engaged in a massive shift toward renewable energy. According to the report Electrification with Renewables produced by the International Renewable Energy Agency (IRENA) in 2019, the renewable share in power generation would climb from 26% in 2018 to 85% in 2050, with up to 60% coming from variable sources such as solar and wind. The same report also forecasts that by 2050 the share of electricity in total final energy consumption will rise from 19% in 2018 to about 44%, with electricity taking on an increasing role in transport and construction (IRENA 2019). A decarbonized electricity – generated by renewable energy sources for higher affordability, sustainability, and efficiency – is thus the ideal candidate energy carrier to forge the path toward the long-term goals of environmental and climate change mitigation. In fact, nowadays the world is witnessing more early moves in parts of the electricity value chain, by players including oil and natural gas companies, tech giants, and car manufacturers – who, to date, have several partnerships with utilities to offer bundled solutions for renewable electricity supply and smart home devices or distributed energy solutions. The plunging demand for oil wrought by the coronavirus pandemic combined with a savage price war in 2020 has further made the fossil fuel industry a very unattractive proposition to investors. In 2021, with the Group of Seven (G7) announcing that it would aim to “protect our planet by supporting a green revolution that creates jobs, cuts emissions and seeks to limit the rise in global temperatures to 1.5 degrees” (European Council 2021), radical changes are anticipated in the very shape of the energy industry itself. The Republic of Kazakhstan, together with the other former Soviet Central Asian republics, is not immune from the impacts caused by this global trend of energy transition. The notion of Kazakhstan as a bridge transcending geographic regions and civilizations is presented as a justification for the republic’s multiple international engagements. Its multi-vector foreign policy, as reflected in its commitment in

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

29

multilateral organizations as well as bilateral relationships, has benefitted the republic without creating adversaries in international politics. Its vast coal, oil, natural gas, and uranium reserves have enabled it to grow in national capabilities and gain leverage in managing difficult relations with neighboring Russia and China and forming strategic partnership with the United States and the European Union, through provoking their counteractions by means of “external balancing.” Home to the world’s 12th-largest oil reserves, Kazakhstan might seem to have low incentives to invest in diversifying away from fossil fuels. However, Kazakhstan’s “very low” rating in the 2020 Climate Change Performance Index highlights the problems created by its aging inefficient Soviet-era energy infrastructure, with coal alone responsible for covering more than 70% of the republic’s electricity demand. When first President Nursultan Nazarbayev announced in 2012 that the republic would implement green economic policies through its Kazakhstan 2050 Strategy development plan, Kazakhstan aspired to become one of the top 30 competitive developed countries in the world by 2050, while gradually “greening” key economic sectors (Kazakhstan 2050 Strategy 2012). Renewable energy policy has since been developed in a decidedly top-down fashion as best illustrated in signing the Paris Climate Change Agreement in 2016 and hosting the “Future Energy” theme of the EXPO 2017 in Nur-Sultan, its capital city – which was known between 1998 and 2019 as Astana. The republic’s Concept on Transition toward Green Economy also sets a bold timeline to move from under 1% wind and solar energy sourcing when it was adopted in 2013 to 3% by 2020, 10% by 2030, and 50% share of alternative and renewable energy sources – including wind, solar, hydro, and nuclear plants – by 2050, promoting a more decentralized, balanced, and environmentally friendly energy supply system (Decree of the President of the Republic of Kazakhstan 2013). With a new legal framework and an auction system for renewable energy projects adopted to attract foreign investment, the Ministry of Energy estimated that renewable energy production has reached around 1370 megawatts by the end of 2020 – equivalent to 3 billion kilowatt-hours (Satubaldina 2020) – accomplishing the 3% target. In 2021, Kassym-Jomart Tokayev – current president of Kazakhstan – ordered his government to take measures to hasten the pace to increase the share of renewable energy in electricity generation to 15% by 2030, an improvement over the 10% goal set by his predecessor. It is noteworthy that Kazakhstan is one of Russia’s partners in the Eurasian Economic Union (EAEU), a key participant in China’s Belt and Road Initiative (BRI), and a significant component of the US Joint Declaration of Partnership and Cooperation by the Five Countries of Central Asia and the United States of America (C5+1). The great powers’ engagement in the republic’s economic, infrastructural, and energy development has been ever present since its independence in 1991, which inevitably gives shape to the transition of the republic, along with the rest of Central Asia. Although concerns about growing electricity demand and deployment of smart technologies for a resilience future animate a great deal of policymaking for common development, mutual benefits, and win-win results across the region, the “tug of war” within the Russia-China-US geopolitical triangle can, in fact, cause a degree of uncertainty over furthering regional cooperation. Kazakhstan’s energy

30

K. W. C. Hor

future – as an underexplored topic – thus should be timely studied by taking into account of the external factors, in particular with regard to intra-regional relations and the great powers’ drive for global presence, influence, and material benefits in Central Asia. The chapter begins with a brief discussion about electricity infrastructure and international relations, followed by an overview of regional relations in Central Asia. The milestones of renewable energy development in Kazakhstan are then laid out, highlighting the legislative and institutional changes that have facilitated the republic’s energy transition since a decade ago. In the latter part of the chapter, regional electricity cooperation in the format of the Central Asia Power System (CAPS) and contemporary great powers’ geopolitical contest in Central Asia are considered. A conclusion is drawn from Kazakhstan’s foreign policy response when faced with a regional pursuit of decarbonized electricity amid great power rivalry.

2

Electricity Infrastructure and International Relations

With the widening cost advantages and spread of renewable energy, increased electrification is not only coming but has become necessary for a decarbonized global economy with policy goals shifting from seeking fewer kilowatt-hours used in 1970s to fewer tons of CO2 emitted in recent years. The 2017 Energy Technology Perspectives report produced by the International Energy Agency (IEA) asserts that electricity will become the driver of the world’s economies in the twenty-first century, rivalling oil as the dominant energy carrier (IEA 2017). The 2019 IRENA report Electrification with Renewables states that electrification of energy services will be pervasive because the combination of widespread electrification and digital technologies on one hand and renewable power, on the other, can become a central pillar of energy and climate policy by offering much more flexibility and control over demand (IRENA 2019). In this connection, electricity grid is a component vital to facilitate a high level of renewable energy penetration, in particular when countries are motivated by the economic incentives in increasing cross-border electricity trade and integrating markets. A “supergrid,” “megagrid,” or “supersmart grid” has been described as a future grid that interconnects various countries and regions with a high-voltage direct current (HVDC) power grid, making it possible to trade high volumes of convention and renewable electricity across great distances (Khalilpour 2019). In order to overcome intermittency for sustainable renewable power generation, geographic diversity is decisive because renewable energy becomes more predictable as the number of renewable generators connected to the grid increases together with a mixture of sources that complement each other to roughly equal out total energy demand over the day. This is technically possible because continental wind energy tends to peak at night, coastal wind energy tends to peak during the day, and solar can peak at various times over the day, depending on which way it is oriented (Fares 2015). Regionalization of energy infrastructure, energy markets, and energy relations is also entailed to minimize long-distance losses

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

31

of electricity and overcome managerial challenges posed by renewable energy intermittency (Scholten 2018). Research on energy regionalism thus needs to turn to more systematic theory building along with comparisons between regions and across energy sources and infrastructure types. However, one should not neglect that establishing cross-border links and power trading agreements requires deep cooperation and mature bilateral relationships. Technical, legal, and policy frameworks, and a network of supporting institutions, are needed to coordinate grid integration and operations. Where these relationships are absent or soured by geopolitical tensions, it is difficult to create the right enabling environment. The EU’s macro-regional strategies – the initiatives to create multicountry coordination structures to strengthen joint collaboration on common challenges by participating countries and regional authorities since 2009 – demonstrate the significance of creating a precondition for establishing a more secure, sustainable, and competitive energy sector in the format of the Energy Union, targeting to cut greenhouse gas (GHG) emissions and cement the EU’s global leadership role in renewable energy (Nunez Ferrer et al. 2019). The Association of Southeast Asian Nations (ASEAN) likewise displays similar degree of political will but in a nonbinding format when bilateral agreements – especially concerning the power sector or electric power transmission in general – are the main form of regional cooperation in ASEAN today. With no political, legal, or institutional instance imposed upon the member countries, the ASEAN Centre of Energy (ACE) plays as a catalyst for economic growth and integration of the ASEAN region by initiating and facilitating multilateral collaborations to achieve the aspirational target of 23% renewable energy share in the primary energy mix by 2025 (ACE 2018). The Economic Community of West African States (ECOWAS), with the goal to overcome a severe energy crisis in the region, joined hands with external entities to establish the ECOWAS Center for Renewable Energy and Energy Efficiency (ECREEE) in 2010 (Hancock 2015). On the contrary, renewable energy cooperation in Northeast Asia between China, Japan, and South Korea is by far more challenging. Although expansion of functional clean energy cooperation can contribute to the buildup of a regional community, political mistrust and presence of intra-regional competition weaken the collective bargaining power of individual countries. The track record of the United States’ entanglement in the Northeast Asian regional affairs also is significant in terms of negative geopolitical implications for regional energy cooperation, especially the region is of major strategic importance for the United States and its allies – with the inclusion of Japan and South Korea – to contain China (Xiangchengzhen and Yilmaz 2020).

3

Regionalism and Energy Regionalism in Central Asia

There is little doubt that a high level of regionalism is exhibited by the EU, ASEAN, and ECOWAS, in which their member countries voluntarily share part, or all, of their decisional authority at an international level, exemplifying the practice of integration

32

K. W. C. Hor

as described by neofunctionalism. According to neofunctionalism, integration is defined as a process “whereby political actors in several distinct national settings are persuaded to shift their loyalties, expectations and political activities towards a new centre, whose institutions possess or demand jurisdiction over the pre-existing national states. The end result of the process of political integration is a new political community, superimposed over the pre-existing ones” (Haas 1968). One notable doctrine is “ramification” or “spillover” effect: a situation in which a given action – related to a specific goal – creates a situation in which the original goal can be assured only by taking further actions, which in turn create a further condition and a need for more action, and so forth (Lindberg 1963). Although crises may delay or even retard integration, the guiding assumption is that, over time, functional spillovers will lead to political spillovers, and eventually supranational activism will produce an upward trend (Hooghe and Marks 2019). Central Asia has a fundamental deficit in this regard despite geographic proximity, close ethnicity and languages, similar religion, shared history, common resources, and Soviet-era infrastructure – such as water and transport links. Since the dissolution of the Soviet Union, several attempts at regional cooperation among the newly formed Central Asian republics – Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, and Uzbekistan – have failed or achieved limited results. One group of academics looks for external factors, arguing that failure of Central Asian integration is caused by the extra-regional economic dependence of the regional economies and the impact of external powers within the region. They consider that Russia and China – in particular – play a twofold, yet critical role in defining the nature of regional cooperation in Central Asia, in which Russia’s more exclusive and deeper regionalism may be contributing to the fragmentation of Central Asia, while China’s economic power has not been sufficient to foster regional cooperation (Krasnopolsky 2015; Krapohl and Vasileva-Dienes 2019). Another group of academics focuses more on the specific regional obstacles for integration, such as rivalry between Kazakhstan and Uzbekistan for the regional leadership, insufficient presence of Kyrgyzstan and Tajikistan , Turkmenistan’s neutrality, unequal distribution of natural resources, and unresolved issues that persist between the Central Asian countries – including border disputes, water sharing, and trade barriers. They assert that these internal factors have implied a political and regulatory framework that drove the Central Asian countries away from pursuing common goals but their own national priorities (Zhambekov 2015; Karatayev et al. 2016). Despite the relatively poor state of cooperation among the countries, Central Asia contains a number of regional groupings and initiatives. Some of these were promoted by international organizations or states from the neighboring regions, while others reflect a common Soviet past or were initiated by Russia. In this regard, regional cooperation in Central Asia is considered to be advancing in such a way as to include overlapping regional organizations with little to account for in bottom-up developments. Whereas in other Asian areas there has been discussion of “regions without regionalism,” Central Asia had been closer to “regionalism without

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

33

a region” in the context of two types of knowledge: knowing that and knowing how – implying a Central Asian state could have different senses of belonging based on various functional sectors in the region (Kerr 2010; Azizov 2017). Regarding energy relations, power trade among Central Asian countries had declined dramatically since the end of the Soviet era because initiatives that promoted regional interdependence were treated as sensitive issues in the context of the newly gained national independence. According to a speech of Uzbekistan President Shavkat Mirziyoyev in summer 2017, “the reasons for why the unified energy system collapsed are, in fact, even more complicated and knotty than this potted history suggests. But at the heart of the problem is a legacy of distrust among regional leaders, and an aspiration by many of the countries to be fully self-reliant” (Putz 2018). Nowadays a change in trend is spotted as the Central Asian leaders are voicing their support for the resumption of the Central Asia Power System (CAPS); especially hydropower surpluses from Tajikistan and the high potential in the deployment of renewable energy sources are acknowledged as crucial for boosting regional energy security. Recent research conducted by the Asian Development Bank (ADB) shows that energy regionalism – referred to as politically led cooperation and integration initiatives between state and nonstate actors and across territorial units that seek to govern energy relationships and deliver energy-related collective goods (Hancock et al. 2020) – is in sight in Central Asia. A recommendation has been made to analyze successful practices of international wholesale electricity markets – such as the functioning Nord Pool and North American Electricity Grid, as well as the projected South Asian Association for Regional Cooperation (SAARC) Market for Electricity – for their applicability in Central Asia (Shadrina 2019). A news release of the ADB in 2019 even asserted that, while institutional and technical capacity upgrade is of the utmost priority for the CAPS to coordinate the increased power trade between the Central Asian countries, this collective system of electricity management can be expanded to include Afghanistan’s electricity grid with ongoing bilateral power trade between Afghanistan and Tajikistan, Turkmenistan, and Uzbekistan, respectively (ADB 2019a).

4

Renewable Energy Development in Kazakhstan

Kazakhstan’s topography is suitable for the development of renewable energy. More than 50% of the republic, particularly in the northern regions, has average wind speeds of between 4 and 6 m/s, making them suitable for utility-scale wind farm development, whereas southern Kazakhstan receives consistently high levels of solar irradiation (Wheeler 2017). However, as the largest economy in Central Asia with the size exceeding that of Western Europe, the republic’s efforts in decarbonized electrification lack behind Kyrgyzstan and Tajikistan – where hydropower contributes over 90% of their total electricity generation. Highly reliant on its significant fossil fuel resources, Kazakhstan is a net exporter of energy and

34

K. W. C. Hor

energy products. Supported by robust oil production, the republic has earned the reputation as one of the world’s most energy-intensive economies with a strong codependency between its energy trade and economic security. Given the overly influential role of oil sector performance in supporting growth in the non-oil economy, for over a decade, Kazakhstan has been determined to undergo structural and institutional reforms to facilitate the development of a vibrant, modern, and innovative tradable non-oil sector for the republic’s future. In fact, it is one of the first Central Asian countries to have built an institutional framework for transition to a green economy, having adopted an ecological code in 2007 and law on supporting the use of renewable energy in 2009. In an effort to learn from international experiences, Kazakhstan became a member of the IRENA and ratified its charter in 2009. In 2012, Kazakhstan began efforts to build a domestic emissions trading system (ETS) that led to the launch of its ETS pilot phase in January 2013 under a framework based on the EU ETS as a tool to aid the republic in switching to clean, more efficient technologies for industry, manufacturing, and electricity generation. In 2013 Kazakhstan adopted the Concept on Transition toward Green Economy, outlining a future development path guided by green energy policies and establishing the Financial Settlement Centre for Support of Renewable Energy Resources to buy all power from renewable energy operators under 15-year power purchase arrangements. Development of hydropower – especially small hydropower stations, solar, and wind – is an important component of this transition, with which the aging infrastructure decommissioned, energy-efficient equipment installed, and strict environmental standards complied. The republic’s hosting of EXPO-2017 under the theme “Future Energy” in 2017 was the result of years-long efforts to seek a renewable energy future. The same year also saw the introduction of an auction scheme, opening a gateway for local and foreign developers to bring in investment, technical experiences, and latest renewable energy technologies. In 2020 the COVID-19 pandemic produced the biggest shock to Kazakhstan’s economy in almost two decades. Low international and domestic demand for industrial fuel, gasoline, and aviation fuel led to a slowdown in the activity of the three major refineries in Atyrau, Shymkent, and Pavlodar. A dip in gross domestic product (GDP) by 3%, higher unemployment, and a weaker tenge currency were other indicators of a declining economy (The World Bank 2020). With the oil sector becoming the weakest link in the republic’s economic growth, the Astana International Financial Centre (AIFC), International Centre for Green Technologies and Investment, and Astana Hub International IT and Startup Hub are some of the most important recent developments to foster innovation, financial solutions, and services in Nur-Sultan. Fully fledged in 2018, these institutions were created to promote Kazakhstan’s accelerated transition to a green economy by fostering technology and best practices, business development, and investments. AIFC’s Green Finance Center (AIFC-GFC), assisted by the European Bank for Reconstruction and Development (EBRD), in particular was launched to develop and promote green finance in Kazakhstan and neighboring countries – the Commonwealth of Independent States (CIS), the EAEU, the Middle East, West China, Mongolia, and Eastern Europe – and provide an important platform that will

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

35

be key to sustainability efforts in the region. Green bonds, climate bonds, and other forward-looking instruments are expected to stimulate development in the following spheres: “renewable energy,” “energy efficiency,” “pollution prevention and control,” “sustainable management of living natural resources,” “terrestrial and aquatic biodiversity preservation,” “clean transportation,” “sustainable water management,” “climate change adaptation,” “eco-efficient products, production technologies, and processes,” and “clean buildings” (AIFC n.d.). Although the ultimate goal of this transition to a green economy is to enable Kazakhstan to achieve the proclaimed goal in the Kazakhstan 2050 Strategy of entering the top 30 developed countries of the world by 2050 (see Sect. 1), AIFC-GFC can be an impetus to Central Asian cooperation as its regulatory framework welcomes partnership deals and investment options toward the choice of eco-friendly solutions in accordance with the global trends.

5

The Central Asia Power System and Extra-regional Powers

As landlocked countries, the post-Soviet Central Asian countries are heavily dependent on their immediate neighbors for access to the rest of the world. Given that water, energy, and other resources are asymmetrically distributed across these countries, they share a number of problems which can be resolved only in the framework of close cooperation, if not integration. The Central Asia Power System (CAPS) – also known as Unified Energy System of Central Asia – is a regional electricity transmission network created in the 1970s during the Soviet Union era and was constituted from the power networks of present-day Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, and Uzbekistan. The system consisted of mainly 30% hydropower plants of Central Asian upstream and 70% thermal power plants of downstream countries (GICA n.d.), which sought to ensure consumers’ energy supply through a jointly operated regional generation and transmission network across all of Central Asia’s diversified energy resources. The end of the Soviet Union resulted in a progressive decline in a crucial function of the CAPS with each country undertaking energy decision-making independently, eroding established practices – including the physical and technical parameters of the infrastructure. Turkmenistan’s disconnection from the CAPS in 2003 and Tajikistan’s disconnection from Uzbekistan in 2009 further plummeted intra-Central Asian electricity trade by 92% between 1990 and 2016 (Russell 2019). Such progressive decline had caused widespread power outages notably in Kyrgyzstan and Tajikistan in winter and resulted in increased hydrocarbon fuel use by Kazakhstan, Turkmenistan, and Uzbekistan in the summer. Reduced coordination had also spurred the countries to focus on independence of their national power systems until 2017 when representatives from the state energy companies of Central Asia met in Kazakhstan to discuss reviving the unified energy grid with the purpose to keep pace with the region’s economic growth and an increasing demand for power. ADB’s approval of a USD$35 million grant in 2018 to reconnect Tajikistan’s electricity system to the once-unified Central Asian power grid via Uzbekistan, followed

36

K. W. C. Hor

by a technical assistance grant of USD$4.5 million in 2019 to investigate on reconnecting Turkmenistan to the CAPS and extending the system into Afghanistan, set the region “on a faster reform path toward more liberal energy markets with greater private sector participation and investment, increased power connections and exchanges between countries, and a strong commitment to tap renewable energy sources and clean technologies” (ADB 2019b). With an estimated USD$400 billion in cumulative investments up to 2030 for the creation of a regional energy market that could benefit billions of people from Europe to East Asia, the extraregional powers look at the resumption of the CAPS with economic and geopolitical interests. The Eurasian Economic Union (EAEU) is the first relatively successful attempt to establish strong multilateral institutions for post-Soviet regional integration, with Russia being incomparably more powerful than other member countries, which include Armenia, Belarus, Kazakhstan, and Kyrgyzstan. The EAEU has been operating as a customs union since 2011 and as an economic union since 2015. Similar to the EU, one point of resemblance when noting the development of the Eurasian integration is the energy component, in particular the transfer of oil and natural gas from Russia to the other EAEU members is both a necessary redistribution mechanism and a condition for getting the potential positive economic effects inside the EAEU. Besides, the EAEU treaty also supports the formation of a common electricity market (Eurasian Economic Commission 2020). With hydro energy from Kyrgyzstan reaching Russia through the territory of Kazakhstan, the CAPS is not only connected with the EAEU power system but can potentially be part of the transnational decarbonized electricity network stretching from western Siberia all the way west to Lisbon based on the World Trade Organization (WTO) framework, increasing the compatibility level of the EAEU and the EU to create a potential space for a pragmatic cooperation. However, although the EAEU common electricity market is expected to be launched by January 1, 2025, western economic sanctions have significantly reduced the Russian investment potential in facilitating the “return” of nonmember Central Asian countries – namely, Tajikistan, Turkmenistan, and Uzbekistan – to Russia’s economic orbit. The capacity expansion of the CAPS requires sizable investments to repair and modernize the Soviet-era power stations and transmission lines, which is met befittingly by China’s Belt and Road Initiative (BRI) with billions of dollars pledged for Central Asian infrastructure development. The BRI aims at reviving the ancient Silk Road by erecting a Eurasian transport-linked corridor for bringing Chinese exports to Europe via land roads in Central Asia, but power grid construction has become key to the BRI’s energy component since 2015 when Chinese President Xi Jinping proposed to establish a global energy network to meet global power demand with clean and green sources at the United Nations Sustainable Development Summit. With electrification gaining importance and China holding more than half of the world’s solar energy capacity, the BRI represents not only China’s financial investment in Central Asia but its export of expertise in renewable system

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

37

interconnection, smart grid technologies, ultrahigh-voltage (UHV) transmission, and electric vehicles to partner countries. When Xi announced in 2020 at the United Nations General Assembly that China would reach carbon neutrality by 2060, forward-looking energy infrastructure with clean energy sources in Central Asia – such as the CAPS and renewable energy development in Kazakhstan – fits well with the BRI’s win-win proposition for a better world rather than conventional fired power plants. China’s long-distance high-voltage (LDHV) power lines, in particular, are genuine enablers for the expansion of renewable energy, by linking regions of high renewable resource – like windy plains or sunny deserts – with distant demand centers and by better balancing demand and supply between grids and regions. One concern, though, is that intra-regional interests of Central Asia may be undermined due to the expansive geographic scope of the BRI. With the EAEU focusing on Russia’s energy interest and the BRI promoting China’s bid for global grid integration, the United States – through the US Agency for International Development (USAID), on the basis of C5 + 1 – launched the Central Asia Regional Electricity Market (CAREM) project in 2018, serving as a platform for an expanded Central Asia-South Asia regional power market in support of greater economic and social development, as well as commercial transactions of electricity between Central Asia and Afghanistan and possibly Pakistan. Three working groups, technical, legal, and regulatory and market, have been created to address issues related to the development of an enhanced regional electricity market in the region. In 2020, USAID launched the “Power the Future (PTF)” program with the goal to accelerate the regional cost-effective, low emission, climate resilience economies, primarily through the deployment of renewable energy and energy efficiency in all five Central Asian countries (USAID 2020a). The same year has seen a USAID workshop for Central Asian Ministries of Energy, utilities, and power system operators to discuss the conceptual principles of market operations, best practices, and alternatives, considering benefits and potential issues in the CAPS context (USAID 2020b), followed by the launch of a new 5-year USD$38.9 million regional energy program “Power Central Asia” to assist national governments, utilities, and other stakeholders to develop domestic energy market reforms, help strengthen the regional electricity market, and promote greater adoption of clean energy technologies from conventional and renewable sources (US Embassy & Consulate in Kazakhstan 2020). Although the United States is not the most influential international player in the region, the vision of CAREM has long been shared by the Central Asia Regional Economic Cooperation (CAREC) Program , which was launched by the ADB in 2001 to achieve a reliable, sustainable, resilient, and reformed energy market in Central Asia. To sum up, renewable energy development in Kazakhstan is inseparable from all these “foreign-initiated” long-term strategic frameworks for the energy sector of Central Asia, and subsequent trends of development in terms of energy security and energy relations appear to be more in the hand of the great powers rather than of the Central Asian countries concerned.

38

6

K. W. C. Hor

Energy Future for Kazakhstan and Central Asia

As the world enters the 2020s, advancement in technology and environmental consciousness continues to render renewable energy a viable alternative to traditional energy in terms of delivering greater common benefits in various aspects such as economic, social, and environmental. Kazakhstan’s energy objective is to raise the supposed 3% renewable energy sourcing in 2020 to the targeted 15% by 2030.

6.1

Energy Security: National Versus Regional

In a region where there is no common definition of energy security due to an uneven distribution of energy resources, energy cooperation among the Central Asia countries seems to be the right action for all parties involved to move on from the distrust that once existed throughout the 2000s. Similar to the purpose of the European Coal and Steel Community (ECSC) for war prevention by tying the member states’ national industries together, from a neofunctionalist perspective (see Sect. 3), a unified power grid and electricity market in Central Asia can ease tensions by increasing trade, trust, and connectivity between the five countries. In this regard, the bilateral electricity deal between Tajikistan and Uzbekistan in April 2018 for the first time in 9 years signaled the beginning of a new era, followed by an agreement to officially resume the CAPS signed by the heads of the stateowned energy enterprises of the five Central Asian countries and Afghanistan in July 2019. 7 months later, in February 2020, an announcement was made by Afghanistan that the construction of the Afghanistan portion of the CASA-1000 electricity transmission project was set to begin. The entire CASA-1000 project – short for the Central Asia-South Asia power project, which initially began in 2015 – represents an expansion of the CAPS to transmit the excess amounts of hydroelectricity Kyrgyzstan and Tajikistan generate in the summer to Afghanistan and Pakistan. In exchange, Pakistan has initiated talks with Tajikistan and Kyrgyzstan for electricity export during winter. The Central Asian countries, together with their subcontinent neighbors, have begun to turn to each other for security of electricity supply. However, it is noteworthy that Central Asia is made up by member countries and nonmember countries of the EAEU. On this basis, how the CAPS will be regulated, what technical standards should be adapted, and how long the negotiations will take are decisive factors at play in energy regionalism. It is not uncommon that, despite hundreds of documents signed by heads of states and governments of the post-Soviet space, they remained “ink on the paper” and were never implemented (Obydenkova 2011). The risk is that, if the negotiations drag on, the CAPS can fall apart before getting all the countries fully engaged. Poor grid infrastructure is another issue across Central Asia. The blackout in Almaty – Kazakhstan’s most populous city with a population of 1.8 million – on July 15, 2019, for over 3 h highlighted the need of a flexible, resilient, efficient, and highly secure electric grid that allows for the real-time optimization of grid

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

39

operations and resources. The fifth generation of mobile technology (5G), in particular, is an important ingredient which allows the grid to better adapt to the dynamics of renewable energy and distributed generation in real time (Elberg 2019). Without substantial grid modernization across the region, not only will power outages continually occur, but renewable energy may best be deployed as off-grid or stand-alone power generation units for isolated consumers only – irrelevant to the CAPS for cross-border electrification and cooperative energy security. Security risks produced by the unpredictability of other states also cannot be underestimated. Afghanistan, as a key country in the CASA-1000, in particular exemplifies the security problems of expanding the Central Asian unified energy grid to the subcontinent where the transmission line is reachable by the terrorists operating in the war-torn country. The periodic political upheavals in Kyrgyzstan, in addition to the Fergana complexities due to disputes over land and water rights between Kyrgyzstan, Tajikistan, and Uzbekistan, pose similar threats to intraregional electricity cooperation. In this connection, while Kazakhstan’s sizable energy market is crucial to the promotion of regional energy flows, cooperative energy security appears challenging in the absence of multilateral rules and regulations, infrastructure resilience, and political stability. When Kazakhstan announced its Concept for the Development of the Fuel and Energy Sector by 2030 in 2014, understandably, there was no mentioning about the contribution that imports of electricity produced from hydropower in Kyrgyzstan and Tajikistan could make to ensure the reliability of electricity supply in the republic. Neither national electricity laws nor development plans recognized the place that its national electricity systems occupied within the CAPS. On a contrary, the report repeatedly stressed the need for the republic to “guarantee national energy security by reinforcing energy self-sufficiency” (Boute 2015), implying the importance of developing local resources and infrastructure to meet domestic electricity needs. Kazakhstan’s ambitious restructuring initiatives in renewable energy thus can be understood as an attempt to seek after energy nationalism as part of nation-building policies promoted by the Kazakhstan 2050 Strategy. Energy regionalism did not appear to be a viable policy option at first glance until gains from foreign direct investment inflows, great power rivalry, and emission mitigation activities have increasingly become main considerable factors.

6.2

Involvement of Extra-regional Entities: Pros and Cons

Electrification is one of the most cost-effective approaches to decarbonize the broader economy, making the resumption of the CAPS the springboard to modernize the grid infrastructure and introduce more renewable energy into the Central Asian energy matrix. Since the CAPS was originally designed to balance the uneven distribution of fossil fuels in Kazakhstan, Turkmenistan, and Uzbekistan and hydro resources in Kyrgyzstan and Tajikistan through electricity sharing (see Sect. 5), the same principle can be applied to facilitate the shift to green energy systems

40

K. W. C. Hor

of countries by relocating carbon emissions from one country to another that can manage emissions more efficiently and eventually reducing them (Kim 2020). However, Central Asian countries currently do not have the capacity to carry out qualitative development without the help of foreign investments, best practices, and technology. Foreign direct investment (FDI) thus is crucial for an increase in technological innovation capabilities. As the most predominant actor among all other Central Asian countries, Kazakhstan has an undisputable role in attracting investors and connecting opportunities to stimulate functional spillover effects both within and between industries in Central Asia. With the AIFC being the ideal platform to handle an influx of investment in building a sustainable energy future at national and regional level (see Sect. 4), the CAPS is where foreign renewable energy developers and green tech companies can participate in partnership with local experts of different Central Asian countries. The best-case scenario is that the efforts of the main FDI providers – namely, Russia, China, and the United States – are mutually complementary in their respective participation in the CAPS, with Russia’s EAEU providing the market forces and legal framework, China’s BRI contributing the finance and technologies, and the United States’ C5+1 introducing the best practices in integrated resource planning for electricity. The challenge, though, is that the involvement of foreign renewable energy developers and green tech companies cannot be understood independently from the crucial geopolitical importance of the region. None of these entities are free from the geopolitical agenda of their home countries, and great power rivalry can be counterproductive to a series of mutually reinforcing processes. While Russia and China have emphasized the compatibility of the BRI with the EAEU, both would vie for power and influence in Central Asia. The CAREM – as a project of the USAID Mission to Central Asia to build a regional energy market – also plays a role in counterweighting the probability of either Russian or Chinese enterprises dominating the modernization of the CAPS. As summarized by the former US Secretary of State Mike Pompeo during his 2020 visit to Central Asia, the region is “where China and Russia are both present” (Wong 2020). The C5+1 cooperation format of the United States thus reduces not only terrorist threats and problems related to Afghanistan as originally intended but gives the Central Asian countries the possibility to balance Russia and China. Another challenge is that renewable energy presents a new type of diplomacy that exposes Kazakhstan’s deficiency in terms of GHG emissions and indigenous innovation to harness renewable energy. With some major players in energy market already replaced by new ones and the basic premise of considering energy issues changed, it is questionable if Kazakhstan can continually gain leverage from its petroleum with the likes of Russia, China, and the United States in the long term. For its own security and global presence, the republic should consider promoting absolute gains for the great powers rather than provoking their counteractions as its foreign affairs strategies. Ideally, the republic’s foreign policy can become consolidative in the face of a variety of energy interests and energy visions from the great powers, the possible synergy between hydrocarbon-rich and water-rich countries for cooperative energy security in Central Asia in the format of the

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

41

CAPS and a coordinated regional approach to meet growing energy demand in the subcontinent in the format of the CASA-1000. After all, the climate-friendly energy revolution appears to be one of the few issues on which the international community has found common ground. The CAPS, on this basis, represents a neofunctionalist platform that formulates a collective solution to their mutual concerns about energy security, capital inflows, and sustainable development after years of limited cooperation among the five Central Asian countries. It has been reported that, since 2017, Kazakhstan has been increasingly looking for opportunities to boost hitherto weak cooperation with its Central Asian neighbors. The two summits of leaders of Central Asian countries held in Kazakhstan and Uzbekistan in 2018 and 2019 called for strengthening cooperation in the energy sector by expanding opportunities of electricity trade and promoting the development of modern electricity transmission infrastructure. According to Kazakhstan’s Foreign Policy Concept for 2020–2030 – approved by a presidential decree in March 2020 – the republic’s aspiration to be a leading state in Central Asia ranks high among the strategic goals in the field of foreign policy (Chebotaryov 2020). Energy regionalism – in the context of either a collective stance for Central Asian interests or a precondition for future regional integration – is in Kazakhstan’s favor to safeguard itself from being secured in any one specific great power’s orbit when faced with the manifold influences of extraregional entities in a time of energy transition.

7

Conclusion

This chapter addresses Kazakhstan’s renewable energy development against the backdrop of its intra-regional and extra-regional relations amid the global transition to net-zero emissions. As a post-Soviet petroleum exporting country, Kazakhstan’s target of raising the share of renewable energy sources to 15% by 2030 and 50% by 2050 signifies that the era of carbon-intensive energy derived from the burning of fossil fuels is coming to a decline, and a cleaner, more reliable energy future based on electrification with renewable energy sources will be the new normal across Central Asia in the coming decades. The republic’s renewable energy development also implies a review of its multi-vector foreign policy when petroleum politics is becoming less relevant to attract foreign investment and less effective in handing competing international actors, not to mention fossil fuels’ negative impacts on the global efforts in emission reduction and environmental protection. This study reflects that, at times, although energy nationalism appears to be a favorable policy option with local resources and infrastructure developed to meet domestic electricity needs, a coordinated regional approach can be more beneficial to Kazakhstan and its Central Asian neighbors for a resilient, cooperative, impactful, and environmentfriendly energy future in the format of the CAPS when faced with an intensified great power rivalry in the region. The latest development indicates that a growing trend toward regional electricity cooperation and interdependence in Central Asia is in sight.

42

8

K. W. C. Hor

Cross-References

Applications of Machine Learning in the Planning of Electric Vehicle Charging

Stations and Charging Infrastructure: A Review Rethinking Renewable Energy Development in the Republic of Kazakhstan from

the Perspectives of International Relations

Glossary Energy regionalism The term is referred to as politically led cooperation and integration initiatives between state and nonstate actors and across territorial units that seek to govern energy relationships and deliver energy-related collective goods (see Sect. 2, 3 & 6). The Belt and Road Initiative The BRI is a significant development strategy launched by the Chinese government with the intention of promoting economic cooperation that covers countries across Asia, Europe, and Africa (see Sect. 5). The Central Asia Power System The CAPS was created in the 1970s by the Soviet Union and was constituted from the electricity networks of presentday Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, and Uzbekistan. The disintegration of the Soviet Union resulted in a drastic decline in a crucial function of the CAPS (see Sect. 3, 5 & 6). The Central Asia-South Asia power project The CASA-1000 is an important step in facilitating clean power export revenues for the Central Asian countries and by alleviating electricity shortages in the South Asian countries (see Sect. 5 & 6). The Eurasian Economic Union The EAEU is an international economic union and free trade zone established by treaty in 2014 and officially beginning on January 1, 2015. Member countries include Russia, Armenia, Belarus, Kazakhstan, and Kyrgyzstan (see Sect. 5). The Joint Declaration of Partnership and Cooperation by the Five Countries of Central Asia and the United States of America This document – initiated in September 2015 which signified the launch of the C5+1 diplomatic format – underscores the commitment of the participating countries to the principles of sovereignty, independence, and territorial integrity, while simultaneously fostering intra-regional cooperation in many areas (see Sect. 5). The Kazakhstan 2050 Strategy A long-term socioeconomic strategy presented by first president of Kazakhstan Nursultan Nazarbayev in 2013 that facilitates a series of reform. The end goal is for the republic to become one of the top 30 most developed nations by 2050 (See Sect. 1).

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

43

References ASEAN Centre for Energy (ACE), (2018) Study on Regional Renewable Energy Cooperation in ASEAN. ASEAN Centre for Energy. January 2018., http://go.aseanenergy.org/mhUTS. Accessed 14 Jan 2021 Asian Development Bank (ADB), (2019a) ADB grant to support increased cross-border energy trading in Central Asia. News Release. January 7, 2019. https://www.adb.org/news/adb-grantsupport-increased-cross-border-energy-trading-central-asia. Accessed 14 Jan 2021 Asian Development Bank (ADB), (2019b) CAREC countries move a step closer to regional energy market after historic ministers’ meeting in Uzbekistan. News Release. September 20, 2019. https://www.adb.org/news/carec-countries-move-step-closer-regional-energy-marketafter-historic-ministers-meeting. Accessed 14 Jan 2021 Astana International Financial Centre (AIFC), (n.d.) Green Finance Centre. Retrieved on October 30, 2020. https://gfc.aifc.kz/uploads/брошюра_AIFC.pdf. Accessed 14 Jan 2021 Azizov, U., Regional integration in Central Asia: From knowing-that to knowing-how. J. Eurasian Stud. 8(2) (2017). July 2017. https://www.sciencedirect.com/science/article/pii/ S1879366517300040. Accessed 14 Jan 2021 Boute, A. (2015) Towards Secure and Sustainable Energy Supply in Central Asia: Electricity Market Reform and Investment Protection Centre of Energy Law. University of Aberdeen. https://www.energycharter.org/fileadmin/DocumentsMedia/Thematic/ Power_Sector_Reform_in_Central_Asia_2015_en.pdf. Accessed 14 Jan 2021 A. Chebotaryov, Special aspects of Kazakhstan’s new foreign policy concept. Central Asian Bureau for Analytical Reporting. April 23, 2020 (2020), https://cabar.asia/en/special-aspectsof-kazakhstan-s-new-foreign-policy-concept#_ftn8. Accessed 7 Apr 2021 Decree of the President of the Republic of Kazakhstan, Concept for transition of the Republic of Kazakhstan to Green Economy. May 30, 2013 (2013), https://www.oneplanetnetwork.org/ sites/default/files/kazakhstan_concept_for_transition_of_the_republic_of_kazakhstan_to_ green_economy.pdf . Accessed 14 Jan 2021 R. Elberg, 5G and power grid singularity – Not if, but when. Utility Analytic Institute. August 13, 2019 (2019). https://utilityanalytics.com/2019/08/5g-and-power-grid-singularity-not-if-butwhen/. Accessed 14 Jan 2021 Eurasian Economic Commission, EEC creates a common electricity market. New Release. April 7, 2020 (2020), http://www.eurasiancommission.org/en/nae/news/Pages/07-04-2020-1.aspx. Accessed 12 Apr 2021 European Council, 2021 G7 Leaders’ communiqué: Our Shared Agenda for Global Action to Build Back Better. Press Release (2021), https://www.consilium.europa.eu/en/press/press-releases/ 2021/06/13/2021-g7-leaders-communique/. Accessed 16 June 2021 R. Fares, Renewable energy intermittency explained: Challenges, solutions, and opportunities. Scientific American. March 11, 2015. (2015), https://blogs.scientificamerican.com/plugged-in/ renewable-energy-intermittency-explained-challenges-solutions-and-opportunities/. Accessed 14 Jan 2021 Global Infrastructure Connectivity Alliance (GICA), Central Asian Power System (CAPS) (n.d.). https://www.gica.global/initiative/central-asian-power-system-caps. Accessed 14 Jan 2021 E. Haas, The Uniting of Europe. (Stanford University Press, Stanford, 1968, First Published in 1958) K.J. Hancock, Energy regionalism and diffusion in Africa: How political actors created the ECOWAS Center for Renewable Energy and Energy Efficiency. Energy Res. Soc. Sci. 5, 2015 (2015) https://www.academia.edu/11002796/Energy_regionalism_and_diffusion_ in_Africa_How_political_actors_created_the_ECOWAS_Center_for_Renewable_Energy_and_ Energy_Efficiency . Accessed January 14, 2021

44

K. W. C. Hor

K. Hancock et al., The Politics of Energy Regionalism, in The Oxford Handbook of Energy Politics, ed. by K. J. Hancock, J. E. Allison, (Oxford University Press, 2020) https:/ /www.researchgate.net/publication/342819652_The_Politics_of_Energy_Regionalism. Accessed 15 Apr 2021 L. Hooghe, G. Marks, Grand theories of European integration in the twenty-first century. J. Eur. Publ. Policy, https://doi.org/10.1080/13501763.2019.1569711. January 2019 (2019). https:// www.tandfonline.com/doi/full/10.1080/13501763.2019.1569711. Accessed 14 Jan 2021 International Energy Agency (IEA), Energy Technology Perspectives 2017: Catalysing Energy Technology Transformations (2017), https://webstore.iea.org/download/direct/1058. Accessed 14 Jan 2021 International Renewable Energy Agency (IRENA), Electrification with Renewables: Driving the transformation of energy services (2019), https://irena.org/-/media/Files/IRENA/Agency/ Publication/2019/Jan/IRENA_RE-Electrification_SGCC_2019_preview.pdf. Accessed 14 Jan 2021 M. Karatayev et al., Renewable energy technology uptake in Kazakhstan: Policy drivers and barriers in a transitional economy. Renew. Sustain. Energy Rev. 66 (2016). December 2016, https://www.sciencedirect.com/science/article/abs/pii/S1364032116303847. Accessed 14 Jan 2021 D. Kerr, Central Asian and Russian perspective on China’s strategic emergence. Int. Aff., 86(1), 127–152 (2010). https://onlinelibrary.wiley.com/doi/epdf/10.1111/j.1468-2346.2010.00872.x. Accessed 15 Apr 2021 K.R. Khalilpour, Chapter 6 – Stranded Renewable Energies, Beyond Local Security, Toward Export: A Concept Note on the Design of Future Energy and Chemical Supply Chains, in (2019) Polygeneration with Polystorage for Chemical and Energy Hubs, ed. by K. R. Khalilpour, (Elsevier Inc., 2019) https://www.sciencedirect.com/topics/engineering/supergrids. Accessed January 14, 2021 S.D. Kim, In Central Asia, a Soviet-era electricity network could power future energy sharing. The Asian Development Blog. October 8, 2020 (2020), https://blogs.adb.org/blog/central-asiasoviet-era-electricity-network-could-power-future-energy-sharing. Accessed 15 June 2021 S. Krapohl, A. Vasileva-Dienes, The region that isn’t: China, Russia and the failure of regional integration in Central Asia. Asia Europe J.. May 28, 2019 (2019), https://link.springer.com/ article/10.1007/s10308-019-00548-0. Accessed 14 Jan 2021 P. Krasnopolsky, Major powers and regionalism in Central Asia. CASI research seminar. American University of Central Asia. January 27, 2015 (2015) . http://web.isanet.org/Web/Conferences/ GSCIS%20Singapore%202015/Archive/8e26ebb3-f4a3-4a13-ac59-adb0ee445aa5.pdf . Accessed 14 Jan 2021 L. Lindberg, The Political Dynamics of European Integration (Stanford University Press, Stanford, 1963) J. Nunez Ferrer et al., Comparative study on the governance structure and energy policies in EU macro-regional strategies. CPES Research Report, No.2019/2. July 2019 (2019), https://www. ceps.eu/wp-content/uploads/2019/07/RR2019-02_EU-macroregional-strategies.pdf. Accessed 14 Jan 2021 A. Obydenkova, Comparative regionalism: Eurasian cooperation and European integration. The case for neofunctionalism? J. Eurasian Stud. 2(2) (2011). July 2011, https:// www.sciencedirect.com/science/article/pii/S187936651100008X. Accessed 14 Jan 2021 C. Putz, Tajikistan to Plug Back into Central Asian Power Grid. The Diplomat. November 28, 2018 (2018), https://thediplomat.com/2018/11/tajikistan-to-plug-back-into-central-asian-power-grid/ . Accessed 14 Jan 2021 M. Russell, Connectivity in Central Asia: Reconnecting the Silk Road. European Parliament. April 2019 (2019), https://www.europarl.europa.eu/RegData/etudes/BRIE/2019/637891/ EPRS_BRI(2019)637891_EN.pdf. Accessed 14 Jan 2021 A. Satubaldina, Nine renewable energy projects to be launched in Kazakhstan by December. The Astana Times. May 19, 2020 (2020), https://astanatimes.com/2020/05/nine-renewable-energyprojects-to-be-launched-in-kazakhstan-by-december/. Accessed 14 Jan 2021

Rethinking Renewable Energy Development in the Republic of Kazakhstan. . .

45

D. Scholten, The geopolitics of renewables – An introduction and expectations, in The Geopolitics of Renewables, ed. by D. Scholten, (Springer, 2018), pp. 1–33 E. Shadrina, Renewable energy in central Asian economies: Role in reducing regional energy insecurity. Asian Development Bank Institute. August 2019. https://www.adb.org/sites/default/ files/publication/522901/adbi-wp993.pdf. Accessed 14 Jan 2021 Kazakhstan 2050 Strategy, Address by the President of the Republic of Kazakhstan, leader of the nation, N. A. Nazarbayev. December 14, 2012 (2012), http://www.akorda.kz/en/addresses/ addresses_of_president/address-by-the-president-of-the-republic-of-kazakhstan-leader-of-thenation-nnazarbayev-strategy-kazakhstan-2050-new-political-course-of-the-established-state. Accessed 14 Jan 2021 The World Bank, Central Asia Electricity Trade brings economic growth and Fosters Regional Co-Operation. October 20, 2020 (2020), https://www.worldbank.org/en/news/feature/2020/ 10/20/central-asia-electricity-trade-brings-economic-growth-and-fosters-regional-cooperation. Accessed 14 Jan 2021 United States Agency for International Development (USAID), Power the future. January 25, 2020 (2020a), http://ptfcar.org/en/power-the-future-2/. Accessed 14 Jan 2021 United States Agency for International Development (USAID), Central Asia Power System Market Alternatives and Methodology for Evaluating Economic Benefits from Regional Trade. February 13, 2020. (2020b), https://ptfcar.org/carem/2020/02/13/usaid-introduces-energy-marketmodels-for-central-asia-2/ & https://ptfcar.org/carem/wp-content/uploads/2020/02/Session-3Activities-and-resources-needed-for-CAREM_Final.pdf. Accessed 14 Jan 2021 US Embassy & Consulate in Kazakhstan, USAID launches ‘Power Central Asia’ to strengthen regional energy sector co-operation. Press Release. October 1, 2020. (2020), https:// kz.usembassy.gov/usaid-launches-power-central-asia-to-strengthen-regional-energy-sectorcooperation/. Accessed 14 Jan 2021 E. Wheeler, Kazakhstan’s renewable energy quest. The Diplomat. May 2, 2017. (2017), https:// thediplomat.com/2017/05/kazakhstans-renewable-energy-quest/. Accessed 14 Jan 2021 E. Wong, US faces tough ‘Great game’ against China in Central Asia and beyond. The New York Times. February 13, 2020. (2020), https://www.nytimes.com/2020/02/13/world/asia/ china-great-game-central-asia-trump.html. Accessed 14 Jan 2021 M. Xiangchengzhen, S. Yilmaz, Renewable energy cooperation in Northeast Asia: Incentives, mechanisms and challenges. Energ Strat Rev 29. May 2020. (2020), https:// www.sciencedirect.com/science/article/pii/S2211467X20300225. Accessed 14 Jan 2021 N. Zhambekov, Central Asian Union and the obstacles to integration in Central Asia. The Central Asia-Caucasus Analyst. January 7, 2015. (2015), https://www.cacianalyst.org/publications/ analytical-articles/item/13116-central-asian-union-and-the-obstacles-to-integration-in-centralasia.html. Accessed 14 Jan 2021

Further Readings Decree of the President of the Republic of Kazakhstan, Concept for transition of the Republic of Kazakhstan to Green Economy. May 30, 2013 (2013) International Renewable Energy Agency, A New World: The Geopolitics of the Energy Transformation (IRENA, 2019) M. Laruelle (ed.), China’s Belt and Road Initiative and its Impact on Central Asia. (George Washington University, Washington, DC, 2018) I. Levine, US Policies in Central Asia: Democracy, Energy, and the War on Terror (Routledge, London, 2016) D. Scholten (ed.), The Geopolitics of Renewables. (Springer, 2018) Vinokurov, E., Introduction to the Eurasian Economic Union (Palgrave Macmillan, 2018)

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations Jun Zheng, Guchuan Zhu, and Meng Li

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Modeling Aggregate of TCL Populations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Dynamics of TCLs Under Thermostat-Based Deadband Control with Forced Switching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 A PDE Aggregate Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 An Equivalent PDE Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Aggregate Power Tracking Control Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Well-Posedness and Stability of the Closed-Loop System . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Well-Posedness Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Simulation Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

48 51 51 52 54 55 57 57 60 63 66 75

This work is supported in part by NSFC under grant NSFC-11901482 and in part by NSERC under grant RGPIN-2018-04571. J. Zheng School of Mathematics, Southwest Jiaotong University, Chengdu, Sichuan, China Department of Electrical Engineering, Polytechnique Montreal, Montreal, QC, Canada e-mail: [email protected] G. Zhu () Department of Electrical Engineering, Polytechnique Montreal, Montreal, QC, Canada e-mail: [email protected] M. Li College of Big Data and Internet, Shenzhen Technology University, Shenzhen, Guangdong, China © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_11

47

48

J. Zheng et al.

Abstract

This chapter presents the development and the analysis of a scheme for aggregate power tracking control of heterogeneous populations of thermostatically controlled loads (TCLs) based on the theory and techniques for partial differential equation (PDE) control. By employing a thermostat-based deadband control with forced switching in the operation of individual TCLs, the aggregated dynamics of TCL populations are governed by a pair of Fokker-Planck equations coupled via the actions located both on the boundaries and in the domain. The technique of input-output feedback linearization is used for the design of aggregate power tracking control, which results in a nonlinear system in a closed loop. As the considered setting is a problem with time-varying boundaries, well-posedness assessment and stability analysis are carried out to confirm the validity of the developed control scheme. A simulation study is conducted to evaluate the effectiveness and the performance of the proposed approach. Keywords

Partial differential equations · Thermostatically controlled loads · Aggregate power control · Well-posedness · Stability

1

Introduction

Control of large populations of thermostatically controlled loads (TCLs), such as refrigerators, air conditioners, space heaters, hot water tanks, etc., has drawn a considerable attention in the recent literature; see, just to cite a few, Angeli and Kountouriotis (2012), Callaway (2009), Laparra et al. (2020), Liu et al. (2016), Liu and Shi (2016), Mahdavi et al. (2017), Tindemans et al. (2015), Totu et al. (2017), and Zhang et al. (2013). The development in this field is mainly motivated by the fact that most of the TCLs exhibit flexibilities in power demand for their operation and elasticities in terms of performance restrictions. Therefore, a large-scale TCL population can be managed as a demand response (DR) resource to provide such features as power peak shaving and valley filling, as well as absorbing fluctuations of renewable energies and enabling dynamic pricing schemes in the context of the smart grid (Deng et al. 2015; Palensky and Dietrich 2011). Most of the TCL population control techniques developed in recent years can be considered as extensions of the traditional method of direct load control (DLC) by exploiting the bidirectional communication capability enabled by the smart grid paradigm. This is also one of the basic assumptions required in the present work. Moreover, thermostatic-based deadband control is one of the most used schemes in the operation of TCLs, which is convenient for supporting two basic types of switching control, namely, fast commutation for power control and slow energy consumption regulation. A fast commutation between ON and OFF states can achieve instantaneous power consumption controls, which can provide, for example,

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

49

auxiliary services, such as frequency control and load following (Liu et al. 2016; Liu and Shi 2016; Mahdavi et al. 2017; Zhang et al. 2013). The long-term energy consumption can be regulated through set-point control, deadband width variation, switching duty cycle adjustment, etc., so that a large population of TCLs can be managed to provide the DR capability (Angeli and Kountouriotis 2012; Bashash and Fathy 2013; Callaway 2009; Soudjani and Abate 2015; Totu et al. 2017). The focus of this work is put on aggregate power control of heterogeneous TCL populations based on the dynamic model of the population, which is in general composed of the microscopic dynamics of individual TCLs and the macroscopic aggregate model of the population. At the level of individual TCLs, the equivalent thermal parameter (ETP) model is the most adopted one for describing the temperature dynamics of TCLs (see, e.g., Koch et al. 2011; Liu et al. 2016; Radaideh et al. 2019; Zhang et al. 2013). At the aggregate level, a pioneer work on the modeling of the aggregated dynamics of TCL populations can be traced back to the one reported by Malhamé and Chong (1985), where a continuum model described by partial differential equations (PDEs) was introduced for the first time. Specifically, it is shown that under the assumption of homogeneous TCL populations with all the TCLs modeled by thermostat-controlled scalar stochastic differential equations, the dynamics of a population can be expressed by two coupled Fokker-Planck equations describing the evolution of the distribution of the TCLs in ON and OFF states, respectively. The PDE-based aggregate model has drawn much attention in the recent literature for different DR applications (Angeli and Kountouriotis 2012; Bashash and Fathy 2013; Callaway 2009; Ghaffari et al. 2015; Laparra et al. 2020; Moura et al. 2013; Tindemans et al. 2015; Totu et al. 2017; Zheng et al. 2020) and has led to mathematically rigorous results obtained by leveraging the rich theory from PDEs. This continuum perspective has also enabled more abstract formulations that are independent of the particular solutions. A more generic stochastic hybrid system model applicable to a wider class of responsive loads is developed by Zhao and Zhang (2018). It should be noted that the diffusive term in the Fokker-Planck equations can capture the heterogeneity of the population at a certain level (Moura et al. 2013). Whereas, a heterogeneous population with high diversity can be divided into a finite number of groups, each of which represents a population with limited variation in its heterogeneity (Ghaffari et al. 2015). Another popular and widely adopted model for describing the aggregated dynamics of TCL populations is based on the Markov chains or state bins representation. The original state-bin model proposed by Koch et al. (2011) is described by a finitedimensional linear time-invariant (LTI) system, which has been further extended to adapt to TCLs of more generic dynamics, e.g., higher-order systems, with different control mechanisms, and to cope with parameter variations (e.g., time-varying ambient temperature) (Hao et al. 2015; Liu et al. 2016; Liu and Shi 2016; Mahdavi et al. 2017; Radaideh et al. 2019; Soudjani and Abate 2015; Zhang et al. 2013). It is interesting to note that, as pointed out by Zhao and Zhang (2018), the statebin model can be seen as the discretization of a PDE-based model over a range of temperature. However, as most of the PDE aggregate models are nonlinear, or semi-linear with a more accurate terminology in the PDE theory, and time-varying,

50

J. Zheng et al.

a PDE model has to be linearized around an operational point, e.g., the set-point corresponding to a reference temperature, with fixed parameters in order to obtain an LTI system. From this point of view, the PDE-based continuum model is indeed a more generic representation of the aggregated dynamics of TCL populations, which can incorporate nonlinearity, time-varying parameters, as well as higher-order ETP models (Zhao and Zhang 2018), within a unified formulation. Indeed, discretizing, and eventually linearizing, the PDE model to obtain a lumped finite-dimensional linear system, also referred as early-lumping, is an approach widely adopted in the control of TCL populations (Angeli and Kountouriotis 2012; Bashash and Fathy 2013; Callaway 2009; Ghaffari et al. 2015; Totu et al. 2017). With this method, the complexity in control design and implementation, as well as the expected performance, should be similar to that carried out with state-bin-based models. Another approach, which allows better taking advantage of PDE-based models, is to perform the control design directly with the original PDE systems. This approach, referred as late-lumping, can preserve the basic property of the original system, in particular the closed-loop stability, and the performance when the designed control is applied (Balas 1978). Another attractive feature of this approach is that the control implementation is computationally tractable and does not suffer from the diminution of discretization step-size, which is often needed for performance improvement. The application of the late-lumping method to the control of TCL populations has been demonstrated recently by Ghanavati and Chakravarthy (2018) and Zheng et al. (2020). In the control design of this work, we adopt the approach of input-output linearization for distributed parameter systems (Christofides and Daoutidis 1996; Christofides 2001; Maidi and Corriou 2014; Zheng et al. 2020), which can avoid discretizing and locally linearizing the model by dealing directly with the original nonlinear PDEs, while coping with time-varying parameters, such as set-point and ambient temperature in a straightforward manner. Specifically, we choose first the aggregate power of the population as the output of the system. We then design a PDE control that renders the input-output dynamics to be a finite dimensional LTI system (see, e.g., Christofides and Daoutidis 1996; Christofides 2001; Maidi and Corriou 2014). Note that this approach is inherited from the well-established technique of input-output linearization for finite dimensional nonlinear system control (Khalil 2002; Krstic et al. 1995; Lévine 2009). The developed control scheme is validated through the application to a typical scenario considered in many works reported in the recent literature (e.g., Angeli and Kountouriotis 2012; Bashash and Fathy 2013; Callaway 2009; Ghaffari et al. 2015; Ghanavati and Chakravarthy 2018; Hao et al. 2015; Koch et al. 2011; Laparra et al. 2020; Liu et al. 2016; Liu and Shi 2016; Mahdavi et al. 2017; Moura et al. 2013; Radaideh et al. 2019; Soudjani and Abate 2015; Tindemans et al. 2015; Totu et al. 2017; Zhang et al. 2013; Zheng et al. 2020), namely, aggregate power control of a large population of heating/cooling appliances that may spread over a wide area. With the thermostat-based deadband control considered in the present work, switchings may happen both in the domain and on the boundaries. Furthermore, the boundaries are time-varying, which represents a more generic setting. It should

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

51

be noted that in addition to the appearance of nonlinear nonlocal terms in the closed-loop system, the nature of time-varying boundaries will make control design, well-posedness assessment, and stability analysis more challenging. Nevertheless, we found that this difficulty could be overcome by transforming the original system into a one with fixed boundaries and, hence, the techniques developed by Zheng et al. (2020) can be applied in well-posedness assessment and stability analysis of the problem considered in this work. Finally, it is interesting to note that the considered setting covers the case where a number of TCLs may move beyond the current deadband boundaries when set-points are changed rapidly, and, hence, they are forced to switch in an unpredictable manner. In this chapter, we will present a rigorous well-posedness and stability analysis, which can provide a theoretical tool for solving this open problem, which cannot be handled by the existing methods. Dealing with a generic setting of TCL populations is commonly adopted in theoretical researches and practical applications, while carrying out control design and analysis in a rigorous manner constitutes a main contribution of this work to the related literatures. In the rest of this chapter, Sect. 2 presents the dynamical model of individual TCLs under thermostat-based deadband control and the PDE aggregate model of heterogeneous TCL populations with moving boundary conditions. Control design is detailed in Sect. 3. Well-posedness analysis and stability assessment of the closedloop system are presented in Sects. 4.1 and 4.2, respectively. A simulation study for evaluating the performance of the developed control scheme is carried out in Sect. 5, followed by some concluding remarks provided in Sect. 6. Notations on function spaces and details of certain technical development are given in appendices.

2

Modeling Aggregate of TCL Populations

2.1

Dynamics of TCLs Under Thermostat-Based Deadband Control with Forced Switching

We consider a large population of TCLs operated in ON/OFF mode. Denote by xi the temperature of the i-th load in a population of size N whose dynamics are described by (see, e.g., Bashash and Fathy 2013; Callaway 2009; Koch et al. 2011) 1 dxi = (xie − xi − si Ri Pi ) , i = 1, . . . , N, dt Ri C i where xie is the ambient temperature, Ri is thermal resistance, Ci is thermal capacitance, Pi is thermal power, and the control signal si (t) takes a binary value from {0, 1}, representing OFF and ON states, respectively. Note that considering cooling systems is only for the sake of notational simplicity. As shown in Fig. 1, there are two types of switchings in a generic setting of thermostat-based control. In the regular operation regime, switchings will occur when the temperature reaches the boundaries of the deadband (x, x), where x and

52

J. Zheng et al.

Fig. 1 Temperature profile starting from a reference point (t0 , x0 ) for a TCL under thermostat-based deadband control with time-varying boundaries and forced switching

x are, respectively, the lower and upper temperature bounds. Forced switchings can be applied inside the deadband for fast power control, as mentioned earlier. Forced switchings can be randomly generated in order to, e.g., avoid the possible power demand oscillations due to synchronization (Callaway 2009; Laparra et al. 2020; Totu et al. 2017; Zheng et al. 2020). Spontaneous interrupts of the operational state of TCLs from the participants of a DR program for any particular motivations can also be classed as forced switchings. The forced switching signal for the i-th TCL, denoted by ri (t), takes also a binary value from {0, 1}, with 1 representing the occurrence of switching and 0 otherwise. The control signal of the i-th TCL integrating the two types of switching actions can then be expressed as

si (t) =

⎧ ⎪ ⎪ ⎨1,

if x ≥ x;

0, if x ≤ x; ⎪ ⎪ ⎩(s (t − ) ∧ r (t)) + (s (t − ) ∨ r (t)), otherwise; i i i i

where “∧” and “∨” represent the Boolean operations AND and OR, respectively, and “+” is the one-bit binary addition with overflow.

2.2

A PDE Aggregate Model

Throughout the chapter, let R, R≥0 , R+ denote the set of all real numbers, nonnegative real numbers, and positive real numbers, respectively. For T ∈ R+ , let QT = (0, 1) × (0, T ), QT = [0, 1] × [0, T ], Q∞ = (0, 1) × R+ and Q∞ = [0, 1] × R≥0 . For t ∈ [0, T ], let ΩT = (x(t), x(t)) × (0, T ) and Ω T = [x(t), x(t)] × [0, T ]. For t ∈ R+ , let Ω∞ = (x(t), x(t)) × R+ and Ω ∞ = [x(t), x(t)] × R≥0 . We denote by ws and wss the first- and second-order derivatives of a function w w.r.t. its argument s, respectively. 1 xe − x − sj RP , j = 0, 1, where R, C, and P are, respectively, Let αj = RC the representative thermal resistance, thermal capacitance, and thermal power of the population, and the control signals s0 (t) = 0 and s1 (t) = 1 represent OFF and ON states. We recall that xe is the time-varying ambient temperature.

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

53

As mentioned earlier, power consumption control of a TCL population under deadband control can be achieved by moving the lower and upper temperature boundaries (Bashash and Fathy 2013; Callaway 2009). The width of the deadband is denoted by Δx = x−x. Fixing Δx, the control input can be chosen as u(t) = x˙ = x, ˙ where x˙ := dx . Denote by δ (x, t) (respectively δ (x, t)) the net flow due to 0→1 1→0 dt the switch of loads at (x, t) from OFF-state to ON-state (respectively from ONstate to OFF-state) inside of the temperature bounds. Under the assumption of mass conservation, that is, the size of the population is constant, there must be δ0→1 (x, t) = −δ1→0 (x, t) := δ(x, t). Let w(x, t) and v(x, t) be the distribution of loads [number of loads/◦ C] at temperature x and time t, over the ON and OFF states, respectively. Then, the dynamics of distribution evolution of a heterogeneous TCL population can be expressed by the following forced Fokker–Planck equations (Bashash and Fathy 2013; Callaway 2009): wt (x, t) = (βwx (x, t) − (α1 (x) − u(t)) w(x, t))x + δ(x, t), (x, t) ∈ Ω∞ ,

(1a)

vt (x, t) = (βvx (x, t) − (α0 (x) − u(t)) v(x, t))x − δ(x, t), (x, t) ∈ Ω∞ ,

(1b)

with β a positive constant denoting the diffusion coefficient, satisfying the following boundary conditions: B1 [w](x, t) := βwx (x, t) − α1 (x)−u(t) − x˙ w(x, t) = σ (t), t ∈ R≥0 , B1 [w](x, t) := βwx (x, t) − α1 (x)−u(t) − x˙ w(x, t) = σ (t), t ∈ R≥0 , B2 [v](x, t) := βvx (x, t) − α0 (x)−u(t) − x˙ v(x, t) = −σ (t), t ∈ R≥0 , B2 [v](x, t) := βvx (x, t) − α0 (x)−u(t) − x˙ v(x, t) = −σ (t), t ∈ R≥0 ,

(2a) (2b) (2c) (2d)

where σ (t) and σ (t) represent the net flux due to the switch of loads on the boundaries x and x, respectively, including the flux due to the switch of loads beyond the current deadband boundaries when set-points are changed rapidly. The initial value conditions are given by w(x, 0) = w 0 (x), v(x, 0) = v 0 (x), x ∈ (x(0), x(0)).

(3)

Clearly, the boundary conditions given in (2) describe the coupling of the w- and v-dynamics on the boundaries. Moreover, as shown in Proposition 5, the property of mass conservation can be assured, i.e., for any t ∈ R≥0 , it holds that Nagg (t) =

x(t)

(w(x, t) + v(x, t)) dx =

x(t)

which represents the size of the population.

x(0)

w 0 (x) + v 0 (x) dx,

x(0)

(4)

54

J. Zheng et al.

2.3

An Equivalent PDE Model

Note that the system defined by (1) and (2) is a problem with moving boundaries for which it will be difficult to perform control system design and analysis. For this reason, we transform this system into an equivalent model with fixed boundaries. x−x Specifically, let z = Δx , and then x(t) ≤ x ≤ x(t) makes 0 ≤ z ≤ 1. Thus, w(x, t) = w(zΔx + x, t) := w (z, t), δ(x, t) = δ(zΔx + x, t) := δ (z, t) and αj (x, t) = αj (zΔx + x, t) := αj (z, t), j = 0, 1, which is expressed by αj (z, t) = with ze =

xe −x Δx .

1 (ze − z)Δx − sj RP , j = 0, 1, RC

Moreover,

wx =

1 1 x˙ w z , wxx = . w zz , wt = w t − w z × 2 Δx Δx (Δx)

Then the system given in (1) becomes

β + α1 − u − x˙ w δ , (z, t) ∈ Q∞ , w z − Δx z

β 1 vz − v − α0 − u − x˙ δ , (z, t) ∈ Q∞ . vt = Δx Δx z

1 w t = Δx

The boundary conditions in (2) become β ˙ w z (1, t) − α1 (1, t) − u(t) − x(t) w (1, t) = σ (t), Δx β w z (0, t) − α1 (0, t) − u(t) − x(t) ˙ w (0, t) = σ (t), Δx β ˙ vz (1, t) − α0 (1, t) − u(t) − x(t) v (1, t) = −σ (t), Δx β vz (0, t) − α0 (0, t) − u(t) − x(t) ˙ v (0, t) = −σ (t), Δx = Let β

t ∈ R≥0 , t ∈ R≥0 , t ∈ R≥0 , t ∈ R≥0 .

1 1 1 1 αj = Δx αj , σ = Δx σ, σ = Δx σ and Δx u, 0 0 0 0 w (z), v (x) = v (zΔx + x) := v (z). Noting that u(t)

β , (Δx)2

u =

w 0 (zΔx + x) := ˙ x(t), we obtain the following system:

w z − ( α1 − 2 u) w z + δ , (z, t) ∈ Q∞ , w t = β vt = β vz − ( α0 − 2 u) v z − δ , (z, t) ∈ Q∞ ,

w 0 (x) = = x(t) ˙ =

(5a) (5b)

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

55

w β z (1, t) − ( α1 (1, t) − 2 u(t)) w (1, t) = σ (t), t ∈ R≥0 ,

(5c)

w β z (0, t) − ( α1 (0, t) − 2 u(t)) w (0, t) = σ (t), t ∈ R≥0 ,

(5d)

β vz (1, t) − ( α0 (1, t) − 2 u(t)) v (1, t) = − σ (t), t ∈ R≥0 ,

(5e)

β vz (0, t) − ( α0 (0, t) − 2 u(t)) v (0, t) = − σ (t), t ∈ R≥0 ,

(5f)

w (z, 0) = w 0 (z), v (z, 0) = v 0 (z), z ∈ (0, 1).

(5g)

It should be noted that transforming the original system (1) into (5) is only for the purpose of simplifying control design and analysis. Whereas, control implementation and operation will be performed in the original frame with moving temperature bounds.

3

Aggregate Power Tracking Control Design

In the considered power load tracking control of the whole TCL population, the weighted total power consumption is chosen as the system output y(t) =

P η

x(t)

(ax + b(t)) w(x, t) dx,

(6)

x(t)

where η is the load efficiency, a is a nonzero constant, and b(t) is a C 2 -function. Note that the weighting function ax + b with a = 0 introduced in system output defined above is to guarantee that the input-output dynamics of the system are well defined in terms of characteristic index (Christofides and Daoutidis 1996; Christofides 2001), which is a generalization of the concept of relative degree for finite dimensional systems (Khalil 2002; Lévine 2009). Moreover, although other forms of weighting function can also be considered, the one given in (6) is a convenient choice that facilitates control design and well-posedness and stability analysis and is meaningful for the application considered in this work as illustrated later in Sect. 5. However, as the boundaries in the computation of the output defined in (6) are time-varying, it is not suitable for control design and analysis. For this reason, we transform the output into a form with fixed boundaries in the normalized coordinates. Specifically, letting a = a(Δx)2 and b = Δx(ax + b), the output can be expressed in the normalized coordinates as y(t) =

P η

1

az + b w (z, t) dz.

(7)

0

˙ = ˙ a u + Δx b. Moreover, we have b˙ = Δx(a x˙ + b) Let yd (t) be the desired aggregate power, which is usually a sufficiently smooth function, and denote by e(t) = y(t) − yd (t) the tracking error. We can derive from (5) and (7) the tracking error dynamics

56

J. Zheng et al.

de P = dt η =

P η +

1

˙ w dz − y˙ (t) az + b w t + b d

0 1 0

P η

P w az + b β z − ( α1 − 2 u) w z dz + η

1

a u + Δx b˙ w dz

0

1

az + b δ dz − y˙d (t).

0

Performing integration by parts on the first term on the right-hand side of the above equation while applying the boundary conditions given in (5), we get P 1 ˙ w dz w a β z − ( α1 − u) w dz + Δx b η 0 0 1 P a + b + az + b δ dz − y˙d (t). σ (t)− b σ (t) + η 0

P de =− dt η

Note that

1 0

1

w z dz = w (1, t) − w (0, t) and let ( β w(1, t) − w (0, t)) −

1

η φ(t) α1 + Δx b˙ w dz + aP

0

u(t) = −

1

0

P a + b Γ (t) = σ (t) − b σ (t) + η

,

az + b δ dz ,

(8)

w dz 1

(9)

0

where φ(t) is an auxiliary control. Then the tracking error dynamics become de (t) = φ(t) − y˙d (t) + Γ (t), e(0) = e0 , dt

(10)

where e0 is the initial regulation error. In the original coordinates, the control is expressed as

x

β(w(x, t) − w(x, t)) − u(t) = −

x

x

,

(11)

w dx

Δx P Γ (t) = η

η φ(t) α1 + b˙ w dx + aP

x

(ax + b) σ (t) − ax + b σ (t) +

x

x

(ax + b) δ dx .

(12)

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

57

Letting φ(t) = y˙d (t) − k0 e(t),

(13)

where k0 > 0 is the controller gain, we obtain de (t) = −k0 e(t) + Γ (t), e(0) = e0 . dt It can be seen from (11) that as the deadband is not empty, the characteristic x index of the input-output dynamics of the system is 1 if x w dx, representing the accumulation of the loads in ON-state, is not null. This condition can be easily fulfilled for a large enough TCL population for which it is reasonable that at least one TCL is in ON-state all the time in the whole population. In addition, due to the fact that at any moment only a small portion of TCLs may change their state between ON and OFF and the boundedness of b can be guaranteed by an appropriate design, it is reasonable to assume that Γ (t) is uniformly bounded, i.e., |Γ (t)| ≤ Γ∞ for all t > 0 with Γ∞ a positive constant. Thus, the control given in (13) guarantees that the trajectory of the system (10) is globally uniformly bounded, as stated later in Theorem 4. Moreover, the amplitude of regulation error e(t) may be reduced by increasing the control gain k0 . In addition, it should be noted that the proposed control scheme is robust with respect to δ(x, t), σ (t), and σ (t), which are treated as disturbances. Finally, as there is no need to compute Γ (t), the TCLs do not need to signal the instantaneous switching operations. This will allow greatly simplifying the implementation and making the control scheme insensitive to timing constraints.

4

Well-Posedness and Stability of the Closed-Loop System

This section is dedicated to addressing two essential properties of the closed-loop system, namely, the well-posedness of the setting in terms of the existence and the uniqueness of the solution and the closed-loop stability.

4.1

Well-Posedness Assessment

It is well known that in the study of classical PDEs (e.g., Laplace equation, heat equations, and wave equations), usually some very specific types of boundary conditions, e.g., Dirichlet, Neumann, and Robin boundary conditions, are associated with these equations. It is often physically obscure whether a boundary condition is appropriate or not for a given PDE. Therefore, this aspect has to be clarified by a fundamental mathematical insight. According to Hadamard’s principe (Hadamard 1923), an initial-boundary value problem (IBVP) of PDEs is said to be “well-posed” if:

58

J. Zheng et al.

(i) It has a solution on a prescribed domain for all suitable boundary data. (ii) The solution is uniquely determined by such data. (iii) The solution is also continuously determined by such data. In PDE theory, the study of well-posedness is mainly focused on the existence and the uniqueness of the solutions. From an application point of view, the importance of the well-posedness is that it assures the validity of a model in the sense that its mathematical formulation complies with the nature of the realworld system. However, unlike finite-dimensional systems described by ordinary differential equations for which the existence of unique solutions of a wide variety of systems can be confirmed under some regular conditions (see, e.g., Khalil 2002, Ch. 3), there do not exist generic results for the well-posedness of PDEs having a general form. Therefore, well-posedness assessment for PDEs is usually carried out on a case-by-case basis. This motivates our study presented in this section. We establish first the well-posedness of the closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13), or equivalently, (5), (7), (8), (9), (10), and (13). For this purpose, in the present work, we always impose the following basic assumptions: (A1) β, P , η, k0 ∈ R+ , a ∈ R \ {0}; (A2) σ , σ ∈ L1loc (R≥0 ), b ∈ C 2 (R≥0 ), yd ∈ C 2 (R≥0 ), x, x ∈ C 1 (R≥0 ), Δx := x − x is a positive constant; (A3) for any T ∈ R+ and t ∈ [0, T ], δ ∈ L1 (ΩT ), α1 , α0 ∈ C 1 ([x(t), x(t)]), w 0 , v 0 ∈ H 2+θ ([x(t), x(t)]) with a constant θ ∈ (0, 1); (A4) for any t ∈ R≥0 , w 0 , v 0 , δ, σ and σ satisfy the following conditions: B1 [w 0 ](x, 0) = σ (0), B1 [w 0 ](x, 0) = σ (0),

(14a)

B0 [v 0 ](x, 0) = −σ (0), B0 [v 0 ](x, 0) = −σ (0), x(0) 0 w (x) dx > 0,

(14b)

x(0)

x(0)

x(0)

w (x) dx + 0

t 0

x(s) x(s)

t

δ(x, s) dx ds + 0

(14c) δ0 σ (s) − σ (s) ds ≥ , 2 (14d)

where δ0 is positive constant. Note that (14a) and (14b) are compatibility conditions for the well-posedness and the conditions given in (14c) and (14d) mean that at any time the TCLs are not all in OFF-state. Moreover, the assumptions on the discontinuity and local integrability of δ, σ , and σ reflect well the nature of the considered problem. However, since δ, σ , and σ are discontinuous, the closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13) (or (5), (7), (8), (9), (10), and (13)) does not admit

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

59

any classical solution that satisfies the equations pointwisely. To address the wellposedness, we consider certain smooth solutions that satisfy the equations in the sense of distribution. It is worth noting that most of the above assumptions are required for assuring the existence of such smooth solutions to the considered problem, which can eventually be relaxed if we consider solutions in a much weaker sense. Definition 1. Let D be an open or closed domain in Ri (i = 1, 2). For two locally integrable functions f and g defined on D, we say that f = g in the sense of distribution, if D f (s)ϕ(s) ds = D g(s)ϕ(s) ds holds for any ϕ ∈ C0∞ (D). Definition 2. (i) We say that (w, v) ∈ C 2,1 (Ω ∞ ) × C 2,1 (Ω ∞ ) is a distributional solution of the closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13), if such (w, v) satisfies (1), (2), and (3) in the sense of distribution. (ii) We say that (w, v) ∈ C 2,1 (Q∞ ) × C 2,1 (Q∞ ) is a distributional solution of the closed-loop system composed of (5), (7), (8), (9), (10), and (13), if such ( w, v) satisfies (5a), (5b), (5c), (5d), (5e), (5f), and (5g) in the sense of distribution. Essentially, the existence of a solution to an IBVP can be established via regularity analysis and a priori estimates of the solution. The result on the existence of a solution to the considered problem is given in the following theorem. Theorem 1 The closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13) admits a distributional solution (w, v) ∈ C 2,1 (Ω ∞ ) × C 2,1 (Ω ∞ ). x−x

By virtue of the variable transformation: z = Δx , Theorem 1 is a direct result of the following proposition, whose proof is provided in Appendix “B: Proof of Proposition 2”. Proposition 2 The closed-loop system composed of (5), (7), (8), (9), (10), and (13) admits a distributional solution ( w , v ) ∈ C 2,1 (Q∞ ) × C 2,1 (Q∞ ). It is worth noting that if δ, σ and σ are continuous, such distributional solutions of the closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13) (or (5), (7), (8), (9), (10), and (13)) become classical solutions that satisfy the equations pointwisely. Regarding the uniqueness of the distributional solution of the closed-loop system, we have the following theorem, whose proof is provided in Appendix “C: Proof of Theorem 3”. Theorem 3 Let (w1 , v1 ), (w2 , v2 ) ∈ C 2,1 (Ω ∞ ) × C 2,1 (Ω ∞ ) be two distributional solutions of the closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13). If for any t ∈ R≥0 , w1 (x(t), ·) − w1 (x(t), ·) = w2 (x(t), ·) − w2 (x(t), ·) in R≥0 , then (w1 , v1 ) = (w2 , v2 ) in Ω ∞ .

60

J. Zheng et al.

4.2

Stability Analysis

We assess first the stability of the error dynamics (10) with the control given in (13). As the regulation error, e(t), is governed by a first-order finite dimensional linear system, it is straightforward to have the following result. Theorem 4 The regulation error e(t) is determined by e(t) = e(0) e−k0 t +

t

Γ (s) e−k0 (t−s) ds.

0

Furthermore, if |Γ (t)| ≤ Γ∞ with a positive constant Γ∞ for all t > 0, then |e(t)| ≤ |e(0)| e−k0 t +

Γ∞ 1 − e−k0 t , ∀t > 0. k0

To assure the closed-loop stability, we need to prove that the internal dynamics composed of (1), (6), and (11) are stable provided Theorem 4 holds true. Toward this aim, we establish some essential properties of the closed-loop solution stated in the following two propositions. Proposition 5 Let (w, v) ∈ C 2,1 (Ω ∞ ) × C 2,1 (Ω ∞ ) be a distributional solution of the closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13). For any t ∈ R≥0 , it holds that:

x(0) t x(s) t w(x, t) dx= w 0 (x) dx+ δ(x, s) dx ds+ σ (s)−σ (s) ds; x(t) x(0) 0 x(s) 0 x(0) t x(s) t x(t) v(x, t) dx= v 0 (x) dx− δ(x, s) dx ds− σ (s)−σ (s) ds. (ii) x(t)

(i)

x(t)

0

x(0)

0

x(s)

Therefore, (4) holds true. Proof. Let w satisfy (5). We have then d dt

1

1

w (z, t) dz =

0

w t (z, t) dz

0 1

= 0

w β z − ( α1 − 2u) w z dz +

= σ (t) − σ (t) +

1

1

δ (z, t) dz

0

δ (z, t) dz.

0

It follows that 0

1

1

w (z, t) dz = 0

w 0 (z) dz +

t 0

1 0

δ (z, s) dz ds +

0

t

σ (s) − σ (s) ds.

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

61

x−x

Using the transformation z = Δx , we obtain the result claimed in (i). The result claimed in (ii) can be obtained in a similar way. Proposition 6 Let (w, v) ∈ C 2,1 (Ω ∞ ) × C 2,1 (Ω ∞ ) be a distributional solution of the closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13). For any t ∈ R≥0 , it holds that: (i)

w(·, t) L1 (x(t),x(t)) ≤ w 0 L1 (x(0),x(0)) +

t

x(s)

δ(x, s) sgn(w(x, s)) dx ds 0

x(s)

t

+

σ (s) sgn(w(x(s), s)) ds 0

t

−

σ (s) sgn(w(x(s), s)) ds; 0

(ii)

v(·, t) L1 (x(t),x(t)) ≤ v 0 L1 (x(0),x(0)) −

t

x(s)

δ(x, s) sgn(v(x, s)) dx ds 0

x(s)

t

−

σ (s) sgn(v(x(s), s)) ds 0

+

t

σ (s) sgn(v(x(s), s)) ds, 0

where sgn(f ) := 1 if f > 0; sgn(f ) := −1 if f < 0; sgn(f ) := 0 if f = 0. x−x

Proof. By virtue of the variable transformation z = Δx , it suffices to let n → ∞ in (26) (see Appendix “B: Proof of Proposition 2”), which leads to the desired result. It should be mentioned that Proposition 6 can still not guarantee the stability of the internal dynamics. In fact by checking Claim (i), it is obvious that if t x w has the same sign as δ over ΩT , then 0 x δ(x, s) sgn(w(x, s)) dx ds = t x 0 x |δ(x, s)| dx ds. However, it is nature to do not impose any assumption on the t x global L1 -integrability of δ. That is to say, 0 x |δ(x, s)| dx ds may tend to ∞ as t t t → ∞. Similarly, 0 |σ (s)| ds and 0 |σ (s)| ds may also tend to ∞ as t → ∞. Consequently, the boundedness of w(·, t) L1 (x(t),x(t)) cannot yet be guaranteed if t → ∞. The same argument also holds true for Claim (ii). To establish the closed-loop stability of the internal dynamics, we note first that due to the fact that the considered problem is conservative in terms of the total number of the TCLs in the population, there must be positive constants M, M such that t) dx dt < M for Ω δ(x, any Lebesgue measurable set Ω ⊂ Ω∞ , and I σ (s) ds < M , I σ (s) ds < M for any Lebesgue measurable set I ⊂ [0, +∞). We have then the following result. Theorem 7 Assume that there exist positive constants M, M such that Ω δ(x, t) dx dt| < M for any Lebesgue measurable set Ω ⊂ Ω∞ , and I σ (s) ds < M ,

62

J. Zheng et al.

σ (s) ds < M for any Lebesgue measurable set I ⊂ [0, +∞). Let (w, v) ∈ I C 2,1 (Ω ∞ ) × C 2,1 (Ω ∞ ) be a distributional solution of the closed-loop system composed of (1), (2), (3), (6), (10), (11), (12), and (13). For any t ∈ R≥0 , it holds that: (i) w(·, t) L1 (x(t),x(t)) ≤ w 0 L1 (x(0),x(0)) + 2M + 2M < +∞; (ii) v(·, t) L1 (x(t),x(t)) ≤ v 0 L1 (x(0),x(0)) + 2M + 2M < +∞. Proof. For any T ∈ R+ , we consider the closed-loop system composed of (5), (7), (8), (9), (10), and (13) over the domain QT . For t = 0, we have immediately w(·, 0) L1 (x(0),x(0)) = w 0 L1 (x(0),x(0)) ≤ w 0 L1 (x(0),x(0)) + 2M + 2M < +∞. (z, s) > 0}, Q− = {(z, s) ∈ For any t ∈ (0, T ], let Q+ = {(z, s) ∈ Qt ; w + Qt ; w (z, s) < 0}, Ω = {(x, s) ∈ Ω t ; w(x, s) > 0} and Ω − = {(x, s) ∈ Ω t ; w(x, s) < 0}, J + = {s ∈ [0, t]; w (1, s) > 0}, J − = {s ∈ [0, t]; w (1, s) < 0}, I + = {s ∈ [0, t]; w(x(s), s) > 0} and I − = {s ∈ [0, t]; w(x(s), s) < 0}. By x−x Proposition 6 and the variable transformation z = Δx , we have 1 w(·, t) L1 (x(t),x(t)) Δx = w (·, t) L1 (0,1) t t ≤ w0 L1 (0,1) + σ (s) sgn( w(1, s)) ds− σ (s) sgn( w(0, s)) ds +

t 1 0

0

0

δ (z, s) sgn( w) dz ds

0

= w0 L1 (0,1) +

J+

σ (s) ds+

J−

σ (s) ds +

Q+

δ (z, s) dz ds −

Q−

δ (z, s) dz ds

≤ w0 L1 (0,1) + δ (z, s) dz ds + δ (z, s) dz ds σ (s) ds + σ (s) ds + J+

J−

1 1 1 0 = w L1 (x(0),x(0)) + σ (s) ds + Δx Δx I + Δx 1 + δ(x, s) dx ds Δx Ω − ≤

Q+

Q−

1 σ (s) ds + δ(x, s) dx ds − + Δx I Ω

1 2M 2M w0 L1 (x(0),x(0)) + + , Δx Δx Δx

which, along with the variable transformation z = can be proved in the same way.

x−x Δx ,

yields Claim (i). Claim (ii)

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

5

63

Simulation Study

We illustrate the developed control scheme through a benchmark problem corresponding to a population of air conditioners (see, e.g., Callaway 2009; Ghanavati and Chakravarthy 2018; Moura et al. 2013, 2014; Zheng et al. 2020). The simulated system contains 10,000 TCLs. The parameters of the system and the controllers are listed in Table 1. As proposed by Callaway (2009), Ghanavati and Chakravarthy (2018), and Moura et al. (2013, 2014), the heterogeneity of this population is generated by the variation of the thermal capacitance of TCLs, which is supposed to follow a log-normal distribution in this study resulting in a difference in term of the thermal constant (RC) of the TCLs up to 3.5 times. Moreover, we consider in this work a case where the ambient temperature varies over time. As the measurement of the ambient temperature is available for control implementation, the actual form of the ambient temperature can be arbitrary. We use then the one shown in Fig. 2 in the simulation study. The diffusivity of the Fokker-Planck equations is taken from Zheng et al. (2020), which is obtained by applying the algorithm developed by Moura et al. (2014). The simulation is performed for a period of 24 hours. The results related to power consumptions are all presented in quantities normalized by the maximal total power consumption of the population. The deadband width is set to Δx = 0.5 ◦ C. Initially, the temperatures of the TCLs are uniformly distributed over [20−Δx/2, 20+Δx/2](◦ C), and their operational state is randomly generated with approximately 50% in ON-state and 50% in OFF-state. In the simulation, the forced switchings are randomly generated independently at the level of each TCL. Moreover, we note that by rewriting the weighting function in the output defined in (6) as ax + b = a(x − xp ), then xp can be interpreted as the set-point. For this reason, in aggregate power tracking control we can fix yd while varying xp , which can be determined by the desired total power consumption of the population (see, e.g., Radaideh et al. 2019). Furthermore, to avoid power consumption oscillations due to fast step set-point variations (Callaway 2009; Ihara and Schweppe 1981; Laparra et al. 2020; Radaideh et al. 2019; Totu et al. 2017), we consider a tracking control scheme with the reference trajectory, xp , a smooth Table 1 System parameters

Parameters R C P xie η β Δx a k0

Description [Unit] Thermal resistance [◦ C/kW] Thermal capacitance [kWh/◦ C] Thermal power [kW] Ambient temperature [◦ C] Load efficiency Diffusivity [◦ C2 /h] Temperature deadband width [◦ C] Weighting function coefficient Control gain

Value 2 ∼N (10, 3) 14 [28, 32] 2.5 0.1 0.5 −1 7.5

64

J. Zheng et al.

Fig. 2 Ambient temperature

Fig. 3 Step set-point variations and smooth reference trajectory

function connecting the initial point at time ti to a desired point at time tf . We choose then a polynomial of the following form taken from Lévine (2009): 4 5 xp (t) = xp (ti ) + xp (tf ) − xp (ti ) τ (t) al τ l (t), t ∈ [ti , tf ],

(15)

l=0

where τ (t) = (t −ti )/(tf −ti ). For a set-point control, the coefficients in (15) can be determined by imposing the initial and final conditions: x˙p (ti ) = x˙p (tf ) = x¨p (ti ) = ... ... x¨p (tf ) = x p (ti ) = x p (tf ) = 0, which yields a0 = 126, a1 = −420, a2 = 540, a3 = −315, and a4 = 70. The reference temperature trajectory used in the simulation is shown in Fig. 3. The control signal is depicted in Fig. 4a. Figure 4b shows the desired temperature trajectory xp and the one generated from the PDE control signal as

t

xref (t) = xref (0) + 0

x˙ref (τ ) dτ = xref (0) +

t

u(τ ) dτ. 0

Note that in order to avoid the chattering induced by fast variations of u(t), the reference sent to the TCLs is the average of the values over the last ten iterations.

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

65

Fig. 4 Dynamics of the TCL population: (a) PDE control signal; (b) desired and generated references; (c) temperature evolutions; (d) aggregate power consumptions

66

J. Zheng et al.

Figure 4c illustrates the temperature evolution of 200 randomly selected TCLs. It can be seen that all the TCLs follow well the reference while respecting the temperature deadband. The variation of the aggregate power consumption of the whole population is shown in Fig. 4d. The results show that the developed control scheme tracks well the desired reference with quite import variations. It is worth noting that since the feedback linearization-based control can cope with parameter variations in a straightforward manner, there is no any concern about the simultaneous variation of set-point and ambient temperature in the considered setting, which is indeed hard to handle with most of the methods reported in the existing literature. Moreover, the purpose of the simulation study presented in this work is to illustrate the essential behaviours of the developed control algorithm, and the emphasis is not put on the performance, although it can be improved by a finer tuning.

6

Concluding Remarks

We have developed in the present work an input-output linearization – governed by a pair of coupled Fokker-Planck equations. Since the thermostat-based deadband control is used in the operation of individual TCLs, the considered problem involves moving boundaries varying with the reference signals. It should be noted that a more generic setting is the case where the upper and lower bounds of the thermostat may vary with different rates. Since the theoretical development, in particular the well-posedness assessment, of this type of systems will involve a very heavy mathematical analysis, it is beyond the scope of this work and will be addressed in a separate work. Other subjects, such as the control of TCL populations with the dynamics of the TCLs described by second-order dynamical models in the framework of partial differential equations, will also be considered in our future work.

A: Notations on Function Spaces Let D be an open or closed domain in Ri (i = 1, 2). C j (D)(j = 1, 2, . . .) consists of all continuous functions u having continuous derivatives over D up to order j inclusively. C ∞ (D) consists of all functions u belonging to C j (D) for all j = 1, 2, . . .. C0∞ (D) consists of all C ∞ -functions defined on D and having a compact support in D. For T ∈ (0, +∞] and a, b ∈ R with a < b, let Q = (a, b) × (0, T ). C 2,1 (Q) consists of all continuous functions u(x, t) having continuous derivatives ux , uxx , ut over Q. For a nonnegative integer m, we also use Dxm u to denote the mth-order derivative of a function u w.r.t. its argument x. Let Kρ be an arbitrary open interval in (−∞, +∞) of length ρ > 0 and Ωρ := Kρ ∩ (a, b). Let l be a nonnegative number. [l] is the largest integer less than or equal to l. H l ([a, b]) is the Banach space whose elements are continuous functions u(x) in [a, b] having in

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

67 (l)

[a, b] continuous derivatives up to order [l] inclusively and a finite norm |u|(a,b) := (j ) (l) u(a,b) + [l] j =0 u(a,b) , where (0)

(j )

(0)

u(a,b) :=|u|(a,b) := sup |u|, u(a,b) := (a,b)

j

(0)

(l)

|Dx u|(a,b) , u(a,b) :=

(j )

(l−[l]) Dx[l] u(a,b) . ([l])

H l,l/2 (Q) denotes the Banach space of functions u(x, t) that are continuous over Q, together with all derivatives of the form Dtr Dxs for 2r + s < l, and have a finite [l] (j ) (l) norm |u|(l) Q := uQ + j =0 uQ , where (0)

(j )

(0)

uQ :=|u|Q := sup|u|, uQ := Q

(l) ux,Q

:=

(τ )

(0)

(l)

(l/2)

(l)

|Dtr Dxs u|Q , uQ := ux,Q +ut,Q ,

(2r+s=j )

(l/2) (l−[l]) Dtr Dxs ux,Q , ut,Q

(2r+s=[l])

ut,Q :=

:=

l−2r−s

2 Dtr Dxs ut,Q

,

0 0 depending only on β 1 0 0 ˙ b ∞ , yd ∞ , y˙d ∞ , x ∞ , α1 ∞ , α1z ∞ , w ∞ , and 0 w dz such that |Gw , p) ∈ [0, 1] × [0, T ] × R × R, 0 (t)| ≤ g0 for all t ∈ [0, T ]. Then for any (z, t, w we infer from Young’s inequality that

δ2 α1 ∞ + g0 α1 ∞ + g0 2 w 2 + p + 0 − w b1 (z, t, w , p) ≤ α1z ∞ + 2 2 2 2 + C1,2 , :=C1,0 p2 + C1,1 w − w ψ1 (z, t, w , p) ≤C1,3 w 2 + C1,4 , where C1,3 , C1,4 are positive constants depending only on g0 , α1 ∞ , σ 0, σ 0. By the compatibility condition (14a) and the definition of w 0 , it follows that w w β z0 (1) + ψ1 (1, 0, w 0 (1)) = β z0 (0) − ψ1 (0, 0, w 0 (0)) = 0.

(20)

Then all the structural conditions specified in Ladyzenskaja et al. (1968, Theorem 7.4, Chap. V) are fulfilled. Therefore, (18) has a unique (classical) solution w 1 ∈ 2+θ,1+ θ2 H (QT ). Moreover, by Ladyzenskaja et al. (1968, Theorem 7.3, 7.2, Chap. V), we have w1 | ≤λ1,1 eλ1 T max{ C1,2 , C1,4 , w 0 (1), w 0 (0)} := m1 , max | w1z | ≤ M1 , max | QT

QT

, C1,0 , C1,1 , and C1,3 , M1 depends only on β , where λ1,1 , λ1 depend only on β (2) 0 α1 ∞ , α1z ∞ , and | w |(0,1) . Furthermore, by Ladyzenskaja et al. (1968, m1 , g0 , (2+θ)

Theorem 5.4, Chap. V), we have | w1 |QT (2) 0 , δ0 , and | w | . m1 , θ , β (0,1)

≤ M 1 , where M 1 depends only on M1 ,

70

J. Zheng et al.

Note that w w β 1z (1, 0) + ψ1 (1, 0, w 1 (1, 0)) = β 1z (0, 0) − ψ1 (0, 0, w 1 (0, 0)) = 0,

(21a)

w 1z (1, 0) = w z0 (1), w 1z (0, 0) = w z0 (0), σ 1 (0) = σ (0), σ 1 (0) = σ (0).

(21b)

By the definition of ψ1 , (20) and (21), we have Gw 1 (0) = Gw 0 (0), which implies that w β z0 (1) − ( α1 (1) − Gw w 0 (1) + σ 1 (0) = 0, 1 (0)) w β z0 (0) − ( α1 (0) − Gw w 0 (0) − σ 1 (0) = 0. 1 (0)) Let k > 1 be an integer. Assuming that for n = k, (18) has a unique (classical) θ solution w k ∈ H 2+θ,1+ 2 (QT ) satisfying (19) and the following equalities: w β z0 (1) − ( α1 (1) − Gw w 0 (1) + σ k (0) = 0, k (0))

(22a)

w β z0 (0) − ( α1 (0) − Gw w 0 (0) − σ k (0) = 0. k (0))

(22b)

We need to prove the existence of a (classical) solution in the case that n = k + 1. Noting that by (16), for all t ∈ [0, T ], it follows that 0

1

1

w k (z, t) dz = 0

w 0 (z) dz +

t 0

1

δk (z, s) dz ds +

0

t 0

σ k (s) − σ k (s) ds. (23)

By the definition of Gwk (t), (14d), (19) and the choice of δn , σ n, σ n , we have |Gw (t)| ≤ g for all t ∈ [0, T ], where g is a positive constant depending only k k k ˙ α1 ∞ , α1z ∞ , on β , P , η, k0 , a, Δx, δ0 , b ∞ , b ∞ , yd ∞ , y˙d ∞ , x ∞ , wk ∞ . Then for any (z, t, w , p) ∈ [0, 1] × [0, T ] × R × R, we obtain w0 ∞ , and , p) ≤Ck+1,0 p2 + Ck+1,1 w 2 + Ck+1,2 , − w bk+1 (z, t, w

(24a)

− w ψk+1 (z, t, w , p) ≤Ck+1,3 w 2 + Ck+1,4 ,

(24b)

where Ck+1,0 , Ck+1,1 , Ck+1,2 , Ck+1,3 , Ck+1,4 are positive constants depending only on gk , σ 0, σ 0, δ0 . α1 ∞ , α1z ∞ , σ k+1 (0) = σ k (0) = σ (0), σ k+1 (0) = σ k (0) = σ (0) and by Noting the fact that (22), we find that the following compatibility conditions hold true: w β z0 (1) − ( α1 (1) − Gw w 0 (1) + σ k+1 (0) = 0, k (0)) w β z0 (0) − ( α1 (0) − Gw w 0 (0) − σ k+1 (0) = 0. k (0))

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

71

Then, by Ladyzenskaja et al. (1968, Theorem 7.4, 7.3, 7.2, Theorem 5.4, Chap. θ V), we conclude that (18) has a unique solution w k+1 ∈ H 2+θ,1+ 2 (QT ) when θ n = k + 1. Therefore, (16) has a unique (classical) solution w n ∈ H 2+θ,1+ 2 (QT ) for every n ≥ 1. Moreover, (19) holds true. Since for any n there exists w n satisfying (16), Gw n (t) is fixed for any fixed n. θ Thus, the existence of a unique (classical) solution vn ∈ H 2+θ,1+ 2 (QT ) to (17) is guaranteed by Ladyzenskaja et al. (1968, Theorem 7.4, Chap. V). Moreover, vn satisfies: (2+θ)

| v |QT

2n , ≤C

(25)

2n is a positive constant which may depend on n. where C θ vn in H 2+θ,1+ 2 (QT ), i.e., we Step 2: We establish uniform estimates of w n and 1n and C 2n are independent of n. First, for any n and any t ∈ [0, T ], we prove that C prove the following L1 -estimates:

t

wn (·, t) 1 ≤ w 0 1 + +

1

t

v 1 − vn (·, t) 1 ≤

0

σ n (s) sgn( vn (1, s)) ds+

0 1

σ n (s) sgn( wn (0, s)) ds

δn (z, s) sgn( wn (z, s)) dz ds,

t

t

0

0

0

−

0

t 0

σ n (s) sgn( wn (1, s)) ds −

(26a) t

0

σ n (s) sgn( vn (0, s)) ds

δn (z, s) sgn( vn (z, s)) dz ds.

(26b)

0

Indeed, for any ε > 0, let

ρε (r) :=

⎧ ⎪ ⎨ |r|, |r| ≥ ε , 3ε r4 3r 2 ⎪ ⎩− + , |r| < ε + 4ε 8 8ε3

which is C 2 -continuous in r and satisfies ρε (r) ≥ |r|, |ρε (r)| ≤ 1 and ρε

(r) ≥ 0. Multiplying (16) by ρε ( wn ), and integrating by parts, we have d dt

1 0

2

ρε ( wn ) dz = − β wnz ρε ( wn ) + + 0

0

1

1

wn ) wn w nz dz α1 − 2Gw n−1 ρε (

δn ρε ( wn ) dz + σ n (t)ρε ( wn (1, t)) − σ n (t)ρε ( wn (0, t)). (27)

72

J. Zheng et al.

Noting that wn

ρε

( wn ) 2

1

=

w n2 ρε

( wn )χ{| wn |>ε} (z) dz

0 1

= 0

1

+ 0

w n2 ρε

( wn )χ{| wn |≤ε} (z) dz

w n2 ρε

( wn )χ{| wn |≤ε} (z) dz

1

=

0

w n2

3(ε2 − w n2 ) χ{| wn |≤ε} (z) dz 2ε3

3 ≤ ε, 2 it follows that

1 0

α1 − 2Gw wn ) wn w z dz n−1 ρε (

1 ≤ α1 − 2Gw n ρε

( wn ) 2 + β wz ρε

( wn ) 2 n−1 w 4β 1 2 α1 − 2Gw ≤ wn ρε

( wn ) 2 + β wz ρε

( wn ) 2 n−1 ∞ 4β ≤Cn ε + β wnz ρε

( wn ) 2 ,

(28)

. where Cn is a positive constant depending only on α1 − 2Gw n−1 ∞ and β By (27) and (28), we have d dt

1

1

ρε ( wn ) dz ≤Cn ε +

0

0

δn ρε ( wn ) dz + σ n (t)ρε ( wn (1, t)) − σ n (t)ρε ( wn (0, t)),

which implies that

1

1

ρε ( wn (z, t)) dz ≤

0

0

ρε ( w 0 (z)) dz + Cn εt +

+ 0

t 0

t

σ n (s)ρε ( wn (1, s)) ds −

0 t

0

1

δn ρε ( wn ) dz dt

σ n (s)ρε ( wn (0, s)) ds.

Letting ε → 0, we obtain

1 0

w n (z, t) dz ≤ 0

1

w 0 (z) dz +

+ 0

t 0

t

1

δn sgn( wn ) dz dt

0

σ n (s) sgn( wn (1, s)) ds −

t 0

σ n (s) sgn( wn (0, s)) ds,

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

73

x−x

which, along with the variable transformation z = Δx , gives (26a). Equation (26b) can be obtained in the same way. Now by (26a) and (26b), for any n ≥ 1 and any t ∈ [0, T ], it follows that

t

wn (·, t) 1 ≤ w 1 + 0

| σ n (s)| ds +

0

t 0

| σ n (s)| ds +

t 0

1

| δn (z, s)| dz ds

0

σ 0T + σ 0T + δ0 T < +∞, ≤ w 0 1 + vn (·, t) 1 ≤ v 0 1 + σ 0T + σ 0T + δ0 T < +∞. By the continuity of w n and the uniform boundedness of wn (·, t) 1 , we deduce that w n is uniformly bounded in n on QT . By (23), (14d), and the choice of δn , σ n, σ n , we have that Gw n (t) is uniformly bounded in n over QT . Furthermore, Cn,0 , Cn,1 , Cn,2 , Cn,3 , Cn,4 in the structural conditions (see (24)) are independent 2n 1n in (19) is actually independent of n. Analogously, C of n. We conclude that C in (25) is independent of n. Step 3: We complete the proof by taking the limit of ( wn , vn ). Indeed, we have θ θ in Step 2 that ( wn , vn ) is uniformly bounded in H 2+θ,1+ 2 (QT )×H 2+θ,1+ 2 (QT ). θ Due to the fact that H 2+θ,1+ 2 (QT ) →→ C 2,1 (QT ), up to a subsequence, there exists ( w , v ) ∈ C 2,1 (QT ) × C 2,1 (QT ) satisfying ( wn , vn ) → ( w , v ) in C 2,1 (QT ) × 2,1 1 δn → δ in L (QT ) and σn → σ, σn → σ in C (QT ) as n → ∞. Noting that L1 (0, T ), we conclude that ( w , v ) satisfies (5a), (5b), (5c), (5d), (5e), (5f), and (5g) in the sense of distribution.

C: Proof of Theorem 3 Indeed, let (w1 , v1 ), (w2 , v2 ) ∈ C 2,1 (Ω T ) × C 2,1 (Ω T ) be two distributional x−x w1 , v1 ), solutions of (1), (2), and (3). By the variable transformation z = Δx , ( 2,1 2,1 ( w2 , v2 ) ∈ C (QT ) × C (QT ) satisfy (5a), (5b), (5c), (5d), (5e), (5f), and (5g) in the sense of distribution. Furthermore, due to the regularity of w 1 and w 2 , W := w 1 − w 2 satisfies the following equations pointwisely: Wz − Wt = β α1 W − 2W Gw w1 (Gw 2 (t) − 2 1 (t) − Gw 2 (t)) z , (z, t) ∈ QT , Wz (1, t) − β α1 (1)W (1, t) − 2W (1, t)Gw w1 (1, t)(Gw 2 (t) − 2 1 (t) − Gw 2 (t)) = 0, t ∈ [0, T ], Wz (0, t) − β α1 (0)W (0, t) − 2W (0, t)Gw w1 (0, t)(Gw 2 (t) − 2 1 (t) − Gw 2 (t)) = 0, t ∈ [0, T ], W (z, 0) = 0, z ∈ (0, 1),

74

J. Zheng et al.

where Gw i (t)(i = 1, 2) is defined as in Appendix “B: Proof of Proposition 2”. Then we have 1 d W 2 = − 2 dt

1 0 1

+ 0

Wz − Wz β α1 W − 2W Gw 2 (t) dz 2Wz w 1 (Gw 1 (t) − Gw 2 (t)) dz.

(29)

It’s clear that there exists a positive constant C0 such that |α1 + 2Gw 2 (t)| ≤ C0 , ∀(z, t) ∈ QT .

(30)

1 1 Noting (14d), 0 w 1 dz = 0 w 2 dz (see Proposition 5), and w 1 (1, t) − w 1 (0, t) = w 2 (1, t)− w2 (0, t) due to the assumptions on w1 , w2 , there exists a positive constant C1 such that 1

2 Gw 1 (t) − Gw 2 (t) =

2 0

k0 b ˙ k0 z + + α1 + Δx b W dz a ≤ C1 W 1 . 1 w 2 dz 0

(31) By (29), (30), (31), Hölder’s inequality, and Young’s inequality, we have C2 1 d ε ε Wz 2 + Wz 2 + 0 W 2 + Wz 2 W 2 ≤ − β 2 dt 2 2ε 2 1 w1 2 · (C1 W 1 )2 + 2ε 1 − ε) Wz 2 + (C02 + C12 ≤ − (β w1 2 ) W 2 , 2ε ) and taking integration, we where ε > 0 is a positive constant. Choosing ε ∈ (0, β have W (·, t) 2 ≤ W (·, 0) 2 · e

t

1 2 2 w (·,s) 2 ) ds 1 0 ε (C0 +C1

= 0,

v -system is which along with the continuity of W yields W ≡ 0 on QT . Since the v2 over QT . Finally, using the variable transformation linear, it is clear that v1 = x−x z = Δx again, we obtain the desired uniqueness of the solution to the considered system.

A PDE-Based Aggregate Power Tracking Control of Heterogeneous TCL Populations

75

References D. Angeli, P.-A. Kountouriotis, A stochastic approach to “dynamic-demand” refrigerator control. IEEE Trans. Control Syst. Technol. 20(3), 581–592 (2012) M.J. Balas, Active control of flexible dynamic systems. J. Optim. Theory Appl. 25(3), 415–436 (1978) S. Bashash, H.K. Fathy, Modeling and control of aggregate air conditioning loads for robust renewable power management. IEEE Trans. Control Syst. Technol. 21(4), 1318–1327 (2013) D.S. Callaway, Tapping the energy storage potential in electric loads to deliver load following and regulation, with application to wind energy. Energy Conv. Manag. 50(5), 1389–1400 (2009) P.D. Christofides, Nonlinear and Robust Control of PDE Systems: Methods and Applications to Transport-Reaction Processes (Birkhuser, Boston, 2001) P.D. Christofides, P. Daoutidis, Feedback control of hyperbolic PDE systems. AIChE J. 42(11), 3063–3086 (1996) R. Deng, Z. Yang, M.-Y. Chow, J. Chen, A survey on demand response in smart grids: mathematical models and approaches. IEEE Trans. Ind. Informat. 11(3), 570–582 (2015) A. Ghaffari, S. Moura, M. Krsti´c, Modeling, control, and stability analysis of heterogeneous thermostatically controlled load populations using partial differential equations. J. Dyn. Syst. Meas. Control 137, 101009 (2015) M. Ghanavati, A. Chakravarthy, Demand-side energy management by use of a design-thenapproximate controller for aggregated thermostatic loads. IEEE Trans. Control Syst. Technol. 26(4), 1439–1448 (2018) J. Hadamard, Lectures on Cauchy’s Problem in Linear Partial Differential Equations, 3rd edn. (Yale University Press, New Haven, 1923) Reprinted (Dover Publications, New York, 1953) H. Hao, B.M. Sanandaji, K. Poolla, T.L. Vincent, Aggregate flexibility of thermostatically controlled loads. IEEE Trans. Power Syst. 20(1), 189–198 (2015) S. Ihara, F.C. Schweppe, Physically based modelling of cold load pickup. IEEE Trans. Power Appart. Syst. PAS-100(9), 4142–4150 (1981) H.K. Khalil, Nonlinear Systems, 3rd edn. (Prentice-Hall, Englewood Cliffs, 2002) S. Koch, J.L. Mathieu, D.S. Callaway, Modeling and control of aggregated heterogeneous thermostatically controlled loads for ancillary services, in Proceedings of Power Systems Computation Conference, Stockholm (2011), pp. 1–8 M. Krstic, I. Kanellakopoulos, P. Kokotovic, Nonlinear and Adaptive Control Design (Wiley, New York, 1995) O.A. Ladyzenskaja, V.A. Solonnikov, N.N. Uralceva, Linear and Quasi-linear Equations of Parabolic Type (American Mathematical Society, Providence, 1968) G. Laparra, M. Li, G. Zhu, Y. Savaria, Desynchronized model predictive control for large populations of fans in server racks of datacenters. IEEE Trans. Smart Grid 1(1), 411–419 (2020) J. Lévine, Analysis and Control of Nonlinear Systems: A Flatness-based Approach (Springer, Berlin, 2009) M. Liu, Y. Shi, Model predictive control of aggregated heterogeneous second-order thermostatically controlled loads for ancillary services. IEEE Trans. Power Syst. 31(3), 1963–1971 (2016) M. Liu, Y. Shi, X. Liu, Distributed MPC of aggregated heterogeneous thermostatically controlled loads in smart grid. IEEE Trans. Ind. Electron. 63(2), 1120–1129 (2016) N. Mahdavi, J.H. Braslavsky, M.M. Seron, S.R. West, Model predictive control of distributed air-conditioning loads to compensate fluctuations in solar power. IEEE Trans. Smart Grid 8(6), 3055–3065 (2017) A. Maidi, J.-P. Corriou, Distributed control of nonlinear diffusion systems by input-output linearization. Int. J. Robust Nonlinear Control 24(3), 389–405 (2014) R. Malhamé, C.-Y. Chong, Electric load model synthesis by diffusion approximation of a highorder hybrid-state stochastic system. IEEE Trans. Autom. Control 30(9), 854–860 (1985)

76

J. Zheng et al.

S. Moura, V. Ruiz, J. Bendtsen, Modeling heterogeneous populations of thermostatically controlled loads using diffusion-advection PDEs, in ASME Dynamic Systems and Control Conference, Palo Alto (2013), p. V002T23A001 S. Moura, J. Bendtsen, V. Ruiz, Parameter identification of aggregated thermostatically controlled loads for smart grids using PDE techniques. Int. J. Control 87(7), 1373–1386 (2014) P. Palensky, D. Dietrich, Demand side management: demand response, intelligent energy systems, and smart loads. IEEE Trans. Ind. Informat. 7(3), 381–388 (2011) A. Radaideh, U. Vaidya, V. Ajjarapu, Sequential set-point control for heterogeneous thermostatically controlled loads through an extended Markov chain abstraction. IEEE Trans. Smart Grid 10(4), 4095–4106 (2019) S.E.Z. Soudjani, A. Abate, Aggregation and control of populations of thermostatically controlled loads by formal abstractions. IEEE Trans. Control Syst. Technol. 23(3), 975–990 (2015) S.H. Tindemans, V. Trovato, G. Strbac, Decentralized control of thermostatic loads for flexible demand response. IEEE Trans. Control Syst. Technol. 23(5), 1685–1700 (2015) L.C. Totu, R. Wisniewski, J. Leth, Demand response of a TCL population using switching-rate actuation. IEEE Trans. Control Syst. Technol. 25(5), 1537–1551 (2017) W. Zhang, J. Lian, C.-Y. Chang, K. Kalsi, Aggregated modeling and control of air conditioning loads for demand response. IEEE Trans. Power Syst. 28(4), 4655–4664 (2013) L. Zhao, W. Zhang, A unified stochastic hybrid system approach to aggregate modeling of responsive loads. IEEE Trans. Autom. Control 63(12), 4250–4263 (2018) J. Zheng, G. Laparra, G. Zhu, M. Li, Aggregate power control of heterogeneous TCL populations governed by Fokker–Planck equations. IEEE Trans. Control Syst. Technol. 28(5), 1915–1927 (2020)

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid Dongsu Kim, Heejin Cho, and Pedro J. Mago

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 US Electric Power Demand Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 US Climate Zones and New Residential Building Permits . . . . . . . . . . . . . . . . . . . . . 2.3 Multifamily Residential Prototype Building (Baseline) . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Multifamily Net-Zero Energy (NZE) Residential Building . . . . . . . . . . . . . . . . . . . . . 2.5 Calculation of Net-Electricity Demand Differences . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 NZE Balances of Multifamily Models in US Climate Zones . . . . . . . . . . . . . . . . . . . 3.2 Net-Electricity Impact of NZE Residential Buildings on a National Scale . . . . . . . . 3.3 Potential Aggregate Impact on the Actual US Electricity Demand Profile . . . . . . . . . 4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

78 80 80 80 81 83 87 89 89 90 93 97 98

Abstract

This chapter demonstrates the potential aggregate impacts of smartly designed energy-efficient residential building (aka net-zero energy (NZE) buildings) implementations on the US electrical grid based on simulation-based analysis results. The aggregate impact of large-scale NZE implementations on the US D. Kim Department of Architectural Engineering, Hanbat National University, Daejeon, South Korea e-mail: [email protected] H. Cho () Department of Mechanical Engineering, Mississippi State University, Mississippi State, MS, USA e-mail: [email protected] P. J. Mago Department of Mechanical and Aerospace Engineering, West Virginia University, Morgantown, WV, USA e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_22

77

78

D. Kim et al.

electrical grid is evaluated through a simulation-based study of prototype residential building models with distributed photovoltaic (PV) generation systems. A comprehensive whole building energy modeling software program, EnergyPlus, is used to simulate the energy and environmental performance of a smartly designed energy-efficient residential building model (i.e., a net-zero energy multifamily low-rise apartment building). The electricity consumption of those residential building models in 12 different US climate locations is estimated to evaluate potential impacts on the US electric grid. Results indicate that adding distributed PV systems to enable annual multifamily NZE performance can significantly increase changes in imported and exported electricity demand from and to the electrical grid during the daytime. However, using electric energy storage (EES) within NZE homes helps reduce the peak electricity demand during the daytime. The stored electricity in the EES can also be used during the evening time. The peak net-electricity differences on the US electrical grid level could potentially be reduced during the daytime and shifted to the evening. Keywords

Net-Zero energy · Electric energy storage · Distributed photovoltaic · Residential buildings Nomenclature

EES NZE PV STC

1

Electrical energy storage Net-zero energy Photovoltaic Solar thermal collector

Introduction

The building and construction sector is responsible for about 36% of final energy use and about 39% of the carbon dioxide emissions globally in 2018 (IEA 2019), and the residential building sector accounts for about 37% of the electricity currently used in the US electrical grid level (EIA 2017). The concepts of zero energy and emission buildings have attracted increasing attention in recent years in the USA to achieve increased energy efficiency and greenhouse emission reductions. One example of this trend is the California Long-term Energy Efficiency Strategic Plan’s (CEESP) goal to have all new residential buildings be net-zero energy by 2020 (California Public Utilities Commission Energy Division 2015). Net-zero energy (NZE) homes try to reduce annual site-energy consumption through energy efficiency technologies, while matching the annual end-use demand to the on-site power generation (Crawley et al. 2009). Li et al. (2013) provided a good overview of NZE homes, involving two strategies: minimizing the need for energy consumption within buildings through energy-efficient measures (EEMs) and adopting renewable energy and other technologies (RETs). Attia et al. (2017) reviewed the societal and

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

79

technical barriers of NZE building implementations and provided recommendations based on available empirical evidence to further lower those barriers in European countries. Kampelis et al. (2017) proposed a comprehensive approach for evaluating the operational performance and minimizing the performance gap of various NZE buildings, including industrial, residential, and commercial buildings. Faris et al. (2017) also reviewed the latest smart technologies and telecommunication techniques to effectively enable NZE homes. In their conclusion, they stated that the intelligent home energy management techniques would provide a solid base to adopt renewable energy systems and then NZE homes with high energy-efficient design. Neves et al. (2021) proposed the path to net-zero energy using geothermal and photovoltaic systems for residential buildings across 12 US climate locations. Kim et al. (2021) demonstrated future renewable design requirement changes of net-zero carbon office building based on potential future climate scenarios in the USA. Their results showed that the cooling energy requirement would increase significantly under the predicted climate conditions and high-performance cooling systems could be considered in the building design to lower the renewable energy generation requirements for the net-zero carbon goal. Kim et al. ( 2020) provided a net-zero energy building design strategy using a highly efficient heating and cooling system (i.e., variable refrigerant flow system) with PV. Their life cycle cost analysis results showed that considering subsidized financing by the federal government for PV installation costs, net-zero energy buildings can be cost-effective in most US climate locations. Numerous studies have investigated how effective NZE building designs can be achieved by considering advanced energy efficiency technologies and the appropriate on-site power generation in a detailed manner. However, there are several concerns for adopting large-scale implementations of NZE homes in the US electrical grid, typically when over-generation within the electrical grid occurs at increased penetration of renewable energy systems associated with NZE homes (Dirks 2010). Seljom et al. (2017) investigated the impact of large penetrations of NZE buildings on the Scandinavian energy system toward 2050 by assuming that all new buildings and parts of the remodeled buildings were nearly NZE buildings, as well as assuming that 50% of the Scandinavian building stock would be NZE buildings by 2050. Their results indicated that increased penetrations of NZE buildings could affect the cost-optimal investments and operation of the gridtied energy facilities, mainly due to the lower heat demand and the increased PV generation. Salom et al. (2014) evaluated the role of NZE buildings on future energy system facilities by investigating the usefulness and relevance of load matching and electrical grid interaction. They concluded that increased penetration of NZE buildings in the electrical grid required detailed understanding of information about the grid topology, the stochasticity of the building consumption, and the appropriate control systems on both buildings and the electrical grid side. Dirks (2010) also investigated the impact of significant implementations of NZE homes in the western grid in the USA. Although there are a few studies performed on the impact of large NZE building penetrations on the electric grid, many issues have not been fully explored, e.g., the impact on the daily load shifting or load shedding. To fill some of the research gaps, this chapter pointed out that widescale NZE residential

80

D. Kim et al.

building implementations could potentially affect significant changes in the national electricity demand profiles and the hourly values of electric generation. The potential effects of increased NZE residential buildings on the large-scale grid connection in the entire USA have not been sufficiently conducted. Therefore, this chapter aims to evaluate and demonstrate the aggregate impact of NZE implementations on the US electrical grid based on a simulation-based analysis. A residential prototype building model (i.e., a multifamily low-rise apartment building) is used to enable NZE building performance using only PV generation (case 1) and PV generation with electrical energy storage (EES) (case 2). To estimate net-electricity demand profiles of NZE residential buildings on a national scale, surveyed residential building permits in 2017 are used. Considering the current US aggregate electricity demand profiles, a comparative analysis is carried out to evaluate the aggregate impact of significant penetration of NZE residential buildings on the US electrical grid.

2

Methodology

2.1

US Electric Power Demand Curves

Current hourly US electricity demand profiles for three representative days of a year are used to compare electricity demand profiles on a national scale. The Energy Information Administration (EIA) provides hourly electrical system operating data from all balancing authorities, including actual electricity demand, day-ahead electricity demand forecast, system electricity net generation, and system total electricity interchange (EIA 2016b). The electrical system operating data (i.e., EIA930 Data) is estimated based on the physical flow of electricity metered at the grid-tied boundaries between electrical systems in several different sectors (e.g., residential, commercial, industrial, and transportation) (EIA 2016a). Aggregated electricity demand monitored through BAs reflects 13 available regions (i.e., Northwest, California, Southwest, Central, Midwest, Texas, Tennessee, Southeast, Florida, Carolinas, Mid-Atlantic, New York, and New England ISO) in the lower 48 states. The 3 representative days in 2017 are obtained by considering the highest daily electricity consumption during each season, such as winter, transition, and summer. Figure 1 illustrates the hourly electricity demand on the US electrical grid for 3 representative days in 2017 (Aug. 22, Jan. 6, and Apr. 28).

2.2

US Climate Zones and New Residential Building Permits

For the national-scale analysis, multifamily models are placed in 14 different locations by considering all US climate zones except for Hawaii and Alaska. The climate zones are classified from 1 to 8 climates, based on the range of climate parameters, such as heating degree days (HDDs) and cooling degree days (CDDs) (Briggs et al. 2003; ASHRAE 2013). In addition, these climate zones are

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

81

Fig. 1 National aggregate electricity demand curves for each representative day

separated into moist (A), dry (B), and marine (C) regions by considering regional weather conditions (e.g., mean temperature of coldest/warmest month and annual precipitation). Table 1 shows abbreviated weighting factors and the number of new residential building permits in each US climate in 2017. Weighting factors assigned to residential buildings in all US climate zones are calculated based on 2010 residential building permit data by climate zone for each state (Tayplor et al. 2015), along with weather locations used to represent the associated climate zone. The highest weighting factor of multifamily homes is 21.8% for Baltimore, MD, followed by 16.3% and 14.7% for Chicago, IL, and Memphis, TN, respectively. The lowest weighting factor is 0.5% for Albuquerque, NM, followed by 0.8% for Phoenix, AZ. Using the weighting factors of multifamily homes and the 2017 residential building permits surveyed by the US Census Bureau, new residential building permits in each climate zone are estimated. The US Census Bureau estimated 446,732 new permits for multifamily homes in 2017 (Bureau of Census 2018). Table 1 lists calculated building permits for multifamily residential buildings by each climate zone.

2.3

Multifamily Residential Prototype Building (Baseline)

The Department of Energy’s (DOE) flagship building energy modeling software, EnergyPlus version 8.6 (US DOE 2016.), is used to model simulated residential buildings and estimate electric energy consumption throughout a full year, based on US climate weather data. DOE provides a multifamily prototype model complying

82

D. Kim et al.

Table 1 Weighting factors and the number of new residential building permits in US climate locations in 2017 (Tayplor et al. 2015) Climate 1A 2A 2B 3A 3B 3C 4A 4B 4C 5A 5B 6A 6B 7

a

TMY3 Location Miami, FL Houston, TX Phoenix, AZ Memphis, TN El Paso, TX San Francisco, CA Baltimore, MD Albuquerque, NM Salem, OR Chicago, IL Boise, ID Burlington, VT Helena, MT Duluth, MN

Weighting factors 1.8% 13.9% 0.8% 14.7% 9.7% 3.7% 21.8% 0.5% 3.6% 16.3% 4.0% 5.9% 1.5% 1.4%

Residential building permits 8059 62,132 3717 65,773 43,190 16,386 97,278 2192 15,983 72,719 17,714 26,488 6485 6145

b

Fig. 2 (a) Floor plan and (b) 3D view of a simulated multifamily prototype model

with the International Energy Conservation Code (2015 IECC). Figure 2 depicts a floor plan and 3D drawing of the multifamily prototype model used in this chapter. Table 2 shows the residential building characteristics that DOE intends to assume for a multifamily prototype model (Tayplor et al. 2015). The multifamily prototype model is configured as a three-story residential building with 18 units (6 units per floor), each having a conditioned floor area of 111.5 m2 and a window area equal to approximately 10% of the conditioned floor area, equally distributed on all sides of the building. There are several different types of space heating equipment in US residential buildings. According to US Census data surveyed in 2014, the most common types of heating equipment in US homes include 57% of natural gas furnace, 39% of electric heat pump, and 2% of hot water or steam. The type of heating systems by census divisions of the USA also varies depending on weather conditions.

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

83

Table 2 Residential prototype building characteristics (Tayplor et al. 2015; US DOE 2016.) Parameter Conditioned floor area Footprint Ceiling height Perimeter length Gross exterior wall area Window area (relative to conditioned floor area) Internal gains Heating system Cooling system Water heating

Multifamily 2006.7 m2 (21,600 ft2 ) total (18 units) 36.5 m by 19.8 m (120 ft. by 65 ft) total, three-story 2.6 m (8.5 ft) 112.7 m (370 ft) 473.8 m2 (5100 ft2 ) per story, or 1421.4 m2 (2040 ft2 ) total 23% of each gross exterior wall area 0.668 kW (54,668 Btu/day) per unit, or 12.016 kW (984,024 Btu/day) total Natural gas furnace and heat pump Central electric air-conditioning Natural gas heating

For example, although natural gas heating equipment for multifamily buildings is commonly used in most census divisions, an electric heat pump system is the most common in south census divisions, such as East South Central and South Atlantic. Based on the 2014 surveyed data (Tayplor et al. 2015), this study assumes that electrical heat pump systems are used in the 1A, 2A, 3A, and 4A climate models, and natural gas furnace systems are used in other climate models for space heating and domestic hot water systems. For cooling systems, central electric air-conditioning is used for multifamily models in all climate zones because the Census data indicates that 98% of multifamily homes built in 2014 had central air-conditioning installed. In the residential prototype model used in this chapter, direct expansion (DX) cooling coils are used to provide space cooling to conditioned zones as central airconditioning by maintaining the required set-point temperature. Residential prototype building models provided by the DOE (US DOE 2016.) have three different foundation types: basement, crawlspace, or slab-on-grade. Since the 2010 Census data indicates that more than 50% of new residential buildings have slab-on-grade-type foundations (Tayplor et al. 2015), residential models of slabon-grade type are used in this study. Input values for the simulation modeling are directly taken from the original version of multifamily prototype building models without any modifications. Although most of the input values, such as building geometry and internal heat gains, are identical in each of the 14 climate zones, some changes occur depending on system capacities and envelope requirements that vary with climate zones, including outdoor air ventilation, supply air flow rates, and U-value/SHGC.

2.4

Multifamily Net-Zero Energy (NZE) Residential Building

To enable NZE for residential buildings in different climate conditions, this study uses a grid-tied photovoltaic (PV) power generation system by considering net

84

D. Kim et al.

metering available within the electrical grid level. To model a PV system and predict on-site electricity generation, the equivalent one-diode model (US DOE 2015a) in EnergyPlus is used. This model employs equations for an empirical equivalent circuit model to determine the current-voltage characteristics of a single module. In addition, this model uses empirical relationships to predict PV operating performance depending on environmental variables, such as PV cell temperature and shading effects. Since the cell temperature is a critical factor for the electrical generation performance of PV panels, EnergyPlus provides several options of PV back-surface integration modes (e.g., decoupled and integrated back surfaces (US DOE 2015b)) to determine PV operating temperature. In this study, the integrated back surface model is used to calculate the PV cell surface temperature. Figure 3 illustrates fixed roof-mounted and ground-mounted PV systems on the south-facing roof and parking areas. The manufactured PV panel, LG230M1C (Solar Design Tool 2017), is used for input parameters of the equivalent one-diode model in EnergyPlus listed in Table 3.

Fig. 3 Multifamily NZE residential building model with a PV system without and with an EES Table 3 PV performance characteristics (Solar Design Tool 2017.) PV performance parameter Cell type STC power rating STC power per unit of area Peak efficiency Panel height Panel width Panel active area Temperature coefficient of power Nominal operating cell temperature (NOCT) Panel heat loss coefficient

Value/information Monocrystalline silicon 230 W 142.9 W/m2 (13.3 W/ft2 ) 14.29% 1.63 m (64.1 in) 0.98 m (38.8 in) 1.61 m2 (17.3 ft2 ) −0.46%/K 45 ◦ C (113 ◦ F) 30 W/m2 -K

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

85

PV systems require the greatest possible radiation energy to maximize the electric power generation through ideal array orientations. However, it is not practical in general to achieve the maximum power generation from PV because of the architectural consideration (i.e., roof design and orientation) and location. Figure 3 provides an example of a multifamily model that uses a gable roof; the PV tile angle on the south-facing gable roof is 22.4◦ . The number of PV panels required on roof areas is obtained by matching the annual electric energy usage to the on-site yearly electricity generation. In addition to the PV panels installed on roof areas, additional PV panels are used on parking areas until each simulated NZE building shows satisfactory annual NZE balances. By considering the fact that the optimum tile angles of PV panels affect how much of the available irradiance the system can collect, the optimum tile angles of PV panels on parking areas for each climate zone are directly taken from reference (Lave and Kleissl 2011). Table 4 lists the number of PV panels on the roof and parking areas. The total maximum power outputs for the multifamily NZE building models in US climate zones are shown in Table 4 as well. Note that several climate zones (i.e., 1A, 2A, 3A, and 4A) require relatively higher PV capacities to enable NZE performances due to the use of electric heat pumps for the space heating system. One of the main components in PV power generation systems is a DC/AC inverter to convert DC voltage delivered by PV power generators to AC voltage used by consumers (Luque and Hegedus 2012). Inverter loading is typically given as the array to inverter ratio. This is the ratio of DC output power rating by the arrayed PV panels at standard test conditions (STC) to the AC output power rating of the inverter (Advanced Energy 2012). Although the range of the inverter ratio is typically around 0.8–1.3 to avoid clipping PV outputs and maximize operation efficiency, the inverter can be operating at high efficiency over a wide range of power outputs due to a relatively flat efficiency curve. Table 4 The number of PV panels and total maximum power outputs of PV systems for multifamily NZE residential buildings

Climate 1A 2A 2B 3A 3B 3C 4A 4B 4C 5A 5B 6A 6B 7

Multifamily PV panels Roof Parking 240 234 240 270 240 60 240 270 240 12 240 6 240 340 240 8 240 104 240 104 240 48 240 104 240 58 240 82

Total maximum power output (kWp;STC) 109.0 117.3 69.0 117.3 58.0 56.6 133.4 57.0 79.1 79.1 66.2 79.1 68.5 74.1

86

D. Kim et al.

The inverter AC power rating used in this study is set to 126 kW for multifamily NZE residential building models based on the inverter ratio and data from the California Energy Commission (CEC) (California Public Utilities Commission Rule 2017). CEC provides detailed information about manufactured inverter products that are used to model the inverter and PV systems in EnergyPlus. With inverter data from CEC and the assumed AC power rating of 126 kW, the manufactured inverters (Yaskawa Solectria 2014; SMA America n.d.), shown in Table 5, are selected and used for the “Electric Load Center: Inverter: Look Up Table” object (US DOE 2015b) in EnergyPlus. Table 5 also lists detailed information about inverter inputs used in this analysis. The impact of electrical energy storage (EES) within NZE residential buildings on the electrical grid is also considered in this chapter. The EES can be used to store surplus electricity and provide the required backup electricity during low or no PV power generation in NZE buildings. In addition, the EES can reduce curtailed electricity demand curves, caused by significant PV power generation, by shifting the peak point of curtailment in the electrical grid level (Yang 2014). To model the EES within a NZE residential building, the simplified EES model (US DOE 2015a) in EnergyPlus is used. This EES model treats the electrical battery (i.e., Li-ion and other battery technologies) as a black box without significant rate limitations, accounting for energy added and removed, and losses due to charge or discharge inefficiencies, in a simplified manner. Table 6 shows the input values of the simple EES in EnergyPlus. There are losses and limits to charging and discharging electricity into the storage, depending on the capacity rates and the efficiency input values of charging and discharging (US DOE 2015b). Maximum storage capacities and power for charging and discharging of the EES are calculated based on manufactured data (GMED 2017.) and maximum PV power outputs listed in Table 4.

Table 5 Inverter input values in EnergyPlus (California Public Utilities Commission Rule 2017) Inverter performance parameter Manufacturer Description Maximum continuous output power Nominal voltage Weighted efficiency Night tare loss Output power at nominal voltage (% of rated)

10% 20% 30% 50% 75% 100%

Value/information SMA America SC125U 126 kW, 480 Vac commercial grid-tied solar PV inverter 126.0 kW 480 V 97.0% 23.92 W 91.5 93.7 94.0 94.5 93.7 93.8

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

87

Table 6 EES input values in EnergyPlus (GMED 2017.) EES performance parameter Maximum storage capacity Nominal efficiency for charging Nominal efficiency for discharging Maximum power for discharging Maximum power for charging

Value 7.2 kWh * (maximum PV output (kWp)/5.2 kWp) 93% 93% 3.0 kW * (maximum PV output (kWp)/5.2 kWp) 3.84 kW * (maximum PV output (kWp)/5.2 kWp)

The EES model developed in EnergyPlus operates in several different modes of operation schemes, including “track facility electric demand store excess on-site,” “track meter demand store excess on-site,” “track charge discharge schedules,” and “facility demand leveling” (US DOE 2015b). For the storage operation in this analysis, the “track facility electric demand storage excess on-site” option in EnergyPlus is used to control the EES charge and discharge status. This method tries to control the EES operation by tracking all facility electric demand used within an entire residential building, while storing any surplus on-site power generation.

2.5

Calculation of Net-Electricity Demand Differences

Combining electric energy consumption used in residential buildings and on-site generation by PV systems yields purchased and sold electricity from and to the electrical grid. Figure 4 depicts the hourly purchased and sold electricity profiles in terms of case 1 and case 2. Based on these hourly profiles, national net-electricity differences for the multifamily model are calculated as follows:

ENet =

NClimate

EEndUse,i − EPurchased,i + ESold,i wNewPermit,i

(1)

i

where EEndUse is the electricity end uses of multifamily baseline models, EPurchased is the imported electricity in NZE buildings from the electrical grid, ESold is the exported electricity in NZE buildings to the electrical grid, and wNewPermit is the new residential building permits in US climate zones in 2017, listed in Table 1. The potential aggregate impact of NZE implementations on the US electrical grid is calculated as follows: EU.S.Net = EU.S.TotalDemand − ENet

(2)

where EU. S. TotalDemand is the actual US aggregate electricity demand for available US regions in terms of all sectors (e.g., residential, commercial, industrial, and transportation), as shown in Fig. 1. ENet indicates the net-electricity differences for multifamily on a national scale.

88

D. Kim et al.

a

b

Fig. 4 An example of electricity net-difference profile for a typical sunny day: (a) case 1 and (b) case 2

Figure 4 shows variations of hourly electricity profiles of two NZE cases: only PV (case 1) and PV + EES (case 2). As seen in this figure, imported (purchased) electricity from the electrical grid is reduced once the electrical output of the PV systems kicks on. It also shows how the excess generated electricity is fed to the electrical grid during the daytime when the on-site electricity generation is higher than the electricity needed. Results indicate that, using the EES with distributed PV systems, the peak of the purchased and sold electricity profiles can be reduced and flattened, resulting in significant changes in the trends of the net-electricity profiles between case 1 and case 2. Such trends of net-electricity difference profiles are further considered in order to estimate hourly variations of net-electricity demand values on a national scale.

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

3

89

Results and Discussion

This section presents and discusses the simulated results of the multifamily NZE models on a national scale and potential aggregate effects on the US electrical grid. First, NZE performance of each NZE case in US climate zones is explained. Second, net-electricity differences of NZE residential building models on a national scale for a typical seasonal period, calculated by Eq. (1), are discussed. Finally, the potential aggregate impact of NZE implementations on the national electrical grid level is evaluated by comparing with actual US electricity demand profiles on three representative days (i.e., winter, transition, and summer days).

3.1

NZE Balances of Multifamily Models in US Climate Zones

NZE balances are focused on regulated electric energy consumption in residential buildings for this analysis. Regulated electric energy consumption covers the enduse electricity required for space heating, cooling, HVAC fans, domestic hot water, interior and exterior lighting, and interior equipment. Natural gas consumption used for space heating and DHW in several climate zones is not considered in this study. Figure 5 shows the scatter plots of NZE performance in each of the 14 climate locations for multifamily models. Annual regulated electricity consumption used in residential building models is compared against annual on-site PV power generation to assess the NZE performance. Most simulated NZE models show acceptable NZE or nearly NZE balances within ±5% of the net-zero energy balance line, as shown in Fig. 5. NZE residential buildings show a wide range of annual electricity consumption depending on climate locations. For example, the highest and lowest

Fig. 5 Annual electric energy consumption vs. annual on-site electricity generation for each multifamily NZE building placed in 14 climate locations

90

D. Kim et al.

electricity values used include 105.5 kWh/m2 -year and 51.42 kWh/m2 -year for Baltimore, MD, and San Francisco, CA, respectively, for multifamily models. Results also indicate that NZE models with the EES show relatively low points from the net-zero energy balance line compared to those that use only PV with net metering. This is mainly because of the conversion losses through the EES within NZE models. It should also be noted that several climate zones (i.e., 1A, 2A, 3A, and 4A) show relatively high electricity usage due to the use of electric heat pumps for the space heating system.

3.2

Net-Electricity Impact of NZE Residential Buildings on a National Scale

This section compares detailed hourly electricity profiles of the multifamily NZE residential buildings for five seasonal days with and without the EES. Overall, simulated electricity demand profiles vary by time of day and by types of NZE residential buildings, as shown in Fig. 6 through Fig. 8. Electricity end-use profiles of the multifamily NZE buildings are estimated based on regulated electric energy consumption, such as space heating, cooling, HVAC fans, domestic hot water, interior and exterior lighting, and interior equipment. Based on the electricity enduse demand profile, hourly net-electricity differences are calculated by Eq. (1) for each seasonal period. Figure 6 shows simulated electricity demand profiles of the multifamily NZE residential buildings with and without the EES, respectively, during 5 typical winter days (Jan. 4 to Jan. 8). The peak electricity end uses include around 12,000 MW at 7:00 am and at 8:00 pm with and without the EES, respectively, during the selected winter period. Hourly electricity end uses of NZE buildings differ from that of baseline models during nighttime, even more during the daytime. This is mainly due to electric energy losses by the inverters and the EES for PV systems, as well as effects on the interior heating load caused by shading on roofs when PV panels are added for NZE buildings. On-site power generation by PV panels installed in the multifamily NZE buildings varies depending on solar radiation levels. For example, maximum on-site power generation for the multifamily NZE buildings includes around 17,200 MW to 31,404 MW for 5 winter days. Daily on-site power generation shows around 106,520 MWh/day and 179,358 MWh/day for the maximum and minimum daily generation values, respectively, on a national scale. As net metering between on-site PV generation and the electrical grid is assumed in this study for all operation hours, imported and exported electricity values are estimated through the net-metering operation. For multifamily baseline models, all electric energy end uses are directly purchased from the electrical grid facility because there is no on-site electricity source. Unlike the baseline model, purchased and sold electricity values from and to the electrical grid facility are estimated based on the end-use electricity requirements and the on-site electricity generation by PV systems, as mentioned in Fig. 4. Figure 6 indicates that the purchased electricity

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

91

Fig. 6 Simulated multifamily electricity demand profiles on a national scale for typical winter period: (a) case 1 and (b) case 2

within NZE residential buildings decreases when the PV systems generate on-site electricity after around 7:00 am. The surplus electricity is then fed to the electrical grid when the on-site generated electricity of NZE buildings is higher than end-use electricity requirements during the daytime. Although PV systems can eliminate most of the purchased electricity used in the multifamily NZE building, it should be noted that the peak purchased electricity does not exhibit any shrunk or shifted patterns because the peak demand for the winter period mostly occurs in the earlier morning and in the later afternoon. Figure 6 (b) shows simulated electricity demand profiles of the multifamily NZE residential building when the EES is considered. Adding the EES within NZE buildings helps reduce the peak electricity demand in the late afternoon during the winter period. The stored electricity during the daytime can be used later in the day, reducing the electricity demand in the late afternoon and the evening, as well as sold electricity to the electrical grid. Figure 6 also represents hourly netelectricity differences, calculated by Eq. (1), between baseline and NZE models for 5 winter days. As seen in this figure, the net-electricity difference of NZE buildings without the EES (case 1) shows almost identical patterns, due to the fact that NZE

92

D. Kim et al.

Fig. 7 Electricity demand profiles of the multifamily NZE building on a national scale for typical transition period: (a) case 1 and (b) case 2

performance is enabled only using PV generation systems. As expected, the netelectricity difference of the NZE buildings with EES (case 2) exhibits flatter curve patterns during the daytime and evening periods. Figure 7 represents simulated electricity demand profiles of the multifamily NZE residential building in case of the selected transition period (April 26 to April 30) with and without the EES, respectively. Unlike the winter period, the peak end-use electricity of baseline and NZE models occurs predominantly in the evening around 7:00 pm, including about 10,750 MW to 11,785 MW at the peak points during the selected transition period. The transition period shows the reduction of the end-use electricity compared to the winter period, typically in the morning due to outside weather conditions. In contrast, daily on-site electricity generation on a national scale shows relatively high values when compared to the winter period, including around 203,395 MWh/day to 257,845 MWh/day during the selected period. PV systems within the multifamily NZE residential building can generate maximum output, including around 27,839 MW and 31,252 MW when solar radiation is at its highest (around 11 am to 12 pm). As expected, with the on-site electricity generation during the

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

93

daytime, the reduction of the purchased electricity from the electrical grid is shown in Fig. 7, and the excess electricity is then fed back to the electrical grid. Figure 7 also illustrates the hourly patterns of the purchased electricity and the surplus electricity profiles when the EES is added to the multifamily NZE residential building. Since the transition period shows relatively high on-site electricity generation and low end-use electricity during the daytime, the EES can store the excess on-site electricity generation to be used later, resulting in the reduction of the purchased electricity throughout the comparison period. In addition, Fig. 7 shows hourly net-electricity differences of NZE residential buildings with and without the EES. As discussed before, the net-electricity difference of NZE buildings without the EES (case 1) mostly shows the same patterns with the on-site electricity by PV systems because adding PV systems is the only source to affect net-electricity difference in NZE buildings. The peak of net-electricity difference of NZE buildings could potentially be reduced and flattened to about 10,000 MW during the daytime and the evening of the day when compared to NZE buildings without the EES, shown in Fig. 7 (a). Figure 8 shows electricity demand profiles of the multifamily NZE residential building on typical summer period (August 20–August 24) with and without the EES, respectively. As seen in this figure, the hourly patterns of the enduse electricity are relatively in a similar fashion of the transition period, but the amount of daily (summer) end-use electricity is much higher than during the transition period. The daily (summer) end-use electricity of the multifamily NZE building includes around 228,643 MWh/day to 250,592 MWh/day throughout this comparison period. However, the daily on-site electricity generation includes around 214,595 MWh/day to 254,762 MWh/day, covering about 105% of the daily end-use electricity of the multifamily NZE building with and without the EES. In addition, the hourly purchased and sold electricity values represent similar patterns to the transition period for most of the comparative times. Compared to other periods, the amount of electricity sold is relatively higher, whereas the amount of purchased electricity has lower values. Figure 8 illustrates how the EES can reduce the hourly patterns of purchased and sold electricity, when compared to case 1, during the summer period. Figure 8 also represents hourly net-electricity differences of the multifamily NZE residential building with and without the EES. Adding the EES within the NZE building helps reduce and flatten the peak of net-electricity differences, including about 20,000 MW for the daytime and the evening throughout the summer period. These hourly net-electricity differences, provided in Fig. 6 through Fig. 8, are used to estimate the US net-electricity demand profiles and compare them with the actual US electricity demand.

3.3

Potential Aggregate Impact on the Actual US Electricity Demand Profile

Hourly electricity demand profiles could potentially be impacted by increased installations of NZE residential buildings. In this section, potential aggregate effects

94

D. Kim et al.

Fig. 8 Electricity demand profiles of the multifamily NZE building on a national scale for typical summer period: (a) case 1 and (b) case 2

of significant NZE implementations on actual US electricity demand profiles for three seasonal representative days (i.e., winter, transition, and summer days) are evaluated and discussed. The actual US electricity demand reflects sectors that include residential, commercial, industrial, and transportation. Figure 9 through Fig. 11 shows the comparison of hourly US electricity demand profiles for actual surveyed data and simulated data, calculated by Eq. (2) in terms of the multifamily NZE residential buildings. Figure 9 shows what the US net-electricity differences would be on a selected winter day for the multifamily case when compared to actual US electricity demand profiles. The actual peak US electricity demand includes about 575,980 MW at 5:00 pm, and daily US electricity demand includes about 12,808,651 MWh/day. As expected, NZE implementations can potentially influence hourly US electricity demand profiles by lowering imported electricity and exporting the on-site generated electricity from and to the US electrical grid. As seen in this figure, the percentage difference of US net-electricity demand is 4.5% (24,457 MW) at the maximum point (12 pm). In contrast, adding the EES helps reduce the peak differences and shift the peak points to 1:00 pm, including

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

95

Fig. 9 Comparison of simulated net-electricity demand differences from actual US electricity demand for the representative winter day (Jan. 6)

3.4% (18,201 MW). Based on these trends of US net-electricity differences, it should be emphasized that, with the EES in NZE residential buildings, US netelectricity differences from actual US electricity demand can be flattened and widened during the daytime and the evening of the day, respectively. On the representative transition day, April 28, the comparison of US netelectricity demand profiles is shown in Fig. 10. The actual peak US electricity demand is about 492,881 MW at 2:00 pm; daily US electricity demand is about 10,400,551 MWH/day. When a wide-scale penetration of the multifamily NZE residential building is considered, the US electricity demand curves are changed during the daytime, and the peak point is shifted to 5:00 pm. From the compared results of percentage differences between the actual US electricity and the simulated net-electricity demand on a national scale, Fig. 10 shows that percentage differences peak in the middle of the day when solar radiation is at its highest (11:00 am), with 6.5% (31,200 MW). As with the EES, Fig. 10 also shows how the peak of the US electricity difference curves is prevented from growth due to PV power generation installed in the multifamily NZE residential building. In addition, it indicates that the percentage difference is reduced during the daytime and the evening of the day, resulting in 5.5% (26,477 MW) at a maximum point (11:00 am). Figure 11 illustrates the comparison of US net-electricity demand profiles on the representative summer day, August 22. The actual peak US electricity demand at about 3:00 pm appears to be about 670,229 MW on the US electrical grid level. Daily US electricity demand includes about 13,308,722 MWh/day on this selected day. Unlike the transition day, the peak point is not shifted, whereas there are changes in the US electricity demand curve during the morning and the evening.

96

D. Kim et al.

Fig. 10 Comparison of simulated net-electricity demand differences from actual US electricity demand for the representative transition day (Apr. 28)

Fig. 11 Comparison of simulated net-electricity demand differences from actual US electricity demand for the representative summer day (Aug. 22)

Figure 11 shows the peak percentage difference includes 4.8% (31,240 MW) at a maximum point (11:00 am). Fig. 11 also illustrates how added EES can change the hourly electricity demand curve and reduce the curtailed demand curve during this summer day. When the EES is added within the multifamily NZE residential building, the peak percentage difference of US net electricity could potentially be

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

97

shifted to 12:00 pm and reduced to 4.3% (32,530 MW). Note that, in Fig. 9 through Fig. 11, using a combination of the EES in the multifamily NZE residential building can effectively reduce significant changes in the US electricity demand curve during the daytime and offer appropriate ways to use on-site generated electricity during the evening, thereby decreasing the surplus electricity to the US electrical grid.

4

Conclusions

This study evaluated the potential aggregate impact of wide-scale NZE implementations on the US electrical grid in a simulation-based analysis. A multifamily prototype model was used to enable NZE performance using on-site PV generation with and without the EES. Net-metering operation was assumed to be available on the electrical grid so that the surplus electricity generated by on-site PV systems could be fed to the electrical grid. New residential building permits in 2017, calculated by using weighting factors of each US climate, were also used to estimate net-electricity demand differences and compared to actual US electricity demand profiles on a national scale. Results from this study indicated that increased implementations of NZE residential buildings could potentially affect the current US electricity demand profiles. The key findings were the following: • Annual electricity consumption in residential prototype building models varied depending on types of space heating and DHW systems and weather conditions throughout US climate locations. Simulated multifamily NZE building models showed around 52–105 kWh/m2 -year of annual electricity consumption in US climate zones, including some conversion losses through PV generation and the EES systems. To enable annual NZE performance in US climate zones, multifamily homes could need around 56–133 kWp of total maximum PV power outputs. • In a national scale analysis regarding new residential building permits surveyed in 2017, daily electricity end uses included around 234,000 MWh/day to 258,600 MWh/day and around 228,600 MWh/day to 250,600 MWh/day for winter and summer periods, respectively. In addition, the peak electricity end uses of the multifamily NZE buildings mostly occurred in the morning and the evenings. • Adding distributed PV systems to enable the multifamily NZE performance could significantly increase changes in imported and exported electricity from and to the electrical grid during the daytime. However, using the EES within NZE buildings helped reduce the peak electricity demand during the daytime and use stored electricity later in the evening. As a result, the peak net-electricity differences on the US electrical grid level could potentially be reduced during the daytime and shifted to the evening. • Comparing hourly electricity demand profiles for the actual US demand versus the calculated net demand indicated that the percentage differences of US net-electricity demand of the multifamily NZE building included about 4.5% and 4.8% without the EES on the representative winter and summer days,

98

D. Kim et al.

respectively, at a maximum point. However, when the EES was added within NZE buildings, the peak percentage differences could be reduced to about 3.4% and 4.3% on the representative winter and summer days, respectively, at a maximum point. Although enabling NZE residential buildings can be an effective means to achieve greater energy efficiency of new or existing buildings and greenhouse emission reductions, high implementations of NZE buildings across US regions would potentially have significant changes on the electric load on the grid and the use of conventional generators within interconnected electrical grids. These changes may have a detrimental effect on grid operations by significantly reducing building electricity demands in certain time periods when energy storage is not implemented on the grid. Therefore, careful consideration and planning should be undertaken before large-scale implementations of NZE buildings. Simulated results from this study could potentially be used to predict the aggregate effects that a significant number of multifamily NZE buildings would have on the US electricity demand profiles at the electrical grid level. Based on this study, future research should include the impact of large-scale penetrations of NZE buildings by considering different NZE design strategies and analyzing costeffectiveness within NZE residential buildings.

References Advanced Energy, DC Loading of PV Powered Inverters. AE Solar Energy. Bend (2012). http://solarenergy.advanced-energy.com/upload/file/pvp/dcloadingofpvpinverters_55-600100-7 5c.pdf. Accessed 1 Sep 2017 F. AlFaris, A. Juaidi, F. Manzano-Agugliaro, Intelligent homes’ technologies to optimize the energy performance for the net zero energy home. Energ. Buildings 153, 262–274 (2017). https:/ /doi.org/10.1016/j.enbuild.2017.07.089 ASHRAE. (2013). ANSI/ASHRAE/IES Standard 90.1–2013: Energy Standard for Buildings Except Low-Rise Residential Buildings (I-P) (ASHRAE (ed.)). ASHRAE S. Attia, P. Eleftheriou, F. Xeni, R. Morlot, C. Ménézo, V. Kostopoulos, M. Betsi, I. Kalaitzoglou, L. Pagliano, M. Cellura, M. Almeida, M. Ferreira, T. Baracu, V. Badescu, R. Crutescu, J.M. Hidalgo-Betanzos, Overview and future challenges of nearly zero energy buildings (nZEB) design in southern Europe. Energ. Buildings 155, 439–458 (2017). https://doi.org/10.1016/ j.enbuild.2017.09.043 R.L. Briggs, R.G. Lucas, Z.T. Taylor, Climate classification for building energy codes and standards: Part 1 – Development process. ASHRAE Trans., 4610–4611 (2003) Bureau of Census, Building Permits by States and Metro Areas in the United States (2018) California Public Utilities Commission Energy Division, New Residential Zero Net Energy Action Plan 2015–2020, (2015) California Public Utilities Commission Rule. Go Solar California. San Francisco (2017). http:// www.gosolarcalifornia.org/equipment/inverters.php. Accessed 1 Sep 2017 D.B. Crawley, S. Pless, P. Torcellini, Getting to net zero energy buildings. ASHRAE J. (2009) J.A. Dirks, The impact of wide-scale implementation of net zero-energy homes on the Western grid. ACEEE Summer Study on Energy Efficiency in Buildings, (2010), pp. 60–75 EIA, EIA-930 Data Users Guide and Known Issues, U.S. Energy Information Administration, Washington (2016a). https://www.eia.gov/realtime_grid/docs/UserGuideAndKnownIssues.pdf. Accessed 6 Nov 2017

Potential Impact of Net-Zero Energy Residential Buildings on the US Electric Grid

99

EIA, U.S. Electric System Operating Data. U.S. Energy Information Administration. Washington (2016b). https://www.eia.gov/todayinenergy/detail.php?id27212. Accessed 6 Nov 2017 EIA, Annual Electric Power Review. U.S. Department of Energy Energy Information Administration (EIA) (2017) GMED, 5.2kW/7.2kWh Most Affordable Solar Battery Storage Solution. Global Mainstream Dynamic Energy Technology. Shanghai (2017). https://cdn.enfsolar.com/Product/pdf/ storage_system/5a7bf6824e06f.pdf. Accessed 20 Aug 2018 IEA, Global status report for buildings and construction 2019. In IEA (2019). https://www.iea.org/ reports/global-status-report-for-buildings-and-construction-2019 N. Kampelis, K. Gobakis, V. Vagias, D. Kolokotsa, L. Standardi, D. Isidori, C. Cristalli, F.M. Montagnino, F. Paredes, P. Muratore, L. Venezia, K. Dracou, A. Montenon, A. Pyrgou, T. Karlessi, M. Santamouris, Evaluation of the performance gap in industrial, residential & tertiary near-zero energy buildings. Energ. Buildings 148, 58–73 (2017). https://doi.org/ 10.1016/j.enbuild.2017.03.057 D. Kim, H. Cho, J. Koh, P. Im, Net-zero energy building design and life-cycle cost analysis with air-source variable refrigerant flow and distributed photovoltaic systems. Renew. Sust. Energ. Rev., 118 (2020). https://doi.org/10.1016/j.rser.2019.109508 D. Kim, H. Cho, P.J. Mago, J. Yoon, H. Lee, Impact on renewable design requirements of net-zero carbon buildings under potential future climate scenarios. Climate 9(1) (2021). https://doi.org/ 10.3390/cli9010017 M. Lave, J. Kleissl, Optimum fixed orientations and benefits of tracking for capturing solar radiation in the continental United States. Renew. Energy 36(3), 1145–1152 (2011). https:// doi.org/10.1016/j.renene.2010.07.032 D.H.W. Li, L. Yang, J.C. Lam, Zero energy buildings and sustainable development implications a review. Energy 54, 1–10 (2013). https://doi.org/10.1016/j.energy.2013.01.070 A. Luque, S. Hegedus, Photovoltaic Science Handbook of Photovoltaic Science and Engineering, 2nd edn. (WILEY, 2012). https://doi.org/10.1002/9780470974704 R. Neves, H. Cho, J. Zhang, Pairing geothermal technology and solar photovoltaics for net-zero energy homes. Renew. Sust. Energ. Rev. 140 (2021). https://doi.org/10.1016/j.rser.2021.110749 J. Salom, A.J. Marszal, J. Widén, J. Candanedo, K.B. Lindberg, Analysis of load match and grid interaction indicators in net zero energy buildings with simulated and monitored data. Appl. Energy 136, 119–131 (2014). https://doi.org/10.1016/j.apenergy.2014.09.018 P. Seljom, K.B. Lindberg, A. Tomasgard, G. Doorman, I. Sartori, The impact of zero energy buildings on the Scandinavian energy system. Energy 118, 284–296 (2017). https://doi.org/ 10.1016/j.energy.2016.12.008 SMA America, SMA America Inverter Ratings (n.d.) Solar Design Tool, LG LG230M1C (230W) Solar Panel. Solar Design Tool. Santa Cruz (2017). http://www.solardesigntool.com/components/module-panel-solar/LG/1067/LG230M1C/specifi cationdata-sheet.html. Accessed 1 Sep 2017 Z.T. Tayplor, W. Mendon, N. Fernandez (2015). Methodology for Evaluating Cost-Effectiveness of Residential Energy Code Changes U.S. DOE, EnergyPlus Engineering Reference: The Reference to EnergyPlus Calculations (Issue Version 8.6) (2015a). https://doi.org/citeulike-article-id:10579266 U.S. DOE, EnergyPlus Input Output Reference: the Encyclopedic Reference to EnergyPlus Input and Output (Issue Version 8.6) (2015b) U.S. DOE, and PNNL, 90.1 Prototype Building Models—Medium Office. U.S. Department of Energy. Washington (2016). https://www.energycodes.gov/development/commercial/ prototype_models#90.1. Accessed 15 May 2016 Y. Yang, Optimization of Battery Energy Storage Systems for PV Grid Integration Based on Sizing Strategy (2014) Yaskawa Solectria, YASAKAWA Solectria Solar Commercial Inverters. Yaskawa Solectria Solar. Lawrence (2014). https://www.solectria.com//site/assets/files/1414/sgi_225-500_ datasheet_december_2016_rev_k.pdf. Accessed 1 Sep 2017

Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network Reza Gharoie Ahangar and Hani Gharavi

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 System Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Wind Turbine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Photovoltaic Panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Fuel Cell and Its Associated Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Objective Function and Its Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Particle Swarm Optimization (PSO) Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

102 103 104 104 105 105 106 107 108 109 109 109 112

Abstract

In this chapter, we introduce a hybrid wind turbine, photovoltaic, and fuel cell energy system intending to minimize the overall cost to increase the designed hybrid system’s efficiency. In our study, the associated costs in the objective function consist of initial investment costs, operational and maintenance costs, and the cost related to loss of load. To find the optimal solution with the nonlinear mixed-integer function, we utilized particle swarm optimization algorithm, and

R. Gharoie Ahangar () Department of Information Technology and Decision Science, G. Brint Ryan College of Business, University of North Texas, Denton, TX, USA e-mail: [email protected] H. Gharavi Innovation & Planning Department, Current Network Team, Eirgrid PLC, Dublin, Ireland e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_23

101

102

R. Gharoie Ahangar and H. Gharavi

also we implemented an approximate reliability model to assess the reliability. The findings show that the overall cost of the designed hybrid system is optimized with the present value of the loss of energy expectation and total cost for 20 years, $0.13 (*10−6 ) and $2.59 (*10−6 ), respectively, which are noticeable, with respect to satisfactory ranges of reliability indexes. Keywords

Energy · Reliability · Hybrid systems · PSO algorithm

1

Introduction

The rapid growth of the population and economic globalization and industrialization highlights the role of energy efficiency in the world (Javed et al. 2020). An increase in population and high energy consumption led to some limitations in the power grid system, such as power quality and load density problems (Zhou et al. 2010). Researchers introduce a distributed system in the network to deal with this problem, which can be a reliable solution for this type of problem in energy networks (Kashefi Kaviani et al. 2009). Nowadays, the renewable energy system (e.g., wind, solar) has a significant contribution to the energy networks’ distribution system (Abdin and Zio 2018; Coppitters et al. 2019; Doepfert and Castro 2021; Krebs-Moberg et al. 2020; Nour et al. 2021; Sultan et al. 2021). However, the importance of renewable energy is quite noteworthy, but there are some uncertainties and drawbacks in their products, such as oversize and overdesigning the system components in an energy network (Georgilakis and Katsigiannis 2009). In addition, the unexpected storm and abnormal weather conditions in the last week of February 2021 in Texas, USA, and the rolling blackouts, which happened after 30 years, proved that we could not rely only on a specific energy source and a combination of different energy system is crucial in a network system. To cope with this uncertainty, one solution can be the incorporation of the different sources (e.g., wind and photovoltaics) to improve the reliability of the energy system. Some other issues that reduce the capability of an energy network system are the economic and environmental problems (Javed et al. 2020). There are some solutions to reduce the costs of economic and environmental issues. For example, introducing a hybrid system including fuel cell and wind turbines or using hydrogen tanks to store the excess power of fuel cell and wind turbines during off peak hours can be some of the available solutions. Another solution can be balancing load during peak and off-peak hours using the hybrid system in order to increase the overall efficiency of the energy network. Some hybrid systems use a superconductor magnetic storage, which helps them control the fluctuation of demand in electrical network. In all methods, the goal is to decrease the amount of damage using the hybrid energy system by damping the power oscillations.

Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network

103

A critical factor in designing an energy system is the reliability of the supply of load (Pavlos et al. 2009). There are also some costs in designing an energy network. The costs of the designed network include investment, operation, and maintenance costs. Moreover, other costs are associated with the loss of load in the network. This chapter introduces the design of a hybrid system consisting of fuel cells, wind turbines, and photovoltaics to solve the objectives in question. This chapter contributes to the literature review and existing related topics by proposing a hybrid energy system that consists of different energy sources. The recent storm in Texas, USA, confirmed the necessity of this type of hybrid system. We used reliability indexes, which are modeled as inequality constraints in the optimization procedure, to develop the energy network. Moreover, we implemented particle swarm optimization (PSO) algorithm to improve the system’s reliability through minimization of cost and sizing of the designed structure. The hybrid optimization objective in our study is a nonlinear mixed-integer problem. We applied the PSO algorithm to minimize the objective function of the study considering the reliability of the designed model. In the end, we tested the performance of the hybrid model to ensure the efficiency of our proposed hybrid model. The rest of the chapter proceeds as follows: in Sect. 2 the hybrid system consisting of wind turbine, photovoltaic panel, and fuel cell with its associated details is explained. In Sect. 3 the reliability is presented. In Sect. 4 the objective function and its constraints are described. In Sect. 5 the PSO algorithm is explained. The analysis of results and conclusion are provided in Sects. 6 and 7, respectively.

2

System Details

Figure 1 demonstrates the schematic of our proposed stand-alone DG system consisting of WT, PV, and FC. It shows that available power in WT and PV flows to the network to feed the load. The excess power that exceeded the load demand in the network conveys to an electrolyzer machine, and this excess power is stored in

Fig. 1 The diagram of the proposed hybrid model in this study

104

R. Gharoie Ahangar and H. Gharavi

a hydrogen tank in the form of hydrogen. However, in case of a power deficit, this trend will be inverse, and the stored hydrogen flows to FC to provide the required electric power to ensure the load is met.

2.1

Wind Turbine

The power curve in a manufacturer models is the needed power for WT. The power curve in this study is formulated in the form of polynomials as follows (Kashefi Kaviani et al. 2009; Ntziachristos et al. 2005):

PW G

⎧ m vW −vcut in ⎪ ⎪ ⎨ PW G,max × vrated −vcut in ; vcut in ≤ vW and vrated ≥ vW = PW G,max + Pf url −PW G,max × (vw −vrated ) ; vrated ≤vW and vf url ≥vW vcut out −vrated ⎪ ⎪ ⎩ 0; vW ≤ vcut in and vW ≥ vcut out (1)

where PWG , max indicates the rated output power of WG and Pfurl shows the cutout speeds of WG’s output. We determined m = 3 in our study (Gharavi et al. 2014), which is the exponent in this equation. The vw denotes the speed of wind with its specific height of WG. The equation below shows the role of the wind turbine’s height, which is stated by exponent law: ref

h = vW × vW

h

α

href

(2)

where α = 0.14 in our study , which is a constant, and it is between 0.14 and 0.25 (Kashefi Kaviani et al. 2009; Ntziachristos et al. 2005).

2.2

Photovoltaic Panel

The conventional model represents the output power of PV panels with solar irradiation data through a set of horizontal and vertical components (Assaf and Shabani 2019; Kashefi Kaviani et al. 2009; Marino et al. 2019): PP V =

G × PP V ,rated × ηP V ,conv 1000

G (t, θP V ) = GV (t) × cos (θP V ) + GH (t) × sin (θP V )

(3) (4)

where G shows the perpendicular radiation of surface in an array (W/m2 ), PPV,rated refers to rated power for each of PV’s array when G = 1000 W/m2 , and θ PV represents the angle of PV panel’s tilt. GH (t) is horizontal, and GV (t) is vertical

Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network

105

components of solar irradiation, and the maximum power point tracker’s (MPPT) efficiency is represented by ηPV,conv (Daghigh et al. 2020).

2.3

Fuel Cell and Its Associated Components

We can model the output power of a full cell by multiplying the efficiency rate of the devices by the intensity of the input hydrogen (Giugno et al. 2020; Kashefi Kaviani et al. 2009; Ntziachristos et al. 2005): PF C−inv = Ptank−F C × ηF C

(5)

The following equation represents the available energy which is available in stored hydrogen tanks (Ye et al. 2020): Etank (t) = Etank (t − 1) + Pel−tank (t) × t − Ptank−F C (t) × t × ηstorage

(6)

where Ptank_FC refers to the transformed power from hydrogen tank in t to the FC. Storage efficiency (ηstorage ) in our study is around 95%, and it shows the losses derived from leakage or pump. Equation (7) depicts the relationship between the amount of hydrogen (kg) being stored in the tank and the energy (kWh) associated with that at the time step t: mstorage (t) =

Estorage (t) H H VH2

(7)

For the fuel cell, electrolyzer, and inventor, we calculated the efficiency of the power in the same way as follows: Pel−tank = Pren−el × ηel

(8)

Pinv−load = (PF C−inv + Pren−inv ) × ηinv

(9)

where ηinv shows the efficiency of an inventor, and we assumed the efficiency is a constant number equal to 90% for the entire process (Kashefi Kaviani et al. 2009).

3

Reliability

There are several reliability indexes, and there is a probability of failure for them, which shows the demand satisfaction in a specific period of time. Some of the most widely used indexes are including equivalent loss factor (ELOF), loss of power supply probability (LOPSP), loss of load expected (LOLE), and loss of energy expectation (LOEE) (Pavlos et al. 2009).

106

R. Gharoie Ahangar and H. Gharavi

However, all the indexes mentioned above are investigated in our study, but the main reliability index in this study is the ELOF index. ELOF is selected based on the number of outages and the associated unserved energy (Garcia and Weisser 2006; Hadidian Moghaddam et al. 2018). In our study, the maximum of ELOF in developed countries is around 0.0001 (Kashefi Kaviani et al. 2009), but to design an autonomous system, the designers restrict the ELOF, and its number is close to 0.01 (Garcia and Weisser 2006): N 1 Q(t) EOLF = N D(t)

(10)

t=1

where D(t) is the load (kWh) and Q(t) refers to energy loss at time t in the network. In our study, the probability of forced outage or failure is involved in WT, PV, and FC components in reliability analysis. We assumed all other types of equipment are fully reliable, with 100% efficiency. To calculate the expectation of reliability index, we utilize an approximate mathematical model, which looks at the average output power of WT units and PV arrays, instead of looking at a single outage of WT or PV arrays (Gharavi et al. 2014; Liu et al. 2020): E [Pren ] =

Pren (s) × fp (s)

(11)

s∈S

The probability of encountering state S is represented by fp (s), and the injected power to DC bus in state S through the renewable generated source is Pren (s): E [Pren ] = NW G × PW G × AW G × NP V × PP V × AP V

(12)

where AWG shows the availability of WG and APV represents the access to PV array in the designed system.

4

Objective Function and Its Constraints

In this optimization problem, we seek to find the best optimum points for a set of parameters to design an economical renewable system. Our objective function parameters consist of the number of WT, number and tilt angle of PV, the full cells and the associated capacity, the electrolyzer, hydrogen tank, and inverter to achieve a minimum net present value (NPV) for the total cost of the designed system. Moreover, the different types of costs in this study’s objective function include the cost of ELOF (NPCloss ), the cost of equipment for initial investment, and operational costs for a span of a 20-year lifetime (Gharavi et al. 2014; Hadidian Moghaddam et al. 2018; Simpson et al. 2020):

Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network

Min Ctotal =

NP C i + NP C loss

107

(13)

i

We can depict the operating and maintaining costs as follows: N P C i = Ni × (CC i + RC i × K + MC i × P W A)

(14)

In this equation, K is a factor that converts the replacement cost to NPV cost, and PWA is another factor, which converts the operational cost to NPV cost (Hakimi et al. 2007). The NPV cost of loss is calculated as follows: NP C loss = LOEE × Closs × P W A

(15)

The optimization is performed while the constraints below are satisfied (Gharavi et al. 2014): 1. ELOF should be in the range of its defined limit. 2. The number of each component should be a positive integer variable. 3. The range of angle in the PV panel should take a value between 0 and 90. There is a relationship between the probability of failure and the associated output power as follows: f ail f ail f ail f ail Pren nW G , nP V = NW G − nW G × PW G + NP V − nP V × PP V

5

(16)

Particle Swarm Optimization (PSO) Algorithm

Kennedy and Eberhart suggested the PSO algorithm in (1995), which is based on collaborative movement of birds to find the most optimum point through the shortest way. This algorithm can handle nonlinear, non-differentiable multiobjective functions. The PSO algorithm starts randomly with a group of particles that have multidimensional characteristics such as specific location in space and velocity. To find the optimum solution in the problem, search space is probed by updating generates (Gharoie Ahangar et al. 2021). Each particle, in each step of a population’s movement, is updated through two best values: Xid , which is the position, and Vid , which is the velocity of ith particle related to dth dimension (Stacey et al. 2003). The first best value for each particle is called the local optimum point, pi , and the second value which is the best value for all particles is called the global optimum point, pg . After finding pi and pg , for the next iteration, Xid and Vid of each particle are updated under the following equations (Parsopoulos 2015):

108

R. Gharoie Ahangar and H. Gharavi (K+1)

Vi

(K)

= ωVi

(K) (K) (K) + c2 r2 pg(K) − Xi + c1 r1 pi − Xi (K+1)

Xi

(K)

= Xi

(17)

(K+1)

+ Vi

(18)

where ω is inertia weight; c1 and c2 are learning factors (or acceleration coeffi(k) (k+1) are cients); r1 and r2 are random number between zero to one; Vi and Vi (k) velocity value of each particle in generation k and k + 1, respectively; and Xi (k+1) and Xi are positions of each particle in generation k and k + 1, respectively. To prevent algorithm divergence, final velocity value of each particle, Vi , is limited to lie in the interval [−Vmax , Vmax ], where Vmax is the maximum velocity. The values for w, c1 , and c2 are 0.7, 2, and 2, respectively. The number of iterations and particles are 500 and 200, respectively. In the last stage, the selection phase, the generated vector will be tested by comparing it with the best vector in the previous iteration, and will be repeated as many as the number of iterations, or the algorithm will terminate after reaching to stop circumstance (Zeugmann et al. 2011).

6

Results

In this study, a hybrid energy system is designed, and its operational procedure is explained. The results generated by MATLAB are presented as follows. Table 1 shows the characteristics of components in our hybrid energy system. The hourly data of wind speed and solar irradiation is collected from Illinois, USA. The test function is the IEEE reliability test system with 50 kW at its peak period. In our

Table 1 The equipment specification in our optimized hybrid system

Equipment Wind turbine Photovoltaic arrays Electrolyzer Hydrogen storage tank Fuel cell Inverter

Maintenance cost Capital cost Replacement ($/unitLifetime ($/KW) cost ($/KW) year) (year) Efficiency (%) Availability (%) 1300 1000 48 20 – 96 250

200

400

20

–

96

300 240

250 200

25 15

20 20

75 95

100 100

300 180

250 150

175 8

5 15

50 90

100 99

Note: The equipment specification is adopted from Gharavi et al. (2014), from Illinois, USA; since renewable equipment costs are decreasing slightly, we expect the same range of costs in 2020. Some part of information is extracted from https://www.fixr.com/costs/solar-panelmaintenance, https://pv-magazine-usa.com/

Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network

109

Table 2 The obtained results in our hybrid system Present value of total cost for 20 years (*$10−6 ) 2.59

Investment cost (*$10−6 ) 2.2

Table 3 Final configuration in our hybrid model

Present value of LOEE cost for 20 year (*$10−6 ) 0.13

ELOF LPSP 0.006 0.008

(KW )

(KW )

LOEE OLE (mWh/year) (h/year) 2.2 328 (KW )

(KW )

(◦ )

NWT

NPV

PEL

MT AN K

PF C

PI N V

θP V

7

221

117.9

1414

39.8

44.3

35.4

study, the energy loss cost is 5.6 ($/kWh), the expected life span of the system is 20 years, and the internal rate of return of system (IRR) is 6%. Table 2 represents the results of our optimization for the costs and reliability indexes. It demonstrates the minimum cost for the standard value of ELOF, which is less than the predetermined value of 0.01 that shows the validity of the obtained results. Table 3 shows the configuration results of our optimization for our hybrid system in this study.

7

Conclusions

In this study, a hybrid wind turbine, photovoltaic, and fuel cell energy system is performed. The goal is to minimize the cost of designing the proposed hybrid system to increase the system’s overall efficiency for a life span of 20 years. In the objective function, the included costs consist of initial investment, operation and maintenance, and load loss. The simulated energy design system is performed in the USA. In our study, a nonlinear mixed-integer optimization problem is solved using the PSO algorithm. For reliability evaluation, an approximate reliability model is utilized to provide a higher efficiency solution. The findings reveal that the system’s overall cost is within satisfactory reliability indexes bound with optimal points. Based on the results, we can also conclude that the equipment’s optimal size can vary with the degree of power. Meanwhile, the electrical network yields more emissions than HGPS; it is equitable to choose larger HGPS equipment capacities. Therefore, the higher the level of emission, the higher the capacity for HGPS equipment.

Appendix Nomenclature A Ainv

Wind speed’s coefficient Probability of inverter availability

110

APV AWT C Closs Cmax Cmin Ctotal CCi CD(mt) CP(t) D D(t) E EHST (t) Emax Emin Etotal EC EENS ELF ELFmax Er fHGPS fsystem G(t) h href HHVH2 I K L LOE(t) LOL(t) LOEE LOLE m mHST (t) MCi n N Ni NPV Nvar NWT

R. Gharoie Ahangar and H. Gharavi

Probability of PV array availability Probability of WT availability Cost function Average cost of loss because of unmet load ($/kWh) Cost function, maximum value Cost function, minimum value Total cost Equipment cost of initial investment i ($/unit) Demand cost in month m and time t ($/kWh) Power purchase cost at time t ($/kWh) Vector description, P-R Load demand at time t (kW) Emission function Stored energy of HST at time t (kW) Emission function, maximum value Emission function, minimum value Total emission (kg) Emission cost ($/kWh) Expectation without energy Equivalent loss factor Equivalent load failure, maximum value Emission rate (kg/kWh) Probability function for WT and PV Probability function for WT, PV, and inverter Incident solar irradiation to panel surface (W/m2 ) Height of WT (m) Height of reference (m) Higher heating value for hydrogen (39.7 kWh/kg) Equipment index Constant Useful lifetime of project Loss of energy at time t Loss of load at time t Loss of energy expectation Loss of load expectation Counter Hydrogen stored mass in HST at time t (kg) Maintenance and operation cost of equipment i ($/kWh) Counter Number of times for the loss of load Number of equipment i in HGPS (kW or kg) Total number of PV Number of problem dimension Number of WT

Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network f ail

nP V f ail nW T NPCi NPCloss Pel-HST Pfurl PFC-inv PHGPS-el PHGPS-inv f ail f ail PH GP S nW T , nP V PHGPS (t) PHST-FC (t) Pinv-load Pload (t) Pnetwork,max Pnetwork,max(mt) Pnetwork (t) PPV PV PPV,rated PV Ps PWT PWT,max Pd Pr(t) PWA Qs R round RCi s T Ts vW vcut in vcut out vrated h vW ref vW β Δt ηel ηFC

111

Failure number of PV Failure number of WT Net present cost for the equipment i ($/kWh) Net present cost for the unmet load ($/kWh) Electrolyzer’s output power (kW) WT’s output power at cutout wind speed (kW) Fuel cell’s output power (kW) Input power for the electrolyzer (kW) Renewable resources injected power to inverter (kW) Renewable power resources injected to DC bus Produced power by HGPS at time t (kW) Power of HST injected to fuel cell (kW) Injected power to the load (kW) Load demand at time t (kW) Electrical network’s maximum power (kW) Electrical network’s maximum power in month m, time t (kW) Electrical network’s purchase power at time t (kW) Array’s output power (kW) Array’s nominal power (kW) Probability of state s occurrence WT’s output power (kW) WT’s maximum output power (kW) Demand price ($/kWh) Electrical network’s power price at time t ($/kWh) Interest rate Loss load (kWh) Same dimension vector as P Function of rounding Replacement cost for the equipment i ($/unit) All possible states Time index Duration of loss of load at state s Wind speed (m/s) Cut-in wind speed for WT (m/s) Cutout wind speed for WT (m/s) Wind speed rate (m/s) Wind speed height h (m/s) Reference height of wind speed (m/s) A number greater than 1 Simulation Time (1 day) Efficiency of electrolyzer Efficiency of FC

112

ηHST ηinv ηPV,conv θ θ PV

R. Gharoie Ahangar and H. Gharavi

Storage system’s efficiency Efficiency of inverter Efficiency of converter DC/DC Add an angle value to x Incident solar irradiation angle (rad)

References I.F. Abdin, E. Zio, An integrated framework for operational flexibility assessment in multi-period power system planning with renewable energy production. Appl. Energy 222, 898–914 (2018). https://doi.org/10.1016/j.apenergy.2018.04.009 J. Assaf, B. Shabani, A novel hybrid renewable solar energy solution for continuous heat and power supply to standalone-alone applications with ultimate reliability and cost effectiveness. Renew. Energy 138, 509–520 (2019). https://doi.org/10.1016/j.renene.2019.01.099 D. Coppitters, W. De Paepe, F. Contino, Surrogate-assisted robust design optimization and global sensitivity analysis of a directly coupled photovoltaic-electrolyzer system under techno-economic uncertainty. Appl. Energy 248, 310–320 (2019). https://doi.org/10.1016/ j.apenergy.2019.04.101 R. Daghigh, H. Oramipoor, R. Shahidian, Improving the performance and economic analysis of photovoltaic panel using copper tubular-rectangular ducted heat exchanger. Renew. Energy (2020). https://doi.org/10.1016/j.renene.2020.04.105 M. Doepfert, R. Castro, Techno-economic optimization of a 100% renewable energy system in 2050 for countries with high shares of hydropower: The case of Portugal. Renew. Energy 165, 491–503 (2021). https://doi.org/10.1016/j.renene.2020.11.061 R.S. Garcia, D. Weisser, A wind–diesel system with hydrogen storage: Joint optimisation of design and dispatch. Renew. Energy 31(14), 2296–2320 (2006). https://doi.org/10.1016/ j.renene.2005.11.003 P.S. Georgilakis, Y.A. Katsigiannis, Reliability and economic evaluation of small autonomous power systems containing only renewable energy sources. Renew. Energy 34(1), 65–70 (2009). https://doi.org/10.1016/j.renene.2008.03.004 H. Gharavi, F. Mohammadi, H. Gharoei, S. Ghanbari-Tichi, R. Salehi, Optimal fuzzy multiobjective design of a renewable energy system with economics, reliability, and environmental emissions considerations. J. Renew. Sustain. Energy 6(5), 053125 (2014). https://doi.org/ 10.1063/1.4898634 R. Gharoie Ahangar, R. Pavur, H. Gharavi, A global optima search field division method for evolutionary algorithms. J. Oper. Res. Soc. 72 (2021). https://doi.org/10.1080/ 01605682.2021.1890531 A. Giugno, L. Mantelli, A. Cuneo, A. Traverso, Performance analysis of a fuel cell hybrid system subject to technological uncertainties. Appl. Energy 279, 115785 (2020). https://doi.org/ 10.1016/j.apenergy.2020.115785 M.J. Hadidian Moghaddam, A. Kalam, S.A. Nowdeh, A. Ahmadi, M. Babanezhad, S. Saha, Optimal sizing and energy Management of Stand-alone Hybrid Photovoltaic/wind system based on hydrogen storage considering LOEE and LOLE reliability indices using flower pollination algorithm. Renew. Energy (2018). https://doi.org/10.1016/j.renene.2018.09.078 S.M. Hakimi, S.M.M. Tafreshi, A. Kashefi, Unit sizing of a stand-alone hybrid power system using particle swarm optimization (PSO). 2007 IEEE International Conference on Automation and Logistics (2007). https://doi.org/10.1109/ical.2007.4339116 M.S. Javed, T. Ma, J. Jurasz, F.A. Canales, S. Lin, S. Ahmed, Y. Zhang, Economic analysis and optimization of a renewable energy based power supply system with different energy storages for a remote island. Renew. Energy (2020). https://doi.org/10.1016/j.renene.2020.10.063

Economical and Reliable Design of a Hybrid Energy System in a Smart Grid Network

113

A. Kashefi Kaviani, G.H. Riahy, S.M. Kouhsari, Optimal design of a reliable hydrogen-based stand-alone wind/PV generating system, considering component outages. Renew. Energy 34(11), 2380–2390 (2009). https://doi.org/10.1016/j.renene.2009.03.020 J. Kennedy, R.C. Eberhart, Particle swarm optimization. Proceeding of IEEE International Conference on neural networks 4, 1942–1948 (1995) M. Krebs-Moberg, M. Pitz, T.L. Dorsette, S.H. Gheewala, Third generation of photovoltaic panels: A life cycle assessment. Renew. Energy (2020). https://doi.org/10.1016/j.renene.2020.09.054 S. Liu, L. Wei, H. Wang, Review on reliability of supercapacitors in energy storage applications. Appl. Energy 278, 115436 (2020). https://doi.org/10.1016/j.apenergy.2020.115436 C. Marino, A. Nucara, M.F. Panzera, M. Pietrafesa, V. Varano, Energetic and economic analysis of a stand-alone photovoltaic system with hydrogen storage. Renew. Energy 142, 316–329 (2019). https://doi.org/10.1016/j.renene.2019.04.079 A.M.M. Nour, A.A. Helal, M.M. El-Saadawi, A.Y. Hatata, A control scheme for voltage unbalance mitigation in distribution network with rooftop PV systems based on distributed batteries. Int. J. Electr. Power Energy Syst. 124, 106375 (2021). https://doi.org/10.1016/j.ijepes.2020.106375 L. Ntziachristos, C. Kouridis, Z. Samaras, K. Pattas, A wind-power fuel-cell hybrid system study on the non-interconnected Aegean islands grid. Renew. Energy 30(10), 1471–1487 (2005). https://doi.org/10.1016/j.renene.2004.11.007 K.E. Parsopoulos, Particle swarm methods. Handbook of heuristics, 1–47 (2015). https://doi.org/ 10.1007/978-3-319-07153-4_22-1 J. Simpson, E. Loth, K. Dykes, Cost of valued energy for design of renewable energy systems. Renew. Energy (2020). https://doi.org/10.1016/j.renene.2020.01.131 A. Stacey, M. Jancic, I. Grundy, Particle swarm optimization with mutation. The 2003 Congress on Evolutionary Computation, 2003. CEC ‘03. 2, 8–12 (2003). https://doi.org/10.1109/ cec.2003.1299838 H.M. Sultan, A.S. Menesy, S. Kamel, A. Korashy, S.A. Almohaimeed, M. Abdel-Akher, An improved artificial ecosystem optimization algorithm for optimal configuration of a hybrid PV/WT/FC energy system. Alex. Eng. J. (2021). https://doi.org/10.1016/j.aej.2020.10.027 Y. Ye, J. Lu, J. Ding, W. Wang, J. Yan, Numerical simulation on the storage performance of a phase change materials based metal hydride hydrogen storage tank. Appl. Energy 278, 115682 (2020) T. Zeugmann, P. Poupart, J. Kennedy, X. Jin, J. Han, L. Saitta, et al., Particle swarm optimization. Encyclopedia of machine learning, 760–766 (2011). https://doi.org/10.1007/978-0-387-301648_630 W. Zhou, C. Lou, Z. Li, L. Lu, H. Yang, Current status of research on optimum sizing of standalone hybrid solar–wind power generation systems. Appl. Energy 87(2), 380–389 (2010). https:/ /doi.org/10.1016/j.apenergy.2009.08.012

Virtual Power Plants and Integrated Energy System: Current Status and Future Prospects Sambeet Mishra, Chiara Bordin, Madis Leinakse, Fushuan Wen, Robert J Howlett, and Ivo Palu

Contents 1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Scope and Key Contributions of the Chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 VPPs, Microgrids, and Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Technical Challenges in a VPP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Mitigation of Power Quality Disturbances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 VPP with IES: An Energy Informatics Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Value of VPP with an IES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Sector Coupling Options for VPPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Storage Options for VPPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Scale of a VPP in an Electricity Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 VPPs of the Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.1 Toward Intelligent VPP Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

116 118 119 122 123 125 127 130 132 134 137 138 139 139

S. Mishra () Tallinn University of Technology, Tallinn, Estonia Danish Technical University, Kongens Lyngby, Denmark e-mail: [email protected] C. Bordin () The Arctic University of Norway, Tromsø, Norway e-mail: [email protected] M. Leinakse · F. Wen · I. Palu Tallinn University of Technology, Tallinn, Estonia e-mail: [email protected]; [email protected]; [email protected] R. J. Howlett KES International, Shoreham-by-sea, UK e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_73

115

116

S. Mishra et al.

Abstract

The power system is undergoing a digitalization, decarbonization, and decentralization. Economic incentives along with resiliency and reliability concerns are partly driving the transition. In the process of decentralization, local energy markets are forming at various places. A virtual power plant (VPP) is a by-product of this digitalization capitalizing on the opportunity to further promote renewable resources, demand-side flexibility, and sector coupling. A VPP enables resilient operation of power system while assembling small- to large-scale generation units and demand-side flexibility. Specifically, during the pandemic uncertainty, virtual work meets virtual power plants. A VPP has two both cyber and physical components. On one side, the physical component presents the operational challenges in terms of security, stability, reliability, and efficiency. On the other side, the cyber component introduces the challenges on communication, computation, security, and privacy. A VPP synthesizes synergies between the cyber and physical components, thereby harnessing the potential in terms of enhancing energy efficiency and reducing the cost. The objective of this chapter is to introduce the virtual power plant and integrated energy system with associated concepts, terminology, and relation thereof. The secondary objective is to categorize the key concepts while highlighting subsequent issues in planning, operations, and control of a VPP with an integrated energy system. Moreover, this chapter knits together the concepts and challenges in realizing virtual power plants with integrated energy systems. Keywords

Cyber-physical energy systems · Energy informatics · Integrated energy systems · Virtual power plants

1

Introduction

Recent developments in renewable energy generation and electrical vehicles (EVs), the widespread use of combined heat and power (CHP) technology, and the emerging power-to-gas (P2G) devices in power systems have provoked significant changes in energy production and consumption patterns and at the same time presented some new opportunities toward an environmentally sustainable energy system. Therefore, the traditional multi-energy network has been endowed with new implications: the electrical power system, natural gas system, electrified intelligent transportation system, and district heating system have formed a closed loop within a broader integrated energy system (IES) (Bai et al. 2016). An IES is a heterogeneous system with a hierarchical and multimodal structure, as well as a multilevel network topology (Wu et al. 2019). The IES can be defined as a system that operates electric power and heating components in an optimal manner, to serve electric, heating, and transportation demands to the end users. The subsets of an IES such as electric

Virtual Power Plants and Integrated Energy System: Current Status and. . .

117

power and heat systems are coupled through CHP (Brown et al. 2018) units and energy hubs (Zhang et al. 2015). An energy hub is a multi-energy system platform with various combinations of multiple kind energy inputs and outputs (Jadidbonab et al. 2020). An energy hub benefits from the synergy of energy carriers. Based on the price levels of various kinds of energy, the operation strategy of an energy hub can be optimized to maximize a specified objective function (Emarati et al. 2020). Energy storage has the potential to act as a linkage among different sectors of an IES (Hemmati et al. 2016) for implementing optimal operation of an IES. The energy storage can broadly be classified into electrical and thermal. Linking the energy storage systems could mitigate the variations from renewable resources alongside of optimal energy conversion to meet the load demand. In a power system, the power supply must meet the load demand in real time; otherwise, frequency and voltage can deviate from the normal operating regime. This deviation could lead to damage of devices, brown outs, outages, and even blackouts. The alternating current (AC) power grid itself has no medium of storage; therefore, the power in and out must be controlled to maintain the power balance. The traditional dispatchable resources can be moderated with a reasonable time delay. The portfolio of generation includes relatively small numbers of dispatchable power plants with large capacity. As renewable generation units take a sizable proportion of the portfolio of generation and its non-dispatchable generation output poses substantial challenges to maintain the balance, then on the one hand, curtailing the renewable energy generation is not a viable option, since the clean energy is wasted, and on the other, some kind of renewable generation such as solar and wind power will depend on its nature, and the production cannot be ramped up on demand. Dispatchable generation resources are necessary to maintain the security of the power supply through load following, frequency and voltage regulation, and reserve power (Papaefthymiou and Dragoon 2016). The renewable energy resources are usually in the form of setting of many small distributed generators which behave independently. A power network with a high share of renewable energy would require power storage units to charge during power surplus and discharge during power deficits – to maintain the power balance. Batteries, a mode of energy storage, are commonly used since they can charge and discharge rapidly to offset any power imbalances. The power grids may install very large battery banks, typically in the size of a dispatchable power unit, or utilize small residential battery units already installed in individual households. The residential batteries serve as backup generators and contribute to the better utilization of the energy. The behind-the-meter battery units can be aggregated together at residential and commercial buildings to form a VPP (Boampong and Brown 2020; El Bakari and Kling 2010; Plancke et al. 2015; Zhang et al. 2019). A VPP can also constitute large battery banks coupled with large wind farms or solar plants. Accordingly, a VPP could serve not only as a backup but also as a support to the grid in the maximum utilization of renewable energy generation (Pudjianto et al. 2007). This leads to significant cost reduction in grid control and motivates new and small renewable energy installations. The electric and heat load demands vary with weather, time of the year, and special occasions. In a typical day, the

118

S. Mishra et al.

peak load appears for a short span of time. To meet the peak load, high capacity of generators are built; however, they are unused most of the time outside of the peak hours. Alternatively, the balance power is purchased at a significant premium from neighbors with excess generation (Mishra et al. 2019). Power can be cheaper if the demand is offset and the load curve is flattened. However, charging the battery banks just after the peak hours might extend the peak hours (Babaei et al. 2019). Therefore, the charging is spread out over a period of time during off-peak hours. Moreover, a VPP can facilitate optimal energy management for distributed energy resources with stochastic renewable energy generation (Alahyari et al. 2020). The coordinated planning and operation, as well as optimal strategies of an IES, have raised widespread concerns regarding a reasonable utilization of multivector energy resources. The IES can also enhance the accommodation capability for renewable energy generation. However, this has not yet been systematically investigated in the existing literature. Furthermore, the potential of different kinds of energy resources as a service for the future electricity market is yet to be investigated.

1.1

Scope and Key Contributions of the Chapter

There is a significant progress in small-scale distributed generation, behind-themeter generation, energy storage options, EVs, and flexibility at the distribution grid level. When aggregated, these small-scale units can unlock the total system flexibility while still being environmentally friendly. In order to facilitate such a transition, these assets need access to the energy market. A VPP is essentially a virtual layer on the top of the physical power units. A VPP provides a platform to access and actively participate in the market for the small-scale units. Consequently, the energy system is becoming decentralized and distributed while the digitization acts as an enabler. This transition also benefits the generators, retailers, distribution, and transmission network providers through demand shifting, reduction in investments, and reduction in network reinforcement costs while lowering reserve power requirements while improving the reliability and security of supply. The energy market framework has experienced a paradigm shift toward peer-to-peer trading in the last few years as reported by numerous studies. New business models, actors, and participants are emerging to engage in a consumer-oriented energy system. However, this shift poses many challenges in the form of decision-making problems. Many pilot projects are ongoing in the USA, Europe, and Australia to realize the VPP’s potential. However, there is a gap in listing the range of issues in terms of decision-making challenges and opportunities for a VPP in an IES. One of the key challenges in deploying VPP is resiliency, specifically how to maintain or improve the grid resiliency with the virtualization of power generation system. A VPP improves resiliency through replacing one large power generation unit with multiple small-scale generators. This process improves the fault tolerance of the system and avoids the cascading of faults. However, this process also brings challenges associated with power quality (PQ). This chapter sets out to summarize the emerging trends alongside providing an overview of the challenges that arise when coupling the concept of the VPP together

Virtual Power Plants and Integrated Energy System: Current Status and. . .

119

with the concept of integrated energy systems. The objective of this chapter is to provide a holistic view on the current state and future directions of VPPs. For this purpose, a big picture of the role of VPPs with an integrated energy system will be presented. The VPP is also discussed as a cyber-physical system with physical and cyber components’ potential connections with IES and with an energy hub as the linkage between them. The power system resiliency in the context of a VPP with an IES is discussed in detail when addressing the challenges and provisions. Moreover, this chapter discusses storage options for VPPs, as well as opportunities for its participation in the energy markets, and directions for intelligent solutions of future VPPs. In summary, the proposed work provides a comprehensive and systematic overview of the challenges and potentials associated with a VPP together with IES. This is achieved by setting a context for decision-making considering the emerging trends in the energy market while considering the technical challenges in the physical layer. The current section establishes the context and key terminology. Section 2 introduces and defines the concept of a VPP compared to the concepts of microgrids and energy hubs while associated technical challenges are discussed in Sect. 3. A VPP with an IES as a cyber-physical system is discussed in Sect. 4. Subsequently the value proposition of a VPP with an IES is discussed in Sect. 5. In Sect. 6, the sector coupling for linking electrical and heating sectors is discussed. The storage options, both thermal and electrical, are discussed in Sect. 7. The VPP participation in the energy markets is addressed in Sect. 8. The future of a VPP based on current trends is covered in Sect. 9. Finally some conclusions are drawn in Sect. 10.

2

VPPs, Microgrids, and Energy Systems

Traditionally, the impediment to entry for participation in the electricity market offsets individual small-scale entities such as a PV plant participating independently. The emergence of distributed and low-investment generation units, the incentives for increasing the share of renewable generation, and the demand-side participation are among the key issues going forward to a carbon-neutral future. This chapter proposes to classify VPPs into the single-owner physical VPP (P-VPP) and multi-owner cyber VPP (C-VPP). A P-VPP has a portfolio of assets which are owned by the VPP. A C-VPP has a portfolio of assets which are owned by different parties. Both VPPs trade as a single unit in the energy market, while C-VPP allows peer-to-peer trading within the VPP through a local market formation. In addition, C-VPP is more flexible in terms of asset types and size. This means heating assets can be a part of the C-VPP and the size of the asset can be dynamically changed if a new party joins the coalition. This latter definition therefore unfolds the concept of VPP with IES that has not yet been discussed and investigated in the literature. Note that both P-VPP and C-VPP are a portfolio of technologies operated through a cyber-platform. However, the former physically owns the assets, while the latter forms multiparty agreements to operate as one entity.

120

S. Mishra et al.

A VPP is a portfolio of decentralized and distributed units of the power network. The units can be power generation, storage, and demand-side flexibility. The objective of a VPP is to collectively trade the transactive energy (power, flexibility, and reserve power) in the electricity market. The cluster of individual assets are pooled together as a portfolio that is called a VPP. An asset can be an individual storage unit or the demand-side flexibility of a consumer. As with a conventional plant, a VPP exerts a degree of control through switches to maintain the system’s stability. A VPP resides between the transmission and distribution system operators. While an individual power plant is limited to the granularity, a VPP can further the scope by integrating smart meter data into the balancing of demand and supply. Typically a VPP receives these control signals from the transmission system operator. A VPP is tightly integrated with the distribution system operator who owns and controls the medium to low-voltage power network. The supervisory control and data acquisition (SCADA) is often used by the system operators for this purpose. Like the conventional power plant, a VPP adjusts the portfolio through smart algorithms, to respond quickly and effectively to price signals from the power and ancillary service markets. In contrast to the conventional power plant, a VPP is decentralized and distributed (Asmus 2010). Due to the modular nature of a VPP, the chances of failure are far lower compared to a conventional large power plants. In addition, the condition of assets can be better integrated to the total risk quotient in the risk portfolio of a VPP. A VPP is essentially grid-connected and centrally controlled. The VPP can be large in size with more flexibility potential. A VPP is dependent on smart meters and associated technology to form a virtual market environment, which is flexible and open to participation. More and more power plants are employing smart algorithms for smart grid demand projection, flexibility calculations, and power generation from weather patterns. A VPP is essentially a software platform that can enable a higher degree of accuracy and effectiveness on this aspect. A microgrid represents a localized and miniature power systems and has a localized control system for efficient energy management. A microgrid can be in the form of an isolated grid, such as an island, or in a grid-connected format. Microgrids engage the inverters and smart switches for effective control. Typically microgrids exist in the form of islands or military bases, where the isolation is due to the expensive cost of connection or intent to remain self-contained units. A VPP may contain a cluster of microgrids or individual units among its assets. An energy hub is a sector coupling instrument linking multiple energy carriers. The energy carriers range from energy generation (such as PV or wind) and energy conversion (power to X or gas to X) to energy storage (i.e., hydrogen, battery banks, and EVs). An energy hub provides local control, flexibility, and accessibility to the overall system. As it happens with microgrids, the energy hub may be deployed in different spatial scales ranging from a building to a city. An energy hub essentially provides the integration of uncertain renewable energy resources. An energy hub facilitates greater flexibility in the overall system while enabling transactive energy. The energy hubs are modular in structure, allowing the integration of new energy carriers or storage units while offering system flexibility.

Virtual Power Plants and Integrated Energy System: Current Status and. . .

121

On one hand, an energy hub can soak the excess electricity from the grid by converting electricity to heat, cold, and other synthetic fluids. Thereby, an energy hub aids in balancing the overcapacity or overproduction and optimal power reserve allocation. An energy hub is in fact a tool for congestion management. On the other hand, an energy hub can act as a reserve power unit when the prices are high or when there is a scarcity in production. They do this by producing electricity from the multitude of renewable sources, heat to power technologies, and synthetic biofuels. Consequently, an energy hub can increase the total efficiency of a system such as a biomass plant, by absorbing electricity and releasing either electricity or heat to aid in peak shaving. In Gerami Moghaddam (2018), the scheduling of a VPP containing multiple smart energy hubs is demonstrated. The VPP expands the geographical scope, while the energy hub acts as a local and physical control unit. The VPP and the energy hub may adopt a leader-follower scheme where the VPP acts as the virtual platform for energy transactions and the energy hub acts as the physical connection enabling, and thereby following, the VPP. The flexibility offered by the energy hub would be the upper capacity/limit for the VPP platform for transactions. A VPP renders an optimal and balanced way to integrate the distributed and decentralized energy resources toward the purpose of sustainable energy. The hierarchy of actors in a power network is presented in Fig. 1. As outlined in the previous paragraphs, a VPP can act as a local energy market where both small-scale units, such as demand-side flexibility, and large-scale units, such as large-scale PV

Large dispatchable generaon

Smaller non-dispatchable generaon

Energy Market TSO DSOs

IES 1

IES 2

IES 3

Energy hub E

VPP Microgrid Fig. 1 Hierarchy of integration of a VPP in power system

IES 4

122

S. Mishra et al.

Heang/ Cooling

Transport Electric

Integrated Energy System

Generaon

Demand

Fig. 2 End-user energy consumption loop

power plant, can participate. Under the energy market, there are TSOs and DSOs. Each DSO has an integrated energy system underneath representing a heating and electric loop. Both loops are tied at the source and origin. The origin could be a CHP plant, and the source is a consumer who might consume heat energy from heating loop and electricity from electric loop. An energy hub presents junctions between the loops that can store and transform from one form to another. VPPs normally operate at the DSO level with assets dispersed across the network. The VPP portfolio can hold assets from a heating and electric network. The VPP operates in the energy market within the TSO on the same level as a DSO while integrating units from multiple DSOs and microgrids. Islanded microgrids can also participate in a VPP. However, since there are no physical links to the grid, they are placed as an extension to the DSOs. Figure 2 presents the enduser energy consumption loops. At one end of the loop, there is demand, and other end is generation forming loops for different types of energy consumption: heating/cooling, transportation, and electric. The loops are proportional to the share of energy consumption. For instance, the energy consumption for heating or cooling is higher in comparison to transport and electric. An IES sets out to tie the loops of generation together, thereby increasing the total system throughput.

3

Technical Challenges in a VPP

The VPP is primarily a software platform with switches. Subsequently, there are a range of challenges that arise in terms of distributed energy and software. A VPP platform needs to collect, process, and render decisions for portfolio management in

Virtual Power Plants and Integrated Energy System: Current Status and. . .

123

real time. The sources range from smart meters to bids from small-scale generation units. However, there are often inconsistencies in the data management from various stakeholders and clients. Where the data structure of a VPP do not fit with a client’s data structure, there is a significant cost of development. In the energy sector, new business models are formed frequently and adapted to the regulations and changing landscape. Accordingly, consistent software protocols need to be in place. A microservice-oriented architecture for a VPP is presented in Wickert et al. (2020). A software architecture to manage the smart metering systems with existing software for outage and workforce management is presented in Vukmirovi´c et al. (2010). Finally, a cloud computing solution for the VPP platform with reduced infrastructure cost for a VPP is presented in Aldegheishem et al. (2020). The output of renewable energy generation is influenced by environmental characteristics like solar radiation and wind speed. This reliance can lead to power quality disturbances (Ray et al. 2012). A high penetration of renewable energy generation can cause voltage fluctuations, current harmonics, voltage harmonics, voltage swell or sag, unbalance, malfunction of protective devices, overloading, and failure of electrical equipment (Ray et al. 2013; Alkahtani et al. 2020). A typical distribution of power quality (PQ) disturbances by duration shows that disturbances lasting less than 1 s far outnumber others in occurrence (Thallam and Heydt 2000). The voltage fluctuations are one of the major power quality concerns that distributed renewable energy generation imposes on the system (Hariri and Faruque 2014). The intermittent nature of renewable energy generation causes fast voltage fluctuations (fast changes in voltage amplitude) (Antoniadou-Plytaria et al. 2017). The voltage deviations are further amplified by the large R/X ratio of mediumvoltage and especially low-voltage networks (Antoniadou-Plytaria et al. 2017). The voltage rises caused by the distributed generation can interfere with the operation of tap changers, since the voltage reference is no longer indicative of the voltage profile of the feeder (Walling et al. 2008). The voltage fluctuations caused by the fast changes in the weather conditions can have detrimental effects on the voltage regulation equipment present in the feeder, such as on-load tap changing transformers, switched capacitor banks , and step voltage regulators (Hariri et al. 2015). Premature wear and tear can occur due to the increased numbers of operations of these devices (Hariri et al. 2015). The distributed renewable energy generation can cause high voltages when interconnected in small residential areas sharing a distribution transformer (Walling et al. 2008). If the transformer primary voltage is already at the upper limit, the DR can reduce the voltage drop through the transformer and secondary conductors, which will cause high voltages to be experienced by other customers on the transformer (Saint and Friedman 2002).

3.1

Mitigation of Power Quality Disturbances

Conventionally, the voltage at service locations is maintained by utilizing fixed designs of the system (e.g., conductor selection, substation, and distribution

124

S. Mishra et al.

transformer tap settings and fixed capacitor banks) and by voltage control equipment such as automatic load tap changers, step-type voltage regulators (SVR), and switched capacitors (Walling et al. 2008). The design of the feeder is based on the assumption that the loading profile follows a predictable pattern: with real power loading on the feeder causing voltage to decrease monotonically from the substation (Walling et al. 2008). SVR controls continuously monitor voltages and load currents to adjust tap positions accordingly (Walling et al. 2008). Capacitors (switched and fixed) compensate reactive current, reducing the current from the source to the capacitor location, resulting in reduced line voltage drop (Walling et al. 2008). The currently used voltage regulation methods with conventional voltage/voltampere reactive (VAR) control devices cannot respond well and promptly to the voltage limit violations that may occur due to renewable energy injection and plugin electric vehicle charging (Antoniadou-Plytaria et al. 2017). One way to mitigate the adverse effects caused by the interaction of renewable energy generation and voltage regulating devices is control coordination (Ranamuka et al. 2013, 2016, 2017; Kulmala et al. 2014; Liu et al. 2012; Viawan and Karlsson 2007). A review of communication-based non-centralized control schemes that can be applied specifically to voltage regulation of distribution networks can be found in Antoniadou-Plytaria et al. (2017). Another mitigation method is the use of a dynamic voltage restorer (DVR), which can be used for handling voltage sags and swells and for damping voltage fluctuations (Alkahtani et al. 2020). Power quality disturbance mitigation by the use of DVR is proposed and discussed in Al-Shetwi et al. (2020). According to Alkahtani et al. (2020), DVR is one of the best devices for mitigating power quality issues in the conventional power system and microgrids. In recent years, static synchronous compensators and static VAR compensator have been used extensively for solving many PQ issues that were caused by renewable energy generation integration (Alkahtani et al. 2020). The use of flexible AC transmission systems devices for PQ improvement in integrated wind energy system is discussed in Yuvaraj et al. (2011). A static VAR compensator and static synchronous compensator are used for overcoming sags (Al-Shetwi and Sujod 2018; Molinas et al. 2008) and compensating reactive power (Ilango et al. 2012; Chavhan et al. 2015). A static synchronous compensator is used for damping voltage fluctuations in Lee et al. (2011) and Chaudhari et al. (2015). A promising solution (Hossain et al. 2018) for mitigating power quality disturbances such as sag, swell, and flicker (Zhao 2016; Edomah 2009; Chen et al. 2013) can be a unified power quality conditioner (Honrubia-Escribano et al. 2015; Khadkikar 2011; Kesler and Ozdemir 2010; Ghosh and Ledwich 2001; Graovac et al. 2007; Khadkikar et al. 2006). A unified power quality conditioner is a complete configuration of hybrid filters, which is identified as a multifunctional power conditioner utilized to compensate for different voltage disturbances (Alkahtani et al. 2020).

Virtual Power Plants and Integrated Energy System: Current Status and. . .

4

125

VPP with IES: An Energy Informatics Approach

The concept of VPPs is highly interdisciplinary, touching upon subjects ranging from engineering to computer science and economics, together with power and energy systems. From this point of view, VPPs fall fully into the novel domain of energy informatics, which aims at “exploring the intersection of informatics, power engineering and energy economics” (Lehnhoff and Nieße 2017). Renewable VPPs, in particular, can make a decisive contribution to the main scope of energy informatics by reaching the two main goals – energy efficiency and renewable energy supply (Goebel et al. 2014) – thus facilitating the global transition toward sustainable and resilient energy systems (Springer). This section aims to present an energy informatics approach to address VPP with IES. Figure 3 illustrates the links between VPPs and computer science in the broad context of energy informatics. As shown in Fig. 3, the concept of the VPP can be described by two main parts. First we identify a physical space, which is represented by the physical resources available for energy production, classified as renewable sources (i.e., wind plants, biomass plants, solar plants) and conventional sources, together with the transmission and distribution grid infrastructures. Within the physical space, sector coupling (power-to-X technologies) has an important role,

Fig. 3 Overview of a VPP in the context of energy informatics

126

S. Mishra et al.

to interconnect and integrate the energy-consuming sectors – transport, industrial, residential, and commercial – with the power-producing sector. The physical space outlined above can be investigated, understood, and controlled through modern mathematical and computer science techniques, which together form part of a so-called cyberspace. A cyberspace can successfully function through four main tasks that are strongly interconnected: • • • •

Learn: Understand the data. Predict: Forecast and generate new data. Model: Build technological mathematical optimization models. Optimize: Utilize the data and the models to make optimal decisions that can positively influence the physical space.

The first task is “learning,” which means to gain a deep understanding of the data gathered from the physical space, identifying patterns and understanding the peculiar properties and trends. Once the data has been understood and handled through the learning process, they can be utilized for the second task which is “predicting.” This refers to the ability to forecast the future based on the information previously obtained in the available dataset. Typical forecasts that are necessary within a VPP are related to demand, weather, price, production, etc. Within the learning and forecast tasks, a key role is played by computer science subjects such as machine learning, big energy data, data analytics, artificial intelligence, and database systems. Once the learning and prediction tasks are over, the key is how to utilize this new generated knowledge to build mathematical optimization models that represent the energy systems and the main technical properties of the resources involved. Finally, the knowledge, the data, and the models can then be utilized within optimization tasks, in order to make optimal decisions that can positively impact the physical space where the VPP resources are located. Such optimal decisions refer to both investment decisions and operational decisions. Investment decisions deal with optimal design which can, for instance, impact the optimal portfolio of a flexibility mix within the energy market. Operational decisions deal with optimal real-time control of the existing VPPs where the main objectives can be (but may not be limited to) profit maximization, together with emission reduction. The optimization tasks involve both applied mathematics (with particular regard to operations research, in the form of mathematical optimization in general and smart energy and power system modeling in particular (Bordin et al. 2020)) and programming skills that allow the development of decision support systems (DSS) tools for the specific VPP application and objectives. Within the DSS tools, key subjects play important roles, such as parallel computing (especially when scalability issues arise), data processing, data integration, data synthesis, and visualization. Figure 4 shows an overview of a DSS for a VPP application and the subjects involved. The physical space and the cyberspace introduced above are linked through two main tasks. “Monitoring” allows the transferring of data from the physical space into the cyberspace (a typical example in this case is represented by the use of

Virtual Power Plants and Integrated Energy System: Current Status and. . .

127

Fig. 4 A proposed decision support system architecture for VPP

sensors, devices, and smart meters). In contrast, “decision and control” takes the decisions developed in the cyberspace and implements them back into the physical space for an optimized management of the VPP. Through monitoring and control, by linking a physical space to a cyberspace, the overall described system defines the so-called cyber-physical energy system (CPES) (Rasmussen et al. 2017). By adding the perspective of CPES to the concept of VPP, the concept of VPP-CPES is identified. On top of the CPES defined above, it is possible to add the Internet of Things (IoT). Through IoT, the VPP-CPES can be connected to the internet, and decisions can be automatized and enhanced. A typical example of a VPP would be the possibility to connect to the market and optimize the operational use of the flexibility mix, through bidding strategies, arbitrage, and demand response. IoT is therefore at the heart of the VPP-CPES, with connections to all the main tasks. Through IoT, the data gathered through monitoring can be uploaded onto the cloud. Moreover, the outcome of the data manipulation (through learning, predictions, and optimization) can be accessed remotely. In addition, control and decisionmaking tasks can be enhanced through cloud computing, thus avoiding the barrier of software installation and maintenance inside the companies’ computers. A cloudbased service can perform software updates seamlessly and can deliver a generic product that can be used by everyone independent of the knowledge in data science.

5

Value of VPP with an IES

VPPs are typically comprised of flexible loads, energy storage units, and dispatchable and non-dispatchable resources. These geographically dispersed resources are aggregated as a VPP participating in the energy market (Morales et al. 2013).

128

S. Mishra et al.

Pico grid • Data mining • Load profiling • Demand side management • Flexibility

Micro grid • Modularity • Flexibility • Resilience • Economic • Clean energy

Macro grid • Reserve power • Grid synchronization • Aggregated

Fig. 5 Grid structure

The power network infrastructure can be categorized into levels such as picogrid, microgrid, and macrogrid, which are interconnected. The power system can be classified in ascending order in terms of size and scale to picogrid, microgrid, and macrogrid. Figure 5 represents the scale and operations of a VPP through picogrid, microgrid, and macrogrid. A VPP can integrate resources from a picogrid to a macrogrid level. Thanks to the recent advances in edge, cloud, and fog computing (Aldegheishem et al. 2020), these resources can be coordinated to operate seamlessly. On a residential scale, picogrid-level operations can be identified, such as data mining and load profiling along with participation in a VPP through load flexibility. The microgrid level is essentially the aggregation of picogrids at the level of residential and commercial buildings. Microgrids present modularity, aggregated flexibility, resiliency, and accessibility to clean resources. Subsequently, the macrogrid level combines the former, picogrid and microgrid, and presents noncritical reserve power, grid synchronization, and aggregated power. This classification aids in planning solutions that fit the requirements of scale and size. For instance, the objectives on a building at the picogrid level are different from those of the macrogrid. Moreover, for a VPP, the scale and size of operation would determine optimal portfolio and operational regime in a decentralized and distributed setting. VPPs can be classified into three categories based on the size, scale, and operational priorities.

Virtual Power Plants and Integrated Energy System: Current Status and. . .

129

VPP

Baery bank Flywheel

Load

Mul-energy mix

Storage

Providing with balance power

Generaon

Operaons based on price signals

Balance power

Trade expected power producon and consumpon

Intra day

Day ahead

Coordinated and opmal operaon Electric Heat Flexible load

Fig. 6 VPP market and scope for optimal portfolio

A VPP enables flexible integration of geographically dispersed and different resources to achieve low emission and low-cost power. A VPP operates in an electricity market driven by price signals resulting from policy measures. For instance, CO2 prices would be consequential to the share of renewable energy-based generation units in the VPP portfolio. Figure 6 presents the operational mechanism of a VPP. A VPP through coordinated planning determines optimal operational regimes in the day ahead, intraday, and balance market. The figure also includes the variables of generation, storage, and load/flexibility that a VPP would optimize. Of course, the VPP may also participate in the reserve market, reactive power support market, and others. The following are among the key performance indicators and value propositions for a VPP with IES: • VPP overcomes uncertainties inherent in non-dispatchable resources such as wind and solar energy resources through collaboration and coordination of operation (Lin et al. 2020). • VPP overcomes risks by avoiding up-front investments and operational failures. • VPP improves flexibility from both demand (Nguyen et al. 2018) and supply sides. • VPP overcomes geographical barriers as in virtual participation. • VPP portfolio leads to cost reduction while maintaining supply-demand balance and impacting production volumes. • VPP portfolio comprises a wide range of technologies in the portfolio, ranging from a picogrid (prosumer) to microgrid level. • VPP integrates participation of the demand side as the ancillary service and flexibility provider in power grid operation. • VPP enables optimal utilization of energy produced from renewable.

130

S. Mishra et al.

• VPP enables accessibility ranging from a very small-scale producer (prosumer) to a midrange wind park to participate. • VPP facilitates emission reduction through participation and awareness. • VPP provides scope for new and innovative, customer-centering, and participatory value stream generation. In addition, it incentives participation of consumers and small producers. An example is represented by coordinated planning of optimal flexibility (Aduda et al. 2017; Mishra et al. 2019). The VPP technology is relatively new and thereby has its own challenges, namely, technology, policy, and market-oriented. Technology challenges include asset management, risk aggregation, software architecture, and privacy-related issues from the energy usage. In addition, there are policy, regulation, law, and framework challenges for defining the scope. Market challenges include a sustainable financial business model development and electricity market reforms concerning VPPs. These challenges are still open questions in relation to the fastevolving power and energy space, in particular within emerging VPPs. Owing to geographical and consumer energy behavior, the regulations would also differ. There, different requirements and potentials exists in different geographical areas, for instance, high potential of PVs in southern EU and high concentration of wind in northern EU or high heating demand in the north and high cooling demand in the south. However, a VPP can enable better market coupling if these issues are solved where consumers also play a role through coordinated decision-making for a decentralized and distributed power grid. A VPP also strengthens the grid resiliency of the decentralized and distributed power network through: • Distributed asset monitoring considering the condition of assets at demand, generation, transmission, energy storage, and measurement • Complementarity control to flatten the overall demand curve and stable power generation capacity • Contingency-aware power flow and grid in-feed • Accessibility for small, large, and EV generation units • Integration of heating and power requirements of a network

6

Sector Coupling Options for VPPs

Sector coupling refers to the concept of a purposeful connection and interaction of energy sectors to increase the flexibility of supply, demand, and storing (Fridgen et al. 2020). Coupling the heating and the power sector can play a key role in a decarbonized power system (Jimenez-Navarro et al. 2020). The sector coupling concept is particularly relevant within the C-VPP definition introduced in Sect. 2. Indeed, a C-VPP is described by a portfolio of assets that are owned by different parties. In this setting, a C-VPP allows peer-to-peer trading within the VPP, through a local market formation. This means heating assets can be part of the C-VPP, and therefore sector coupling options arise. Power-to-X technologies represent a way

Virtual Power Plants and Integrated Energy System: Current Status and. . .

131

to achieve sector coupling. Within the physical space identified in Fig. 3, the sector coupling options are embedded within the power-to-X module, as the interface that connects the renewable and conventional sources, with the residential, commercial, industrial, and transport sectors. The coupling between the heating and the power sector is identified by two main processes within the so-called power-to-X methods: • Power to heat refers to the conversion of electrical energy into heat. This is allowed, for instance, through the use of heat pumps. • Power to gas refers to the use of surplus wind generation to produce hydrogen through electrolysis. In addition, the broader definition of sector coupling also includes the possibility of coupling the energy and the transportation sector (Robinius et al. 2017). In particular, the so-called power to mobility refers to the use of batteries of electric cars as a buffer. This is also known as vehicle-to-grid technology, through which energy is pushed back to the power grid from the battery of electric cars. For the specific purpose of a VPP, we are more interested in the coupling between the heating and the power sectors, which has the highest potential within the frame of C-VPP described earlier in this chapter. When including heating concepts within the C-VPP frame, two main technologies show promising potential: low-temperature district heating and organic Rankine cycles. District heating systems provide the heat generated in a centralized location to a set of users for their residential and commercial heating requirements such as space heating and water heating. The heat is often obtained from a cogeneration plant burning fossil fuels or biomass. Heat distribution is generally obtained by using hot water or steam flowing through a closed network of insulated pipes and heat exchange stations at the users’ locations (Bordin et al. 2016). District heating can be linked to electricity systems through cogeneration of electricity and heat and through power-to-heat production in large-scale heat pumps. Low-temperature district heating refers to district heating where the network supply temperature is reduced down to approximately 50 degrees or even less (ultralow-temperature district heating). Low-temperature district heating offers new possibilities for greater energy efficiency and utilization of renewable energy sources, which lead to reduced consumption of fossil fuel-based energy (Ancona et al. 2019). The integration of low-temperature district heating within C-VPP solutions can therefore enhance the renewable potential of a VPP. From this perspective, low-temperature district heating represents one key option to move from the conventional concept of VPP toward a more holistic concept of VPP with an integrated energy system, like the C-VPP outlined earlier in this chapter. An organic Rankine cycle has been recognized as a promising technology for conversion of heat into electricity since it can be designed for operation at low temperatures with suitably selected working fluids (Freeman et al. 2017). An organic

132

S. Mishra et al.

Rankine cycle uses an organic, high molecular mass fluid with a liquid-vapor phase change, occurring at a lower temperature than the water-steam phase change. The fluid allows Rankine cycle heat recovery from lower-temperature sources (i.e., biomass combustion, industrial waste heat, geothermal heat, solar ponds, etc.). The low-temperature heat is converted into useful work, which can itself be converted into electricity. Various sources can provide the heat for the organic Rankine cycle, from solar radiation to biomass combustion, geothermal heat, or waste heat from industry. Hence, a small-scale organic Rankine cycle can be an appropriate option for sunny remote areas (Lizana et al. 2020) where, combined with photovoltaic panels, it can be a suitable alternative for electricity generation. Moreover, a tailored model and fine management can provide both electricity and thermal energy for the local inhabitants (Pereira et al. 2018). The organic Rankine cycle also plays an important role in waste heat valorization in the process industry since the residual heat can be converted to electricity. It has been proven that coupling waste heat recovery with a district heating network can provide flexibility to the electricity generation. This flexibility can be utilized by a VPP, to compensate for the variable output of renewable energy sources (Zwaenepoel et al. 2013). It is therefore possible to implement strategies to balance variable renewable production with industrial waste heat, instead of compensating the power fluctuations only by traditional power plants such as gas and coal. In summary, both low-temperature district heating and the organic Rankine cycle can actively and successfully contribute to renewable VPP solutions, by expanding the concept of VPPs toward the broader concept of VPP with integrated energy systems.

7

Storage Options for VPPs

Electricity storage has been a pivotal point in the power system – specifically to harness the full potential of the renewable energy resources. The key energy storage technologies in practice in the smart energy system are explored in Ajanovic et al. (2020). Some of the technologies are placed on transmission level, and others are in the distribution system level owing to the capacity and size of initial investment. In Table 1, a classification of electrical energy storage technologies is provided. A VPP enables the participation of a single battery unit at distribution level alongside large-scale battery banks. EVs present a spatially dynamic energy storage solutions that may also participate in the VPP. In Argade et al. (2018), charging of EVs as a VPP is explored. These technologies through VPP can provide services for peak demand shifting and voltage and frequency control. A VPP, being decentralized and distributed, presents a better value for the utilization of the total energy potential in the system. VPP can respond quickly to the variation in renewable energy resources such as utilizing the peak energy generation and scope for small storage units of EVs, to participate where there is less or no RES-based energy available.

Virtual Power Plants and Integrated Energy System: Current Status and. . .

133

Table 1 Energy storage technologies (Wong et al. 2011; Inage 2009) Storage Maximum technologies power (MW/h) Pumped 3000 hydro Compressed 1000 air Li-ion 100

Lead-acid

Drain time

4 to 16 h

Life span Energy (Yrs/cycles) density (Wh/lt) 30 to 60 0.2–2

70–85

Power network level Transmission

2 to 30 h

20 to 40

6 – Feb

40–70

Transmission

1 min to 8 h

2 to 3

200–400

85–95

100

1 min to 8 h

3 to 5

50–80

80–90

Flow-based

100

hours

20–70

60–85

Hydrogen

100

Flywheel

20

Mins to week Seconds to minutes

12,000 to 14,000 cycles 5 to 30

Transmission and distribution Transmission and distribution Transmission and distribution Transmission

20 to 30

Efficiency (%)

600 at 200 25–45 bar 20 to 80 70–95

Transmission

A sizable proportion of energy consumption is due to heating and cooling demand. Thermal energy storage therefore can play a significant role in providing system-wide flexibility through low-cost storage of heat. In Kiviluoma et al. (2017), authors indicate that 45% of the total energy usage in EU is attributed to heating. In Wong and Pinard (2016), the authors perform a case study to demonstrate how thermal energy storage coupled with variable renewable energy capacity can attain goals to reduce the environmental impacts while better utilizing the energy potential in RES. Converting electricity to heat generates losses due to the Carnot cycle. However, the energy contents in the fuel can be converted to useful heat more efficiently. This high conversion efficiency can facilitate RES fluctuations and unlock flexibility. Surplus production during off-peak hours can be converted to heating fuels. Then the value of surplus energy rises to the value of fuel. The conversion efficiency in, for example, a heat pump is higher than that of the direct resistance heaters based on electricity (Chua et al. 2010). The heating demands can be broadly classified into space heating and water heating. A district heating system is primarily used for heating demand due to the low cost. In Mateu-Royo et al. (2020), it is demonstrated how high-temperature heat pump can be effectively integrated to a district heating network as both a source and a sink. Note that the heating in this context is also applicable to cooling. In the EU, the sources of heat include natural gas, coal, oil, and biomass. Combined heat and power (CHP) power plants are rising in share. A case study on the impact of electricity prices on energy flexibility for a heat pump and thermal storage in a residential building is examined by the authors in

134

S. Mishra et al.

Table 2 Thermal energy storage potential (Sarbu and Sebarchievici 2018; Hauer 2011) Thermal energy storage Sensible (hot water) Phase change material Chemical reactions

Capacity (kWh/t) 10–50

Power (MW)

Efficiency (%)

0.001–10.0

50–90

Storage period day/months

Cost (e/kWh) 0.1–10

50–150

0.001–1.0

75–90

hours/months 10–50

120–250

0.01–1.0

75–100

hours/days

8–100

Mateu-Royo et al. (2020). The authors highlight how the flexibility potential is unlocked with better utilization of the boiler tank. In Yohanis et al. (2006), the authors discuss the solar water heating potential across different regions in Europe. With growing urbanization and habitation, the space cooling requirements are rising in Europe. A cubit meter of water changing 55–95 ◦ C offers around 58 kWh of thermal energy storage for a moderate residential house in a typical winter day. However, house insulation also plays a significant role in this setting. An adequately insulated house provides load flexibility between 2 and 12 h while maintaining occupant comfort. In Denmark, for example, the large wind power generation is coupled with district heating network. Surplus power generation during off-peak hours can be stored through a combination of electric heaters and heat storage with CHP as a balancing unit. Table 2 presents the thermal energy storage system potential.

8

Scale of a VPP in an Electricity Market

The scope for a VPP ranges from distribution to transmission levels in a power system. Within the power systems, VPPs are interconnected and share a common power market. VPP integrates more RES in the distribution and transmission networks. Many RES are placed into distributed generation in the low- to medium-voltage grid. The grid networks can be classified as radial and meshed. Radial networks are often found in suburban areas, whereas the meshed networks is found in urban areas. Often these networks have a few large-scale centralized generators. Placing variable RES-based distributed generation in this network may cause issues such as an islanding effect due the variations in the power generation from wind or solar and insulation damage due to high voltage at the end of the distribution network. A VPP formed with a selection of RES can have complementary effects in mitigating the fluctuations from wind and solar. A dispatchable generator, such as a hydropower unit, will further increase the scope of the VPP. In addition, inclusion of energy storage units or demand-side flexibility can further decrease the overall fluctuations. The VPP portfolio ranges from small scale (single battery unit, residential demand-side flexibility, rooftop PV) to large scale (wind parks,

Virtual Power Plants and Integrated Energy System: Current Status and. . .

135

PV parks, and hydropower). The VPP can also be formulated in different levels from small and regional to distributed and wide. This flexibility enables the VPP to increase the scope of participation and coordination. Furthermore, since VPP is not geographically constrained, islands, rural areas, and cities can also participate as in residential, commercial, and industrial consumers. Currently there are multiple demonstration projects around the world at a distribution system level, to study the feasibility and profitability. However, the geographic scope of VPPs is directly related to the power market and actual grid control. Cross-border VPPs can play a crucial role in further integrating the power networks and energy prices through increasing market competition. There is a wide choice of markets that are accessible to a VPP such as the spot market, reserve power market, and ancillary service market. Example of VPPs are Tesla energy in Australia, Next Kraftwerke in Germany, and PREMIO in France, to name just a few as found in Table 3. In place of installing new dispatchable or non-dispatchable generation units to meet the rising demand, the system can benefit from the exploitation of flexibility. However, the scope of flexibility varies from region to region. For example, there is a high degree of system flexibility in Norway due to the hydropower plants. Furthermore, the pattern of demand-side flexibility is different in northern and southern Europe due to the climatic conditions. A VPP provides a solution to pool together flexibility from both supply and demand side. In Fig. 7, a framework listing market services and products alongside responsible parties is presented. Constraint management services aid system operators (DSO and TSO) optimizing the grid operation in a flow-based market coupling. Adequacy services are designed to increase the security of supply through managing the total capacity to offset the long-term peak and nonpeak power demand. This service is requested by TSO and balance responsible party (BRP), while the capacity service provider (CSP) is the trading party. Wholesale services, controlled by BRP, aim to reduce the cost of electricity purchase in day-ahead, intraday markets, and cost adjustment in balancing mechanisms. Balancing services include frequency regulation mechanisms. This service is requested by TSO and traded by a balance service provider (BSP). This mechanism pools together the demand-side flexibility in the system. A VPP can participate in this mechanism in coordination with the DSO and TSO for peak load management and reserve power allocation. For example, a single EV can participate in a VPP in different locations. Based on the scale of VPP, it can be unilateral on distribution level, or bilateral contracts on TSO, or cross-border level. Moreover, a VPP can expand both the demand- and supplyside flexibility to broaden the scope. The tariff is a monthly fee in addition to the absolute power consumption for the maximum hour average power output in kW during a month. VPP by utilizing the flexibility can flatten the curve and reduce the peak input from the external grid. Consequently, the cost to the TSO is reduced. Table 3 presents a list of planned and operational VPPs corresponding to their capacities. It is clear that more and more power companies and start-ups are forming VPPs.

136

S. Mishra et al.

Table 3 Planned and operational VPPs around the world Company Next Kraftwerke-Ecotricity (VPP) Kiwi Power (modelling electricity demand the key) Statkraft (Statkraft) Open Energi (Open energi – dynamic dem2.0 distributed energy platform) Flexitricity (Quinbrook invests in flexible generation) Moixa (Moixal) E.ON-Thyssenkrupp (E.onthyssenkrupp bring hydrogen to the electric market – energy live news) UK Power Reserve (UKPR) (UK’s largest independent distributed energy supplier sold for $216m) Limejump (Limejump gets all clear to chase big six in UK grid balancing market – rethink) EDF (Edf offers 866 mw at VPP auction) Next Kraftwerke (Next kraftwerke connects 2 mw battery to its VPP in Belgium – pv magazine international) Centrica’s Restore (Centrica’s restore business launches 32 mw ‘virtual power plant’ in Belgium – centrica plc) E.ON-Thyssenkrupp (E.onthyssenkrupp bring hydrogen to the electric market – energy live news) sonnenVPP (Software for virtual power plants powered by residential solar-plus-storage) Steag (A4 large scale storage options under special consideration of 6 × 15 mw battery example) AutoGrid (Navigant research names autogrid) Energy2market (e2m announces strategic partnership with swytch to collaborate on proof of concept pilot – with e2m) LichtBlick (Storagethe rise of the virtual power plant – energy storage news) Next Kraftwerke (Next kraftwerke records strong portfolio growth in 2017) WEMAG (Building a battery power plant wemag – aggreko) AGL “Bring Your Own Battery” (What is a virtual power plant?) Simple Energy SA (Simply energy virtual power plant – Australian renewable energy agency) Tesla Energy Plan (Sa virtual power plant – tesla Australia) Origin Virtual Power Plant (First 5 mw virtual power plant from major utility origin gets funding in victoria) Plico Energy VPP (W.a. community virtual power plant confirms $50m Swiss investment – one step off the grid) Ausgrid “Power2U” (Evegen virtual power plant) ENGIE-KiWi (Engie) ENEL (Enel increases ’virtual power plant’ to 157 mw – Vermont business magazine) Enbala (VPP) Sunrun (repowering-clean-sunrun) PG&E (Pge program will transform hundreds of homes into a virtual power plant – news releases – pge)

Capacity Countries (MW) 6.9 UK 300 2000 60 540 17 600 533 150 866 2

FR BE

32

BE

600

DE

2 800 5000 3500 5 1000 5 5 6

AU

250 5 6.5 1 4.1 157 350 295 200

US

Virtual Power Plants and Integrated Energy System: Current Status and. . .

Services

Constraint management

Adequacy

Wholesale

137

Balancing

Voltage control

Capacity payment

Day ahead optimization

Frequency containment

Grid capacity management

National capacity market

Intraday optimization

Automatic frequency restoration

Congestion management

Strategic reserve

Self/Passive balancing

Manual frequency restoration

Controlled islanding

Hedging

Generation optimization

Replacement reserve

Requesting party

DSO

TSO

TSO

BRP

Trading party

CMSP

CSP

BRP

TSO

BRP

BSP

Fig. 7 Scope for VPP in market and products for flexibility (Van der Veen et al. 2018)

9

VPPs of the Future

With reference to Table 3, investments are flowing in for capacity expansion along with new VPPs. Apart from the PV-based capacities, there is a surge in storage capacity or battery bank-oriented VPP along with demand-side flexibility. Reshaping the peak demand is among the primary objectives among new VPPs. This facilitates new and emerging technologies such as hydrogen for storage. Energy policies and regulations are adapting to encourage the share of RES in the total energy mix. VPPs will further catalyze the adoption and integration of RESbased distributed and decentralized sources and demand-side flexibility. Reduction of unitary energy price and climate neutrality are key motivators behind the adoption of VPP technologies. With distributed generation-based VPPs comes the challenge of resiliency and reliability. The sector coupling of power and energy sectors further increase such challenges. Risks owing to loss of load expectation and energy expected not served in a VPP are discussed in Sadeghian et al. (2020). Since a VPP is primarily a software solution with hardware switches, security challenges become prominent. In this report Johnson (2017), the authors describe the scope of VPPs to provide a range of grid support services. The cybersecurity and cyber threats associated with VPPs are discussed in Venkatachary et al. (2020). Strategies to mitigate cyber threats and communication disruptions in a VPP are discussed in Li et al. (2018). A draft framework for cybersecurity for smart grids in EU is presented in Pavleska et al. (2020).

138

S. Mishra et al.

A VPP is essentially a bottom-up framework where small- and medium-scale generation units form a centralized entity to participate in the market. However, asset management and ownership arise as a challenge as opposed to the conventional top-down power plants. In Sony (2019), a Lean sigma approach is discussed for the power sector. Lean Sigma (LS) is a method to improve the quality to transfer the value while increasing profit and maintaining a competitive advantage. An LS approach can be applied to VPPs for deriving value for small-scale generation units and consumer participation for the demand-side flexibility. Through this, it is possible to unlock an incremental and continuous growth potential for stake holders, from both the demand and the supply side. A VPP of the future would employ intelligent strategies for fast response to volatile prices and adapt to weather changes. EVs can participate in a geographically distributed manner to charge or discharge in coordination with peak demand. From this point of view, a VPP can ensure high interoperability such as API, protocols, and data exchanges in order to accommodate assets of different technologies and ensure optimal communication between the assets. Along with scalability, a VPP of the future would be implemented with sophisticated systems that allow for the rapid identification of operational strategies including preventive, corrective, and restorative actions that ensure high reliability of the technologies involved.

9.1

Toward Intelligent VPP Solutions

Intelligent and self-learning algorithms can be integrated to the proactive operational planning of the VPP. The consumer devices such as charging plugs, television, etc. are becoming smart with cognitive learning of user behavior. A VPP can learn and be responsive to this behavior, customized to each household to adapt to fluctuations from weather, prices, generation, and the likes, in other regions. Smart control strategies can be implemented in the VPP to facilitate the larger grid for self-healing, frequency, and voltage control. A VPP can enable the formation of cognitive twins (CT) (Lu et al. 2020). CT have augmented semantic capabilities for identifying the dynamics to formulate interlinks between VPPs. Beyond residential and commercial demand, industrial processes such as manufacturing can also be adapted to a VPP (Rozanec and Jinzhi 2020). Future VPPs can enable human-machine interaction through reasoning- and learning-based approach in industries (Zhai et al. 2020). The power grid control strategies are in transition from conventional centralized to distributed. The last few years have seen rapid advances in communication technologies and cognitive devices. Therefore, the control strategies need to further develop toward cognitive and responsive control while being distributed. A VPP as a platform can facilitate these transformations toward a transactive energy. Since VPPs are not limited to geography while wide in scale and level, virtual control strategies can be further incorporated to enhance the VPP. Even though all vectors of energy demand are weather-aware, these vectors are not aware of each other. If the vectors are aware of each other, the scope of flexibility can be further improved while being more coordinated. This feature can be provided

Virtual Power Plants and Integrated Energy System: Current Status and. . .

139

by a VPP. The devices we use (i.e., fridge, computers, etc.) are also becoming more energy-efficient. Energy-efficient manufacturing as a part of industry 4.0 is also on the horizon. A VPP can also facilitate such projects through trading values in terms of how the price is set for a device. Beyond the price and quantity relationship, in economy the quality and resiliency of curves can be further integrated by a VPP.

10

Conclusion

This chapter has explored the scope of virtual power plants in an integrated energy system. Detailed discussions on the definition of the VPP, classification of VPPs, and evolution are provided. Challenges associated with the VPP both in context of power system, cybersecurity, and asset management are discussed. The IES-oriented VPP in the context of sector coupling is also explored. The current and planned VPPs around the world with an insight to the scale of VPPs are discussed. Subsequently, the future scope of VPP considering intelligence, smart control, and lean sigma process optimization techniques is also brought into discussion. This chapter sets out in an investigative understanding of the state of the art while shedding light on the future outlook of VPP. The review and discussions brought forward in this chapter have underlined the scope, opportunity, and challenges for a virtual power plant in an integrated energy system. The proposed scheme expands the opportunity to low-cost power generation through integration of distributed energy systems with high share of renewable resources. Furthermore, the virtual power plants have been categorized into physical and cyber components. The challenges related to security and privacy are addressed. In the outlook, the policy measures are required along with regulatory reforms to increase the scope for a virtual power plant. Specifically sector coupling within the integrated energy systems among electric power system and heating system requires further investigation. Beyond this, energy can be also traded like a currency, forming an energy bank where the value of energy is traded as a currency in the bank. Acknowledgments This work is supported by the Estonian Research Council grant PUTJD915.

References A4 large scale storage options under special consideration of 6 × 15 mw battery example, http:// www.eqmagpro.com/wp-content/uploads/2017/06/VGB-Congress_Benesch.pdf. Accessed on 21 Oct 2020 K. Aduda, T. Labeodan, W. Zeiler, G. Boxem, Demand side flexibility coordination in office buildings: a framework and case study application. Sustain. Cities Soc. 29, 139–158 (2017) A. Ajanovic, A. Hiesl, R. Haas, On the role of storage for electricity in smart energy systems. Energy 200 117473 (2020) A. Alahyari, M. Ehsan, M. Moghimi, Managing distributed energy resources (DERs) through virtual power plant technology (VPP): a stochastic information-gap decision theory (IGDT) approach. Iranian J. Sci. Technol. Trans. Electric. Eng. 44(1), 279–291 (2020). https://doi.org/ 10.1007/s40998-019-00248-w

140

S. Mishra et al.

A. Aldegheishem, R. Bukhsh, N. Alrajeh, N. Javaid, FaaVPP: fog as a virtual power plant service for community energy management. Futur. Gener. Comput. Syst. 105, 675–683 (2020) A.A. Alkahtani, S.T. Alfalahi, A.A. Athamneh, A.Q. Al-Shetwi, M.B. Mansor, M. Hannan, V.G. Agelidis, Power quality in microgrids including supraharmonics: issues, standards, and mitigations. IEEE Access 8, 127104–127122 (2020) A.Q. Al-Shetwi, M.Z. Sujod, Modeling and control of grid-connected photovoltaic power plant with fault ride-through capability. J. Solar Energy Eng. 140(2), 1–8 (2018) A.Q. Al-Shetwi, M. Hannan, K.P. Jern, A.A. Alkahtani, A. PG Abas, Power quality assessment of grid-connected pv system in compliance with the recent integration requirements. Electronics 9(2), 366 (2020) M.A. Ancona, M. Bianchi, L. Branchini, A. De Pascale, F. Melino, A. Peretto, Low temperature district heating networks for complete energy needs fulfillment. Int. J. Sustain. Energy Plan. Manag. 24, 33–42 (2019) K.E. Antoniadou-Plytaria, I.N. Kouveliotis-Lysikatos, P.S. Georgilakis, N.D. Hatziargyriou, Distributed and decentralized voltage control of smart distribution networks: models, methods, and future research. IEEE Trans. Smart Grid 8(6), 2999–3008 (2017) S.G. Argade, V. Aravinthan, I.E. Büyüktahtakın, S. Joseph, Performance and consumer satisfaction-based bi-level tariff scheme for EV charging as a VPP. IET Gener. Transm. Distrib. 13(11), 2112–2122 (2018) P. Asmus, Microgrids, virtual power plants and our distributed energy future. Electric. J. 23(10), 72–82 (2010) S. Babaei, C. Zhao, L. Fan, A data-driven model of virtual power plants in day-ahead unit commitment. IEEE Trans. Power Syst. 34(6), 5125–5135 (2019) L. Bai, F. Li, H. Cui, T. Jiang, H. Sun, J. Zhu, Interval optimization based operating strategy for gas-electricity integrated energy systems considering demand response and wind uncertainty. Appl. Energy 167, 270–279 (2016). https://doi.org/10.1016/j.apenergy.2015.10.119 R. Boampong, D.P. Brown, On the benefits of behind-the-meter rooftop solar and energy storage: the importance of retail rate design. Energy Econ. 86, 104682 (2020) C. Bordin, A. Gordini, D. Vigo, An optimization approach for district heating strategic network design. Eur. J. Oper. Res. 252(1), 296–307 (2016) C. Bordin, A. Håkansson, S. Mishra, Smart energy and power systems modelling: an IOT and cyber-physical systems perspective, in the context of energy informatics. Proc. Comput. Sci. 176, 2254–2263 (2020) T. Brown, D. Schlachtberger, A. Kies, S. Schramm, M. Greiner, Synergies of sector coupling and transmission reinforcement in a cost-optimised, highly renewable European energy system. Energy 160, 720–739 (2018). arXiv:1801.05290, https://doi.org/10.1016/j.energy.2018.06.222 Building a battery power plant wemag – aggreko, https://www.aggreko.com/en-pg/case-studies/ utilities/building-a-battery-power-plant-wemag#challenge. Accessed on 21 Oct 2020 Centrica’s restore business launches 32 mw ‘virtual power plant’ in Belgium – centrica plc, https://www.centrica.com/news/centricas-restore-business-launches-32-mw-virtualpower-plant-belgium. Accessed on 21 Oct 2020 P. Chaudhari, P. Rane, A. Bawankar, P. Shete, K. Kalange, A. Moghe, J. Panda, A. Kadrolkar, K. Gaikwad, N. Bhor et al., Design and implementation of statcom for reactive power compensation and voltage fluctuation mitigation in microgrid, in 2015 IEEE International Conference on Signal Processing, Informatics, Communication and Energy Systems (SPICES) (IEEE, 2015), pp. 1–5 S.T. Chavhan, C. Bhattar, P.V. Koli, V.S. Rathod, Application of statcom for power quality improvement of grid integrated wind mill, in 2015 IEEE 9th International Conference on Intelligent Systems and Control (ISCO) (IEEE, 2015), pp. 1–7 C.-I. Chen, Y.-C. Chen, C.-N. Chen, A high-resolution technique for flicker measurement in power quality monitoring, in 2013 IEEE 8th Conference on Industrial Electronics and Applications (ICIEA) (IEEE, 2013), pp. 528–533 K.J. Chua, S.K. Chou, W. Yang, Advances in heat pump systems: a review. Appl. Energy 87(12), 3611–3624 (2010)

Virtual Power Plants and Integrated Energy System: Current Status and. . .

141

e2m announces strategic partnership with swytch to collaborate on proof of concept pilot – with e2m, https://www.e2m.energy/en/news-entry/Strategic-Partnership-with-Swytch.html. Accessed on 21 Oct 2020 Edf offers 866 mw at VPP auction, no additional power tenders on horizon – icis, https://www. icis.com/explore/resources/news/2007/11/29/9300807/edf-offers-866-mw-at-vpp-auction-noadditional-power-tenders-on-horizon/. Accessed on 21 Oct 2020 N. Edomah, Effects of voltage sags, swell and other disturbances on electrical equipment and their economic implications, in IEEE Proceedings of 20th International Conference on Electricity Distribution (IET, 2009), pp. 1–4 K. El Bakari, W.L. Kling, Virtual power plants: an answer to increasing distributed generation, in IEEE PES Innovative Smart Grid Technologies Conference Europe, ISGT Europe (2010). https://doi.org/10.1109/ISGTEUROPE.2010.5638984 M. Emarati, F. Keynia, M. Rashidinejad, A two-stage stochastic programming framework for riskbased day-ahead operation of a virtual power plant. Int. Trans. Electric. Energy Syst. 30(3) (2020). https://doi.org/10.1002/2050-7038.12255 Enel increases ‘virtual power plant’ to 157 mw – Vermont business magazine, https://vermontbiz. com/news/2018/march/01/enel-increases-virtual-power-plant-157mw. Accessed on 21 Oct 2020 Engie, kiwi power partner on virtual power plants in US – s&p global market intelligence, https://www.spglobal.com/marketintelligence/en/news-insights/latest-news-headlines/ engie-kiwi-power-partner-on-virtual-power-plants-in-us-59480758. Accessed on 21 Oct 2020 E.on and thyssenkrupp bring hydrogen to the electric market – energy live news, https://www.en ergylivenews.com/2020/06/30/e-on-and-thyssenkrupp-bring-hydrogen-to-the-electric-market/. Accessed on 21 Oct 2020 Evegen virtual power plant, https://www.evergen.com.au/. Accessed on 21 Oct 2020 First 5 mw virtual power plant from major utility origin gets funding in victoria, Australia, https:// www.energy-storage.news. Accessed on 21 Oct 2020 J. Freeman, I. Guarracino, S.A. Kalogirou, C.N. Markides, A small-scale solar organic Rankine cycle combined heat and power system with integrated thermal energy storage. Appl. Thermal Eng. 127, 1543–1554 (2017) G. Fridgen, R. Keller, M.-F. Körner, M. Schöpf, A holistic view on sector coupling. Energy Policy 147, 111913 (2020) I. Gerami Moghaddam, Scheduling a smart energy hub-based virtual power plant using benders decomposition to considering power system constraints. Int. Trans. Electric. Energy Syst. 28(10), e2608 (2018) A. Ghosh, G. Ledwich, A unified power quality conditioner (UPQC) for simultaneous voltage and current compensation. Electric Power Syst. Res. 59(1), 55–63 (2001) C. Goebel, H.-A. Jacobsen, V. del Razo, C. Doblander, J. Rivera, J. Ilg, C. Flath, H. Schmeck, C. Weinhardt, D. Pathmaperuma et al., Energy Inf. Bus. Inf. Syst. Eng. 6(1), 25–31 (2014) D. Graovac, V. Katic, A. Rufer, Power quality problems compensation with universal power quality conditioning system. IEEE Trans. Power Delivery 22(2), 968–976 (2007) A. Hariri, M.O. Faruque, Impacts of distributed generation on power quality, in 2014 North American Power Symposium (NAPS) (IEEE, 2014), pp. 1–6 A. Hariri, M.O. Faruque, R. Soman, R. Meeker, Impacts and interactions of voltage regulators on distribution networks with high pv penetration, in 2015 North American Power Symposium (NAPS) (IEEE, 2015), pp. 1–6 A. Hauer, Storage technology issues and opportunities, international low-carbon energy technology platform, in, Proceedings of the Strategic and Cross-Cutting Workshop “Energy Storage— Issues and Opportunities”, Paris, vol. 15 (2011) R. Hemmati, H. Saboori, S. Saboori, Stochastic risk-averse coordinated scheduling of grid integrated energy storage units in transmission constrained wind-thermal systems within a conditional value-at-risk framework. Energy 113, 762–775 (2016). https://doi.org/10.1016/ j.energy.2016.07.089

142

S. Mishra et al.

A. Honrubia-Escribano, T. García-Sánchez, E. Gómez-Lázaro, E. Muljadi, A. Molina-Garcia, Power quality surveys of photovoltaic power plants: characterisation and analysis of grid-code requirements. IET Renew. Power Gener. 9(5), 466–473 (2015) E. Hossain, M.R. Tür, S. Padmanaban, S. Ay, I. Khan, Analysis and mitigation of power quality issues in distributed generation systems using custom power devices. IEEE Access 6, 16816– 16833 (2018) K. Ilango, A. Bhargav, A. Trivikram, P. Kavya, G. Mounika, M.G. Nair, Power quality improvement using statcom with renewable energy sources, in 2012 IEEE 5th India International Conference on Power Electronics (IICPE) (IEEE, 2012), pp. 1–6 S.-I. Inage, Prospects for large-scale energy storage in decarbonised power grids. Int. Energy Agency 3(4), 125 (2009) M. Jadidbonab, B. Mohammadi-Ivatloo, M. Marzband, P. Siano, Short-term self-scheduling of virtual energy hub plant within thermal energy market. IEEE Trans. Ind. Electron. 1 (2020). https://doi.org/10.1109/tie.2020.2978707 J.-P. Jimenez-Navarro, K. Kavvadias, F. Filippidou, M. Paviˇcevi´c, S. Quoilin, Coupling the heating and power sectors: the role of centralised combined heat and power plants and district heat in a European decarbonised power system. Appl. Energy 270, 115134 (2020) J.T. Johnson, Design and evaluation of a secure virtual power plant. Technical report, Sandia National Lab.(SNL-NM), Albuquerque, NM (United States) (2017) M. Kesler, E. Ozdemir, Synchronous-reference-frame-based control method for upqc under unbalanced and distorted load conditions. IEEE Trans. Ind. Electron. 58(9), 3967–3975 (2010) V. Khadkikar, Enhancing electric power quality using UPQC: a comprehensive overview. IEEE Trans. Power Electron. 27(5), 2284–2297 (2011) V. Khadkikar, A. Chandra, A. Barry, T. Nguyen, Analysis of power flow in UPQC during voltage sag and swell conditions for selection of device ratings, in 2006 Canadian Conference on Electrical and Computer Engineering (IEEE, 2006), pp. 867–872 J. Kiviluoma, S. Heinen, H. Qazi, H. Madsen, G. Strbac, C. Kang, N. Zhang, D. Patteeuw, T. Naegler, Harnessing flexibility from hot and cold: heat storage and hybrid systems can play a major role. IEEE Power Energy Mag. 15(1), 25–33 (2017) A. Kulmala, S. Repo, P. Järventausta, Coordinated voltage control in distribution networks including several distributed energy resources. IEEE Trans. Smart Grid 5(4), 2010–2020 (2014) T.-L. Lee, S.-H. Hu, Y.-H. Chan, D-statcom with positive-sequence admittance and negativesequence conductance to mitigate voltage fluctuations in high-level penetration of distributedgeneration systems. IEEE Trans. Ind. Electron. 60(4), 1417–1428 (2011) S. Lehnhoff, A. Nieße, Recent trends in energy informatics research. IT-Inf. Technol. 59(1), 1–3 (2017) P. Li, Y. Liu, H. Xin, X. Jiang, A robust distributed economic dispatch strategy of virtual power plant under cyber-attacks. IEEE Trans. Ind. Inform. 14(10), 4343–4352 (2018) Limejump gets all clear to chase big six in UK grid balancing market – rethink, https://rethinkresearch.biz/articles/limejump-gets-all-clear-to-chase-big-six-in-uk-grid-balancing-market/. Accessed on 21 Oct 2020 L. Lin, X. Guan, Y. Peng, N. Wang, S. Maharjan, T. Ohtsuki, Deep reinforcement learning for economic dispatch of virtual power plant in internet of energy. IEEE Internet Things J. 7, 6288– 6301 (2020) X. Liu, A. Aichhorn, L. Liu, H. Li, Coordinated control of distributed energy storage system with tap changer transformers for voltage rise mitigation under high photovoltaic penetration. IEEE Trans. Smart Grid 3(2), 897–906 (2012) J. Lizana, C. Bordin, T. Rajabloo, Integration of solar latent heat storage towards optimal smallscale combined heat and power generation by organic Rankine cycle. J. Energy Storage 29, 101367 (2020) J. Lu, X. Zheng, A. Gharaei, K. Kalaboukas, D. Kiritsis, Cognitive twins for supporting decisionmakings of internet of things systems, in Proceedings of 5th International Conference on the Industry 4.0 Model for Advanced Manufacturing (Springer, 2020), pp. 105–115

Virtual Power Plants and Integrated Energy System: Current Status and. . .

143

C. Mateu-Royo, S. Sawalha, A. Mota-Babiloni, J. Navarro-Esbrí, High temperature heat pump integration into district heating network. Energy Convers. Manag. 210, 112719 (2020) S. Mishra, C. Bordin, A. Tomasgard, I. Palu, A multi-agent system approach for optimal microgrid expansion planning under uncertainty. Int. J. Electric. Power Energy Syst. 109, 696–709 (2019) ‘modelling electricity demand the key’ as UK comes out of lockdown: Kiwi power – s&p global platts, https://www.spglobal.com. Accessed on 21 Oct 2020 Moixa to build virtual power plant as first phase of UK smart energy project – pv magazine international, https://www.moixa.com/. Accessed on 21 Oct 2020 M. Molinas, J.A. Suul, T. Undeland, Low voltage ride through of wind farms with cage generators: statcom versus SVC. IEEE Trans. Power Electron. 23(3), 1104–1117 (2008) J.M. Morales, A.J. Conejo, H. Madsen, P. Pinson, M. Zugno, Integrating Renewables in Electricity Markets: Operational Problems, vol. 205 (Springer Science & Business Media, New York, 2013) Navigant research names autogrid as #1 virtual power plant platform provider in 2020 – autogrid, https://www.auto-grid.com/products/virtual-power-plant/. Accessed on 21 Oct 2020 Next kraftwerke connects 2 mw battery to its VPP in Belgium – pv magazine international, https://www.pv-magazine.com/2018/07/13/next-kraftwerke-connects-2-mw-battery-toits-vpp-in-belgium/. Accessed on 21 Oct 2020 Next kraftwerke records strong portfolio growth in 2017, https://www.next-kraftwerke.com/news/ next-kraftwerke-records-strong-portfolio-growth-in-2017. Accessed on 21 Oct 2020 H.T. Nguyen, L.B. Le, Z. Wang, A bidding strategy for virtual power plants with the intraday demand response exchange market using the stochastic programming. IEEE Trans. Ind. Appl. 54(4), 3044–3055 (2018) Open energi – dynamic demand 2.0 distributed energy platform, https://www.openenergi.com/. Accessed on 21 Oct 2020 G. Papaefthymiou, K. Dragoon, Towards 100% renewable energy systems: uncapping power system flexibility. Energy Policy 92, 69–82 (2016). https://doi.org/10.1016/j.enpol.2016. 01.025 T. Pavleska, H. Aranha, M. Masi, G.P. Sellitto, Drafting a cybersecurity framework profile for smart grids in EU: A goal-based methodology, in European Dependable Computing Conference (Springer, 2020), pp. 143–155 J.S. Pereira, J.B. Ribeiro, R. Mendes, G.C. Vaz, J.C. André, Orc based micro-cogeneration systems for residential application–a state of the art review and current challenges. Renew. Sustain. Energy Rev. 92, 728–743 (2018) Pge program will transform hundreds of homes into a virtual power plant – news releases – pge, https://www.portlandgeneral.com. Accessed on 21 Oct 2020 G. Plancke, K. De Vos, R. Belmans, A. Delnooz, Virtual power plants: definition, applications and barriers to the implementation in the distribution system, in International Conference on the European Energy Market, EEM, Aug 2015 (IEEE Computer Society, 2015). https://doi.org/ 10.1109/EEM.2015.7216693 D. Pudjianto, C. Ramsay, G. Strbac, Virtual power plant and system integration of distributed energy resources. IET Renew. Power Gener. 1(1), 10–16 (2007). https://doi.org/10.1049/ietrpg:20060023 Quinbrook invests in flexible generation, grid support and demand response, https://www. quinbrook.com. Accessed on 21 Oct 2020 D. Ranamuka, A. Agalgaonkar, K. Muttaqi, Online voltage control in distribution systems with multiple voltage regulating devices. IEEE Trans. Sustain. Energy 5(2), 617–628 (2013) D. Ranamuka, A.P. Agalgaonkar, K.M. Muttaqi, Examining the interactions between dg units and voltage regulating devices for effective voltage control in distribution systems. IEEE Trans. Ind. Appl. 53(2), 1485–1496 (2016) D. Ranamuka, A.P. Agalgaonkar, K.M. Muttaqi, Examining the interactions between dg units and voltage regulating devices for effective voltage control in distribution systems. IEEE Trans. Ind. Appl. 53(2), 1485–1496 (2017)

144

S. Mishra et al.

T.B. Rasmussen, G. Yang, A.H. Nielsen, Z. Dong, A review of cyber-physical energy system security assessment, in 2017 IEEE Manchester PowerTech (IEEE, 2017), pp. 1–6 P.K. Ray, S.R. Mohanty, N. Kishor, Classification of power quality disturbances due to environmental characteristics in distributed generation system. IEEE Trans. Sustain. Energy 4(2), 302–313 (2012) P.K. Ray, S.R. Mohanty, N. Kishor, J.P. Catalão, Optimal feature and decision tree-based classification of power quality disturbances in distributed generation systems. IEEE Trans. Sustain. Energy 5(1), 200–208 (2013) repowering-clean-sunrun.pdf, https://www.sunrun.com/sites/default/files/repowering-cleansunrun.pdf. Accessed on 21 Oct 2020 M. Robinius, A. Otto, P. Heuser, L. Welder, K. Syranidis, D.S. Ryberg, T. Grube, P. Markewitz, R. Peters, D. Stolten, Linking the power and transport sectors – part 1: The principle of sector coupling. Energies 10(7), 956 (2017) J.M. Rozanec, L. Jinzhi, Towards actionable cognitive digital twins for manufacturing, in International Workshop On Semantic Digital Twins (SeDiT 2020), Greece (2020) Sa virtual power plant – tesla Australia, https://www.tesla.com/en_au/sa-virtual-power-plant. Accessed on 21 Oct 2020 O. Sadeghian, A. Oshnoei, R. Khezri, S. Muyeen, Risk-constrained stochastic optimal allocation of energy storage system in virtual power plants. J. Energy Storage 31, 101732 (2020) R. Saint, N. Friedman, The application guide for distributed generation interconnection-the nreca guide to IEEE 1547, in 2002 Rural Electric Power Conference. Papers Presented at the 46th Annual Conference (Cat. No. 02CH37360) (IEEE, 2002), pp. D2–1 I. Sarbu, C. Sebarchievici, A comprehensive review of thermal energy storage. Sustainability 10(1), 191 (2018) Simply energy virtual power plant – Australian renewable energy agency, https://arena.gov.au/ projects/simply-energy-virtual-power-plant-vpp/. Accessed on 21 Oct 2020 Software for virtual power plants powered by residential solar-plus-storage, https://www.pvmagazine.com/. Accessed on 21 Oct 2020 M. Sony, Lean six sigma in the power sector: frog into prince. Benchmarking: Int. J. 26, 356– 370 (2019) Springer, Aims and scope of energy informatics journal, https://energyinformatics.springeropen. com, online; Accessed 2020 Statkraft unveils 1gw virtual power plant in UK, intends to double capacity by summer – current news, https://www.statkraft.com/newsroom/news-and-stories/archive/2020/vppbalancing-services/. Accessed on 21 Oct 2020 Storage and the rise of the virtual power plant – energy storage news, https://www.energy-storage. news/blogs/storage-and-the-rise-of-the-virtual-power-plant. Accessed on 21 Oct 2020 R. Thallam, G. Heydt, Power acceptability and voltage sag indices in the three phase sense, in 2000 Power Engineering Society Summer Meeting (Cat. No. 00CH37134), vol. 2 (IEEE, 2000), pp. 905–910 UK’s largest independent distributed energy supplier sold for $216m, https://www.eenewspower. com/news/uks-largest-independent-distributed-energy-supplier-sold-ps216m. Accessed on 21 Oct 2020 A. Van der Veen, M. Van der Laan, H. De Heer, E. Klaassen, W. Van den Reek, Flexibility value chain (2018) S.K. Venkatachary, J. Prasad, R. Samikannu, A. Alagappan, L.J.B. Andrews, Cybersecurity infrastructure challenges in IoT based virtual power plants. J. Stat. Manag. Syst. 23(2), 263– 276 (2020) F.A. Viawan, D. Karlsson, Combined local and remote voltage and reactive power control in the presence of induction machine distributed generation. IEEE Trans. Power Syst. 22(4), 2003– 2012 (2007) VPP, Why Ecotricity has entered the flexibility market – theenergyst.com, https://theenergyst.com/ ecotricity-enters-flexibility-market/. Accessed on 21 Oct 2020

Virtual Power Plants and Integrated Energy System: Current Status and. . .

145

VPP.pdf, https://www.enbala.com/wp-content/uploads/2018/01/VPP.pdf. Accessed on 21 Oct 2020 S. Vukmirovi´c, A. Erdeljan, F. Kuli´c, S. Lukovi´c, Software architecture for smart metering systems with virtual power plant, in MELECON 2010-2010 15th IEEE Mediterranean Electrotechnical Conference (IEEE, 2010), pp. 448–451 W.a. community virtual power plant confirms $50m Swiss investment – one step off the grid, https://onestepoffthegrid.com.au/w-a-community-virtual-power-plant-confirms-50mswiss-investment/. Accessed on 21 Oct 2020 R. Walling, R. Saint, R.C. Dugan, J. Burke, L.A. Kojovic, Summary of distributed resources impact on power delivery systems. IEEE Trans. Power Delivery 23(3), 1636–1644 (2008) What is a virtual power plant? https://www.solarpowerworldonline.com/2017/09/virtual-powerplant/. Accessed on 21 Oct 2020 M. Wickert, S. Liebehentze, A. Zündorf, Experience report: first steps towards a microservice architecture for virtual power plants in the energy sector, in Joint Post-proceedings of the First and Second International Conference on Microservices (Microservices 2017/2019), Schloss Dagstuhl-Leibniz-Zentrum für Informatik (2020) S. Wong, J.-P. Pinard, Opportunities for smart electric thermal storage on electric grids with renewable energy. IEEE Trans. Smart Grid 8(2), 1014–1022 (2016) Y. Wong, L. Lai, S. Gao, K. Chau, Stationary and mobile battery energy storage systems for smart grids, in 2011 4th International Conference on Electric Utility Deregulation and Restructuring and Power Technologies (DRPT) (IEEE, 2011), pp. 1–6 Q.-H. Wu, J. Zheng, Z. Jing, X. Zhou, Large-Scale Integrated Energy Systems. Energy Systems in Electrical Engineering (Springer, Singapore, 2019). https://doi.org/10.1007/978-981-13-6943-8 Y. Yohanis, O. Popel, S. Frid, B. Norton, Geographic variation of solar water heater performance in Europe. Proc. Inst. Mech. Eng. Part A: J. Power Energy 220(4), 395–407 (2006) V. Yuvaraj, S. Deepa, A.R. Rozario, M. Kumar, Improving grid power quality with facts device on integration of wind energy system, in 2011 Fifth Asia Modelling Symposium (IEEE, 2011), pp. 157–162 Z. Zhai, J.F. Martínez, N.L. Martínez, V.H. Díaz, Applying case-based reasoning and a learningbased adaptation strategy to irrigation scheduling in grape farming. Comput. Electron. Agric. 178, 105741 (2020) X. Zhang, M. Shahidehpour, A. Alabdulwahab, A. Abusorrah, Optimal expansion planning of energy hub with multiple energy infrastructures. IEEE Trans. Smart Grid 6(5), 2302–2311 (2015). https://doi.org/10.1109/TSG.2015.2390640 G. Zhang, C. Jiang, X. Wang, Comprehensive review on structure and operation of virtual power plant in electrical system (2019). https://doi.org/10.1049/iet-gtd.2018.5880 Y. Zhao, Electrical power systems quality. Univ. Buffalo (2016) B. Zwaenepoel, J. Vansteenbrugge, T. Vandoorn, G. Van Eetvelde, L. Vandevelde, Renewable energy balancing with thermal grid support, in 16th International conference on Process Integration, Modelling and Optimisation for Energy Saving and Pollution Reduction (PRES 2013), vol. 35 (AIDIC Servizi, 2013), pp. 535–540

Reliability Analysis of Smart Grids Using Formal Methods Mohamed Abdelghany and Sofiène Tahar

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Reliability Modeling Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Reliability Analysis Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Proposed Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 State of the Art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Event Tree Reliability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Cause-Consequence Diagram Reliability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Functional Block Diagram Reliability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Monte Carlo Simulation Reliability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Reliability Analysis in HOL4 Theorem Proving . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Formal Reliability Analysis of Smart Power Grids . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Proposed Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Smart Grid Power System Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

148 150 151 151 151 152 152 153 154 154 155 156 157 158 161 161

Abstract

Smart grids (SG) are complex integrated electric networks, where failures in any zone of the network can cause widespread catastrophic disruption of supply. In recent years, there has been a significant proliferation in the use of renewable energy sources, such as wind/solar systems, for SG power generation due to global warming, pollution, as well as economic and energy security concerns. However, the main obstacle that these energy systems face is their intermittent

M. Abdelghany () · S. Tahar Department of Electrical and Computer Engineering, Concordia University, Montreal, QC, Canada e-mail: [email protected]; [email protected]; [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_81

147

148

M. Abdelghany and S. Tahar

nature, which greatly affects their ability to deliver constant power to the grid. While this raises several reliability-related concerns, existing sampling-based simulation tools, such as the Monte Carlo approach, cannot guarantee absolute accuracy of the reliability analysis results due to their inherent incompleteness. Therefore, in this chapter, we propose a novel approach that uses formal methods for the accurate and sound reliability analysis of SG systems. This new methodology overcomes the incompleteness of simulation-based analysis and the error-proneness of manual mathematical analysis. In particular, we use higherorder logic (HOL) theorem proving, which is a computer-based mathematical reasoning tool, where we developed a library of fundamental concepts of reliability analysis techniques, such as event trees, functional block diagrams, and cause-consequence diagrams. This library allowed us to conduct formal system-/subsystem-level reliability analysis and determine absolute accuracy of important SG reliability indices, such as system/customer average interruption frequency and duration (SAIFI, SAIDI, and CAIDI), as well as energy indices, such as Energy not Supplied Index (ENS) and loss of energy expectation (LOEE). In order to demonstrate the effectiveness of our proposed methods, we conducted the formal system-/subsystem-level reliability analysis of the standard IEEE 3/39/118-bus electrical power generation/transmission/distribution networks. The results of the proposed formal analysis are extremely useful for the electrical power planners/designers to accurately quantify SG reliability improvements and satisfy the total demand within acceptable risk levels. Keywords

Power system reliability · Smart grids · Renewable energy resources · Formal methods · SAIFI · SAIDI · CAIDI · ENS · LOLE · LOEE

1

Introduction

Due to the complex and integrated nature of real-world smart grids (SG) (Keyhani and Albaijat 2012), as shown in Fig. 1, failures in any part of the network can cause catastrophic accidents, such as severe damage of expensive equipment, serious injury to people, and huge economic loss. Therefore, the central safety inquiry in SG power systems is to identify all possible risk consequences given that one or more sudden events could happen at system/subsystem level. Spare or redundant critical components in highly critical SGs have been inbuilt in order to ensure adequate and acceptable continuity of power service without failures. However, major discussion points regarding reliability during the decision-making process at critical design stage are as follows (Bucher et al. 2013): (1) How much redundancy and at what cost? (2) Should the reliability be increased, maintained at existing levels, or allowed to degrade? (3) On what accuracy should the decision be made? It is evident that reliability and economics are related to each other, as shown in

Reliability Analysis of Smart Grids Using Formal Methods

Solarr Energy

Automated Substation

149

Smart Energy Manager

Utility

Thermal Power Plants Residential Loads

Industrial Loads

Communication C i ti Gateways

Electric Vehicles

Commercial Loads

Wind Energy

Fig. 1 A smart grid power system

a

b

Reliability (R)

Annual Cost

1.0

Total Optimum

R

Utility

C

Consumer Investment Cost (C)

Incremental Cost-Reliability

System Reliability

Total Reliability Costs

Fig. 2 Smart grid cost-reliability relationship. (a) Incremental cost reliability. (b) Total reliability costs

Fig. 2a (Allan 2013), i.e., increased investment ΔC, is required in order to improve reliability ΔR. The increment cost of reliability ΔC/ΔR is one of the ways of deciding whether an investment in the SG power system is worth it or not. The basic concept of reliability-cost/reliability-worth assessment can be presented by the cost/reliability curves, as shown in Fig. 2b (Allan 2013). It can be observed that as reliability increases, the investment cost generally increases, while the consumer costs associated with failures decrease. From the reliability analysis of SGs, we can obtain an optimum target level of reliability and costs, as shown in Fig. 2b (Allan 2013). Therefore, to make decision-making of the optimal design for a specific SG power system, planners/designers require a probabilistic risk assessment at the critical design stage.

150

1.1

M. Abdelghany and S. Tahar

Reliability Modeling Methods

ˇ Since the late 1960s, various types of reliability modeling methods (Cepin 2011) have been developed to determine the probabilistic risk assessment of SG power systems. These include predominantly graph theory-based approaches such as fault trees (FT) (Javadi et al. 2011), reliability block diagrams (RBD) (Boussahoua and Elmaouhab 2019), event trees (ET) (Muzik and Vostracky 2018), causeconsequence diagrams (CCD) (Andrews and Ridley 2002), and functional block diagrams (FBD) (Papazoglou 1998a), as shown in Fig. 3. FTs mainly provide a graphical model, using logic gates OR/AND/NOT, for analyzing the factors causing a complete SG system failure upon their occurrences only (see Fig. 3). On the other hand, RBDs provide a schematic structure, using series/parallel configurations, for analyzing the success relationships of SG system components that keep the entire power system reliable only (Fig. 3). In contrast to FTs and RBDs, ETs provide a complete risk tree model for all possible complete/partial failure and successconsequence scenarios at the system level simultaneously so that one of these possible sudden events can occur in the entire SG system, as shown in Fig. 3. More recently, an approach has been proposed to conduct ET analysis in conjunction with FTs to identify all subsystem failure events in a critical SG system and their cascading dependencies on the entire power network. This analysis method is known as cause-consequence analysis, using a combined hierarchical structure of causeconsequence diagrams (CCD), as shown in Fig. 3. Moreover, ET analysis can be used to associate failure and success events with all subsystems of the safetycritical SG power grid in more complex hierarchical structures, such as functional block diagrams (FBD). An FBD (Fig. 3) is a graphical representation of the detailed

Fault Trees (FT)

Reliability Block Diagrams (RBD)

Event Trees (ET)

Reliability Modeling Methods

Cause Consequence Diagrams (CCD)

Functional Block Diagrams (FBD)

Fig. 3 Reliability modeling methods of smart grid power systems

Reliability Analysis of Smart Grids Using Formal Methods

151

SG functionality and the functional relationship between all its subsystems (i.e., generation, transmission, and distribution) that are represented as functional blocks (FB). All subsystem-level ETs associated with their corresponding FBs are then composed together to build a subsystem-level ET model of a given smart power grid system under reliability study. Therefore, the probabilistic risk assessment of the occurrence of consequence accident events using ETs/CCDs/FBDs can be used for all required system-/subsystem-level improvements and satisfy the total reliability demand within acceptable risk levels.

1.2

Reliability Analysis Methods

The reliability analysis of SG power systems can be calculated using a variety of methods, among them analytically based paper-and-pencil methods and samplingbased Monte Carlo simulation (MCS) methods being the most popular. The former one represents the system by a mathematical model and evaluates the reliability indices from this model manually using direct numerical solutions. However, when real-world smart power grids with complex operating procedures have to be modeled, the resulting analysis can therefore lose some of its significance due to the possibility of human error-proneness, and it was a very cumbersome effort to perform reliability analysis manually. For that reason, many of designers use sampling-based MCS approach for faster computation, which uses random algorithms to predict the real functional behavior of critical SG systems and estimate the average value of reliability parameters.

2

Problem Statement

The existing analysis methods for reliability assessment of critical SG systems compromise the accuracy or completeness of the reliability parameter evaluation during the design stage. Therefore, this could lead to undesirable inaccuracies in the obtained risk results, which can be deemed fatal for SG power systems that consequently could lead to the occurrence of unexpected sudden failures. The smallest error in the safety-critical smart power grids can cause disastrous consequences in human lives as well as huge financial losses. A more accurate and safer way to the error-prone informal reasoning of the predictive reliability evaluation would be the use of formal mathematical verification as per recommendations of safety standards, such as IEC 61850 (Mackiewicz 2006) and ISO 26262 (Palin et al. 2011).

3

Proposed Solution

In this chapter, we propose a novel approach that uses formal methods (Hasan and Tahar 2015), based on theorem proving, for accurate and sound ET-/FBD/CCD-based reliability analysis of large-scale SG systems. The proposed formal

152

M. Abdelghany and S. Tahar

analysis technique allows us to perform a predictive verification of all possible SG system-/subsystem-level failure and reliability probabilistic risk expressions simultaneously. Theorem proving is a formal verification technique (Hasan and Tahar 2015), which is used for conducting the proof of mathematical theorems constructed in higher-order logic (HOL) (Hasan and Tahar 2015) based on computerized proof tools, called theorem provers. In particular, we use HOL4, which is a well-known interactive theorem prover with the ability of verifying a wide range of mathematical HOL expressions. The main advantage of using HOL formulation form is that all defined functions and verified theorems are generic, which allow us to use them for any given input system data and for N system components. Moreover, the main characteristic of HOL4 is that its core consists only of a few axioms and inference rules and any further theorem should be verified based on proven theorems. This ensured the soundness of the SG system model analysis. Moreover, since the system properties are proven mathematically within HOL4, no approximation is involved in the analysis results. These features make HOL4 suitable for carrying out the reliability analysis of the safety-critical SG power systems, which require sound verification results. To the best of our knowledge, this is the first work that develops a library for the mathematical modeling and reasoning framework of system-/subsystem-level consequence risk analysis (based on ETs, FBDs, and CCDs) using HOL4 and that uses library on real-world SG power system applications, where we provide significant improvements compared to all existing reliability analysis methods in terms of scalability, expressiveness, accuracy, and time.

4

State of the Art

In this section, we present all the related literature review and the proposed novel framework in this chapter.

4.1

Event Tree Reliability Analysis

Event tree (ET) reliability analysis has been developed in the mid-1970s (Rasmussen 1974) for the probabilistic risk assessment of all possible sudden accident risks that can occur in nuclear power plants in the generation sector of power systems. Then, in the late 1990s, Papazoglou in Papazoglou (1998b) was the first researcher to lay down the mathematical foundations of ETs to replace their graphical representation for probabilistic risk analysis of nuclear power plants. Since that time, many researchers have analyzed SG power systems using ETs for the probabilistic risk assessment at the design stage. For instance, Dialynas and Koskolos (1994) used ET analysis to model the operational consequence behavior of the high-voltage direct current (HVDC) power transmission system components as well as to deduct all possible system failure and reliability modes simultaneously. Muzik and Vostracky (2018) used the notion of ETs in analyzing

Reliability Analysis of Smart Grids Using Formal Methods

153

all emergency risk possibilities of a microgrid (MG) power system near the city of Pilsen, Czech Republic. Phulpin et al. in (2011), used the ET analysis results to improve the control strategy of HVDC transmission power flow and consequently the system sustainability aftermath of a large disturbance. In (2004), Peplow et al. used an ET diagram to evaluate the probability of all possible consequences of sudden accident events or terrorist attacks causing the contamination with radioactive material, which would make a large surrounding area uninhabitable for thousands of years. Ku and Cha, in (2011), used graph theory of ETs to determine some significant reliability indices of catenary of electric railways. However, the reliability analysis done in all the abovementioned work is done purely analytically based using a paper-and-pencil approach. A major limitation in the mathematical manual approach is the possibility of human error-proneness for large-scale realworld SG systems as well as the cumbersome effort and large amounts of time to perform the reliability analysis manually. On the other hand, there exist several commercial software tools for building the graph theory of ETs for probabilistic risk assessment of critical SG systems, such as ITEM (ITEM Software 2021) and Isograph (Isograph Software 2021). However, these commercial tools also require from the user to manually draw the ET model based on two states only of each component (success or failure) due to an explosion of outcome possible test cases. This limitation could be not suitable for real-world complex SG power systems that usually require to assign multistates of complete/partial failure and reliability events to each component.

4.2

Cause-Consequence Diagram Reliability Analysis

There exist some techniques that have been developed for subsystem-level reliability analysis of SG power systems. For instance, Papadopoulos et al. in (2011) have developed a software tool called HiP-HOPS (Hierarchically Performed Hazard Origin and Propagation Studies) for subsystem-level failure analysis to overcome classical manual failure analysis of complex SG systems and prevent human errors. HiP-HOPS can automatically generate the subsystem-level fault tree (FT) model and perform failure modes, effects, and critical analyses (FEMCA) from a given subsystem models, where each subsystem component is associated with its failure rate or failure probability (Papadopoulos et al. 2011). Currently, HiP-HOPS lacks the modeling of multistate system components and also cannot provide generic mathematical expressions that can be used to predict the reliability of an SG system based on any probabilistic distribution, like exponential/Weibull/Poisson (Kabir et al. 2019). Similarly, Jahanian in (2019) has proposed a new technique called failure mode reasoning (FMR) for identifying and quantifying the failure modes for SG power systems at the subsystem level. However, according to Jahanian (2019), the soundness of the FMR approach needs to be proven mathematically. On the other hand, cause-consequence diagram (CCD) reliability analysis typically uses FTs to analyze failures at the subsystem levels combined with an ET consequence diagram to integrate their cascading failure/reliability dependencies on the entire

154

M. Abdelghany and S. Tahar

system (Andrews and Ridley 2002). CCDs are categorized into two general methods ˇ for the ET linking process with the FTs (Cepin 2011): (1) small ET diagram and large subsystem level FT and (2) large ET diagram and small subsystemlevel FT. Both methods are used for the probabilistic risk assessment of industrial applications. For example, Andrews and Ridley, in (2002), used the former method of CCD reliability analysis (i.e., small ET and large FTs) to determine all the probabilistic risk assessment of high-integrity protection systems (HIPS). Also Andrews, in (2001), used latter approach (i.e., large ET and small FTs) to determine all possible complete/partial failure and reliability events at the subsystem level of a pressure tank system that contains a motor control center (MCC) with a startup and shutdown sequence in addition to its required operational phase. In (2006), Vyzaite et al. applied both methods the CCD method for reliability analysis of nonrepairable phased missions at the subsystem level. However, the subsystem-level CCD reliability analysis done in all the abovementioned framework is done purely analytically based using a paper-and-pencil approach. This implies that it is very difficult to apply the manual CCD analysis on complex SG power systems, where planners/designers require n-level cause-consequence analysis corresponding to n-subsystems (i.e., n-level ET model and n-level FT models).

4.3

Functional Block Diagram Reliability Analysis

In the late 1990s, Papazoglou developed the fundamentals of FBD subsystemlevel reliability analysis in Papazoglou (1998a) for more hierarchical ET structures, which is done purely analytically using the mathematical manual paper-and-pencil approach. For instance, Papakonstantinou et al., in (2013), used FBD analysis to determine all safety classes of a boiling water reactor (BWR) and steam turbine generator in a nuclear power plant generation system. Since that time, FBD analysis has not improved much due to the complexity that planners/designers are facing of building complex ET structure models manually during the design stage. A computer simulation program can be written in any modern language to automate the FBD reliability analysis proposed by Papazoglou. However, both of these analysis methods either lack detailed proof steps and are not scalable for n-level reliability analysis of real-world complex SG systems or use approximation algorithms for faster computation. Therefore, these approaches could introduce undesirable inaccuracies that can be deemed fatal for SG power systems.

4.4

Monte Carlo Simulation Reliability Analysis

Due to the limitations of the analytically mathematical manual approach, planners and designers started to consider sampling-based Monte Carlo simulation (MCS) for faster reliability analysis computation, which is built-in behind based on predictive consequence analysis using random algorithms to estimate the probability of failure and reliability (Allan 2013). For instance, in (1994), Billinton and

Reliability Analysis of Smart Grids Using Formal Methods

155

Sankarakrishnan used the random-based MCS to predict the reliability studies of a composite generation and transmission systems containing HVDC link. In (2001), Shelemy and Swatek used MCS to model of the performance of Manitoba Hydro Nelson River HVDC transmission lines. Bae et al., in (2012), proposed a technique to evaluate the reliability of distribution systems with multiple interconnected microgrids (MG) based on MCS. In (2013), Ranjbar used MCS-based reliability assessment method for reliability analysis of hybrid integrated MGs. In (2015), Shi and Bazzi applied MCS to determine the reliability of MGs for high penetration photovoltaic (PV) and wind turbine (WT) clean renewable energy systems. In Ansari et al. (2016), the authors used MCS to evaluate the reliability of smart power grids including various types of distributed green generation and energy storage systems (ESS) as well as prioritized loads to be supplied. Nanou et al., in (2016), used the MCS method to enhance power dispatch control under stochastic operating conditions for multiterminal HVDC transmission power grids. In (2000), Bevilacqua et al. used MCS to perform a predictive FEMCA analysis at the subsystem level of electrical power plants in a smart power grid. However, MCS-based reliability analysis approaches lack the rigor of detailed proof steps for reliability indices and may not be scalable for large-scale SG systems due to an explosion of the generated test cases, which requires a large amount of computing time. A more accurate and safer way to the error-prone informal reasoning of the predictive reliability evaluation would be the use of formal mathematical verification, which we propose in this chapter.

4.5

Reliability Analysis in HOL4 Theorem Proving

Only a few work have previously considered using formal methods for reliability analysis of SG power systems. For instance, N`yvlt and Rausand in (2012) used Petri nets for ET analysis to model the complete/partial system-level failure and success-consequence events. The authors proposed a new method based on Pinvariants to obtain a model of cascading dependencies in ETs (N`yvlt and Rausand 2012). However, according to the same authors, they are not able to obtain verified expressions from the generated ET model (N`yvlt and Rausand 2012). Ortmeier et al. in (2005) developed a framework for Deductive Cause-Consequence Analysis(DCCA) using a model checker SMV to verify the CCD proof obligations. However, according to the authors, there is a problem of showing the completeness of DCCA due to the exponential growth of the number of proof obligations with complex systems that need cumbersome proof efforts (Ortmeier et al. 2005). To overcome all the abovementioned limitations of existing analysis methods, we are proposing a novel approach of using theorem proving to obtain a verified system/subsystem-level failure and reliability probabilistic expressions. Prior to this work, proposed in this chapter, there were three notable projects for building formal infrastructures in HOL to formally model and analyze FTs and RBDs. For instance, Ahmad (2017) used the HOL4 theorem prover to formalize ordinary (static) FT and RBD structures. Elderhalli (2019) had formalized dynamic

156

M. Abdelghany and S. Tahar

versions of FTs and RBDs in HOL4. These formalizations have been used for the reliability analysis of several engineering systems. However, they formally analyze either a critical system static/dynamic failure or static/dynamic success only. Conversely, in this chapter, we developed a formally verified ET/CCD/FBD library in HOL4, which allowed us to a analyze all possible complete/partial failure and success-consequence risk events simultaneously at system and subsystem level, which makes our framework the first of its kind.

4.6

Formal Reliability Analysis of Smart Power Grids

Only a few researchers have recently considered using formal methods to analyze the reliability of SGs. For instance, Mahmood et al., in (2016), developed a framework for the reliability assessment of power grid components (Fang et al. 2011) with backup protection using the probabilistic model checker PRISM. Similarly, in (2013), Khurram et al. presented a foundational model for relaybased protected components in power distribution systems using the PRISM model checker. Also, in (2017), Sugumar et al. were the first to use formal analysis via the UPPAAL model checker (Larsen et al. 1997) to design and validate the energy management system (EMS) for a microgrid (MG) system that consists of high penetration of solar PV systems. Also, Sugumar et al., in (2019), used UPPAAL for the verification of a supervisory EMS, which provides much stronger confidence in the correctness of the EMS design than conventional approaches. Recently, Badings et al., in (2021), used model checking for the predictive verification of smart grids (SG) incorporating WTs and ESSs to overcome the need for samplingbased MCS and to be used by transmission system operator (TSO). However, all the abovementioned model-checking tools face a combinatorial blowup of the state space, commonly known as the state explosion problem (Hasan and Tahar 2015). Moreover, in (2017), Li et al. used the formal analysis, based on the continuous reachability analyzer (CORA) MATLAB toolbox, for the predictive verification of networked MG systems’ stability in the presence of heterogeneous uncertainties induced by high penetration of RES generation. However, CORA has a limit of only providing probabilistic failure/reliability expressions of integrated N MGs at each MG component level. On the other hand, theorem proving has been successfully utilized for the formal reliability analysis of smart power grids. For instance, Ahmad et al. in (2020a) used theorem proving to generate a capacity outage probability table (COPT) (Allan 2013) in order to estimate the overall capacity of the generation system. Also, Ahmad et al. in (2020b) used RBDs/FTs to determine the reliability/failure of various intelligent embedded devices for an automated substation. However, the work in Ahmad et al. (2020b) is limited to formally analyzing either the failure or the reliability of smart power grids compared to the work we propose which provides a comprehensive formal analysis considering both failure and success risk states of smart power grid components simultaneously at the system/subsystem level.

Reliability Analysis of Smart Grids Using Formal Methods

5

157

Proposed Methodology

Figure 4 depicts an overview of the proposed novel methodology for SG reliability analysis. This methodology allows us to formally verify system-/subsystem-level failure and reliability expressions for SG power systems, which can be evaluated and used in critical decision analysis. The core component of this methodology is the HOL4 libraries of ETs, FBD, and CCDs as depicted in container, enclosed by a box. The first step in our proposed methodology is the SG power system diagram, description, and its subsystem specifications provided by the reliability engineers. The second step is to choose which technique is suitable for the SG power system reliability analysis, i.e., either using ET analysis for system-level analysis or using FBD/CCD for subsystem-level analysis, as shown in Fig. 4. To perform the formal system-level ET analysis using our core ET library (ET structure, reduction properties, and probabilistic theorems), the reliability engineer has to provide the reliability requirements for the SG power system under reliability study, as shown in Fig. 4 with green arrows. While performing the formal subsystemlevel FBD analysis using our FBD library (FBD structure, FBD-ET translation, and probabilistic theorems), the user has to model the SG subsystem-level FBD with all its decomposed FBs, as shown in Fig. 4 with blue arrows. Similarly, using our CCD library (CCD structure, CCD reduction, and probabilistic theorems) requires the SG subsystem RBD or FT models, as shown in Fig. 4 with red arrows. Based on our rich new ET/FBD/CCD definitions and novel formulations, proposed in this research work, a user with some basic knowhow about HOL4 can easily verify all possible system-/subsystem-level safety classes of reliability and failure consequence expressions based on any given probabilistic distribution, like exponential/Weibull/Poisson, corresponding to the given SG power system description. The last step is the probabilistic computation of the formally verified reliability/failure expressions using new defined standard metalanguage (SML) functions. We can use the verified probabilistic results to determine significant reliability and energy indices (Keyhani and Albaijat 2012), such as forced outage rate (FOR), System Average Interruption Frequency Index (SAIFI), System Average Interruption Duration Index (SAIDI), Customer Average Interruption Duration Index (SAIDI), Average Service Availability Index (ASAI), Average Service Unavailability Index (ASUI), Energy not Supplied Index (ENS), Average System Curtailment Index (ASCI), loss of load expectation (LOLE), loss of energy expectation (LOEE), and Energy Index of Reliability (EIR), to help the planners/designers in making effective decisions at the critical design stage. Lastly, we propose a new Functional Block Diagram and Event Tree Modeling and Analysis (FETMA) Software for industrial engineers, which is mainly build based on our verified mathematical equations in HOL4 to overcome the limitations of all existing analysis methods. In order to demonstrate the practical effectiveness of our proposed methodology, we use our framework to conduct a formal system-/subsystem-level reliability and safety analysis of real-world smart power systems in the major sectors (Keyhani and

158

M. Abdelghany and S. Tahar

Smart Grid System NO

Refinement

Required? System

System

Subsystems

Diagram

Description

Specifications

YES FBD Decomposition (Subsystem Level)

ET Library Reliability Engineer

CCD Library

FBD Library

ET Structure

FBD Structure

CCD Structure

Reduction Properties

FBD-ET Translation Probabilistic Theorems

CCD Reduction

Probabilistic Theorems

Probabilistic Theorems

System Level FBD Model

Smart Grid Applications Power Generation

Reliability

System-Level

Subsystem-Level

Requirements

Reliability Analysis

Reliability Analysis

Probabilistic

Power Distribution Power Protection

Subsystems RBD/FT Models

Distributions

Power Transmission

Formally Verified System/Subsystem-Level Smart Grid Reliability Expressions

Probability Computation

HOL4 Theorem Prover

SML Functions

FETMA Software ET Reliability Analysis FBD Reliability Analysis

Smart Grid Reliability and Energy Indices (FOR, SAIFI, SAIDI, CAIDI, ASAI, ASUI, ASCI, ENS, LOLE, LOEE, EIR)

Probability Computation

Fig. 4 Proposed methodology

Albaijat 2012), (i) power generation plants, (ii) power transmission grids, (iii) power distribution networks, and (iv) power protection systems, as shown in Fig. 4.

6

Smart Grid Power System Applications

In this section, we utilize our proposed novel methodology on real-world SG power system applications, as follows: – In Abdelghany and Tahar (2020), we conducted the formal system-level ET reliability analysis of the standard IEEE 3-bus composite bulk power system incorporating 50% distributed renewable energy resources (RES) systems, as shown in Fig. 5. Also, we performed the predictive verification of some significant reliability indices, such as SAIFI, SAIDI, and CAIDI, for the entire power grid. – In Abdelghany et al. (2021), we conducted the formal system-level ET reliability analysis of different load locations in the standard IEEE 118-bus transmission line system representing a portion of the American electric power system (in the Midwestern USA), as shown in Fig. 6. Moreover, we determined an important SG energy indices, such as ENS, ASCI, LOLE, LOEE, and EIR, for the transmission power system under reliability study. – In Abdelghany and Tahar (2021), we conducted the formal subsystem-level CCD reliability analysis of the IEEE 39-bus electrical power network incorporating 50% truly carbon-neutral or emission-free green power generation, as shown

Reliability Analysis of Smart Grids Using Formal Methods

159

Steam Generator Power Plant G1

Solar Generator Power Plant G2

M1

M2 M3

Substation 1

Substation 2 L1

L2 Substation 3

Load B

Load A Load C

Fig. 5 IEEE 3-Bus Composite Bulk Power System

B8 G

B9

B 47

B 10

B7

B 41

B 40

B 42

B 45

G

G

B 46

Zone 1 B5

B 48 B 43

B 32

B 11

B6

G

B 49

G

G

B 12

B4

Load A

B 29 B 31

B 30 B 28 B3

B 27

B 26

B 14

B2

B 15

G1 B 34

B 35

B 24 B 23

B 22

TL3

TL1

B 51

TL6

B 36

B 55 TL8

TL9

TL7

G

B 60

B 57

B 56

TL10

G2 B 18

B 53

B 58

G

G3

TL11

Zone 3

B 66

B B 62

B 16

B 52

B 54

TL2 B 59

B 37

B 19

B 50

B 38

TL4

B 95

B 25 B 17

B 44

B 33

G

TL5

B 13

B 39

B 63

G

B 105 B 103

B 107

B 65

B 64

B1

B 104

Load C

B 86

B 20

B 106 G

B 21 G

B 84

B 85

B 71 B 67

TL22 B 83

B 69

B 73

B 87

TL23

B 74

B 82 B 89

B 75 B 81

G

B 76

B 79

B 91 TL28

B 93 B 103

TL33

B 99 TL29

G

G8

G

B 92

TL31

B 96 TL32 B 97

TL30

Zone 2

B 108

G7 B 104

B 102

B 100

TL27

B 77 B 78

B 88

TL26

B 90

B 80

B 89

TL25

G9 G

B 70

B 101

TL24

B 72

B 68

G4

B 111

G

TL16

G5

B 112

G

TL17

TL21

B 94 B 109 TL14 TL13 TL12

TL19

B 113

TL20 TL18

TL15

G

B 115

B 117

G6

B 110 B 114

Load B

B 116

B 118

Fig. 6 IEEE 118-bus transmission power grid system

in Fig. 7. Also, we automatically generated the forced outage rate (FOR) probabilistic expressions of all power generation units and the network reliability index SAIDI complex expression at each generation subsystem level. – In Abdelghany et al. (2020), we apply our FETMA software on a probabilistic risk ET analysis of the distance protection fault tripping circuits in different zones of SG power transmission lines, as shown in Fig. 8. We built a decision

160

M. Abdelghany and S. Tahar

G7

G9 Bus 37

Bus 26

Legend

Bus 6

Bus 25

Load B

Bus 30

TL

Bus 29

Bus 2

Load A Bus 18

Power Flow

Bus 38

Generator

G5

Bus 27

S/S

Bus 1 Bus 17 Bus 39

Bus

Load D

Bus 3 Bus 4

Bus 9

Bus 15

Bus 21

Bus 5

Bus 22

Bus 14

Bus 13

Bus 12

Bus 11

Bus 8

Power Outage

Bus 16

G10

Bus 24

Bus 35

Bus 19

G4

Load C

Bus 7

Bus 10 Bus 31 G1

Bus 32 G2

Bus 34 G6

Supplied From

A

G9, G5

B

G7, G9

C

G1, G2

D

G6, G3, G8, G4

Bus 23

Bus 20

Bus 28

Load

Bus 33

Bus 36 G8

G3

Fig. 7 IEEE Standard 39-bus generation network

Legend CB

Circuit Breaker

TC

Fault Trip Circuit

CT Current Transformer PT Potential Transformer

Zone2 (120%)

Zone3 (220%)

Zone1 (80%)

CT1

Fault

CB1

CB2

CT2

Transmission Line TC1

PT1

Distance Relay 1

TC2

PT2

Distance Relay2

Fig. 8 Power transmission distance protection

tree describing the process of selecting the redundancy for critical transmission protection components. Moreover, the CPU time for the transmission protection step-wise ET analysis in FETMA is much faster than existing tools, like the commercial Isograph ET analysis tool.

Reliability Analysis of Smart Grids Using Formal Methods

7

161

Conclusions

We proposed a novel methodology based on formal methods to conduct the predictive verification of system-/subsystem-level reliability analysis of smart grids (SG) as an accurate alternate analysis approach. For that purpose, we built a formally verified complete library using HOL4 theorem proving for analyzing largescale reliability risk modeling methods, i.e., event trees (ET), cause-consequence diagrams (CCD), and functional block diagrams (FBD). We demonstrated the practical effectiveness of our proposed methods by conducting the formal system/ subsystem-level reliability analysis of the standard IEEE 3-bus composite bulk power system, IEEE 118-bus transmission system, IEEE standard 39-bus generation network, and power transmission distance protection scheme. Moreover, we accurately determined significant reliability and energy indices, such as SAIFI, SAIDI, CAIDI, ASAI, ASUI, ENS, ASCI, LOLE, LOEE, and EIR.

References M. Abdelghany, S. Tahar, Event tree reliability analysis of electrical power generation network using formal techniques, in Electric Power and Energy Conference (IEEE, Edmonton, 2020), pp. 1–7 M. Abdelghany, S. Tahar, Cause-consequence diagram reliability analysis using formal techniques with application to electrical power networks. IEEE Access 9, 23929–23943 (2021) M. Abdelghany, W. Ahmad, S. Tahar, S. Nethula, ETMA: an efficient tool for event trees modeling and analysis, in IEEE International Systems Conference (IEEE, Montreal, 2020), pp. 1–8 M. Abdelghany, W. Ahmad, S. Tahar, Event tree reliability analysis of safety critical systems using theorem proving. IEEE Syst. J. (2021). [Online]. Available: https://doi.org/10.1109/JSYST. 2021.3077558 W. Ahmad, Formal dependability analysis using higher-order-logic theorem proving. Ph.D. dissertation, National University of Sciences and Technology, Islamabad, 2017 W. Ahmad, O. Hasan, F. Awwad, N. Bastaki, S. Hasan, Formal reliability analysis of an integrated power generation system using theorem proving. IEEE Syst. J. 14(4), 4820–4831 (2020a) W. Ahmad, O. Hasan, S. Tahar, Formal reliability and failure analysis of ethernet based communication networks in a smart grid substation. Form. Asp. Comput. 32(1), 71–111 (2020b) R. Allan, Reliability Evaluation of Power Systems (Springer Science & Business Media, New York, 2013) J. Andrews, Reliability of sequential systems using the cause-consequence diagram method. J. Process Mech. Eng. 215(3), 207–220 (2001) J. Andrews, L. Ridley, Application of the cause-consequence diagram method to static systems. Reliab. Eng. Syst. Saf. 75(1), 47–58 (2002) O. Ansari, N. Safari, C. Chung, Reliability assessment of microgrid with renewable generation and prioritized loads, in Green Energy and Systems Conference (IEEE, 2016), pp. 1–6 T. Badings, A. Hartmanns, N. Jansen, M. Suilen, Balancing wind and batteries: towards predictive verification of smart grids, in NASA Formal Methods Symposium (Springer, 2021), pp. 1–18 M. Bevilacqua, M. Braglia, R. Gabbrielli, Monte Carlo simulation approach for a modified FMECA in a power plant. Qual. Reliab. Eng. Int. 16(4), 313–324 (2000) Z. Bie, P. Zhang, G. Li, B. Hua, M. Meehan, X. Wang, Reliability evaluation of active distribution systems including microgrids. IEEE Trans. Power Syst. 27(4), 2342–2350 (2012)

162

M. Abdelghany and S. Tahar

R. Billinton, A. Sankarakrishnan, Adequacy assessment of composite power systems with HVDC links using Monte Carlo simulation. IEEE Trans. Power Syst. 9(3), 1626–1633 (1994) B. Boussahoua, A. Elmaouhab, Reliability analysis of electrical power system using graph theory and reliability block diagram, in Algerian Large Electrical Network Conference (IEEE, 2019), pp. 1–6 M. Bucher, R. Wiget, G. Andersson, C. Franck, Multiterminal HVDC networks – what is the preferred topology? IEEE Trans. Power Delivery 29(1), 406–413 (2013) ˇ M. Cepin, Assessment of Power System Reliability: Methods and Applications (Springer Science & Business Media, London, 2011) E. Dialynas, N. Koskolos, Reliability modeling and evaluation of HVDC power transmission systems. IEEE Trans. Power Delivery 9(2), 872–878 (1994) Y. Elderhalli, Dynamic dependability analysis using HOL theorem proving with application in multiprocessor systems. Ph.D. dissertation, Concordia University, 2019 X. Fang, S. Misra, G. Xue, D. Yang, Smart grid – the new and improved power grid: a survey. IEEE Commun. Surv. Tutor. 14(4), 944–980 (2011) O. Hasan, S. Tahar, Formal verification methods, in Encyclopedia of Information Science and Technology (IGI Global, 2015), pp. 7162–7170 Isograph Software, 2021. [Online]. Available: https://www.isograph.com ITEM Software, 2021 [Online]. Available: https://itemsoft.com/eventtree.html H. Jahanian, Failure mode reasoning, in International Conference on System Reliability and Safety (IEEE, 2019), pp. 295–303 M. Javadi, A. Nobakht, A. Meskarbashee, Fault tree analysis approach in reliability assessment of power system. J. Multidiscip. Sci. Eng. 2(6), 46–50 (2011) S. Kabir, K. Aslansefat, I. Sorokos, Y. Papadopoulos, Y. Gheraibia, A conceptual framework to incorporate complex basic events in HiP-HOPS, in Model-Based Safety and Assessment, vol. 11842 (Springer, 2019), pp. 109–124 A. Keyhani, M. Albaijat, Smart Power Grids (Springer Science & Business Media, Berlin/Heidelberg, 2012) A. Khurram, H. Ali, A. Tariq, O. Hasan, Formal reliability analysis of protective relays in power distribution systems, in Formal Methods for Industrial Critical Systems (Springer, Berlin/Heidelberg, 2013), pp. 169–183 B. Ku, J. Cha, Reliability assessment of catenary of electric railway by using FTA and ETA analysis, in Environment and Electrical Engineering (IEEE, 2011), pp. 1–4 K. Larsen, P. Pettersson, W. Yi, UPPAAL in a nutshell. Int. J. Softw. Tools Technol. Transfer 1(1–2), 134–152 (1997) Y. Li, P. Zhang, P. Luh, Formal analysis of networked microgrids dynamics. IEEE Trans. Power Syst. 33(3), 3418–3427 (2017) R. Mackiewicz, Overview of IEC 61850 and benefits, in Power Engineering Society General Meeting (IEEE, Montreal, 2006), pp. 623–630 A. Mahmood, O. Hasan, H. Gillani, Y. Saleem, S. Hasan, Formal reliability analysis of protective systems in smart grids, in Region 10 Symposium (IEEE, 2016), pp. 198–202 V. Muzik, Z. Vostracky, Possibilities of event tree analysis method for emergency states in power grid, in International Scientific Conference on Electric Power Engineering (IEEE, 2018), pp. 1–5 S. Nanou, O. Tzortzopoulos, S. Papathanassiou, Evaluation of an enhanced power dispatch control scheme for multi-terminal HVDC grids using Monte-Carlo simulation. Electric Power Syst. Res. 140, 925–932 (2016) O. N`yvlt, M. Rausand, Dependencies in event trees analyzed by petri nets. Reliab. Eng. Syst. Saf. 104, 45–57 (2012) F. Ortmeier, W. Reif, G. Schellhorn, Deductive cause-consequence analysis. IFAC Proc. Vol. 38(1), 62–67 (2005) R. Palin, D. Ward, I. Habli, R. Rivett, ISO 26262 safety cases: compliance and assurance, in IET Conference on System Safety, Birmingham (2011), pp. 1–6

Reliability Analysis of Smart Grids Using Formal Methods

163

Y. Papadopoulos, M. Walker et al., Engineering failure analysis and design optimisation with HiPHOPS. Eng. Failure Anal. 18(2), 590–608 (2011) N. Papakonstantinou, S. Sierla, B. O’Halloran, Y. Tumer, A simulation based approach to automate event tree generation for early complex system designs, in Design Engineering Technical Conferences and Computers and Information in Engineering Conference, vol. 55867 (American Society of Mechanical Engineers, 2013), pp. 1–10 I. Papazoglou, Functional block diagrams and automated construction of event trees. Reliab. Eng. Syst. Safety 61(3), 185–214 (1998a) I. Papazoglou, Mathematical foundations of event trees. Reliab. Eng. Syst. Saf. 61(3), 169–183 (1998b) D. Peplow, C. Sulfredge et al., Calculating nuclear power plant vulnerability using integrated geometry and event/fault-tree models. Nucl. Sci. Eng. 146(1), 71–87 (2004) Y. Phulpin, J. Hazra, D. Ernst, Model predictive control of HVDC power flow to improve transient stability in power systems, in International Conference on Smart Grid Communications (IEEE, 2011), pp. 593–598 A. Ranjbar, Reliability analysis of modern hybrid micro-grids. Ph.D. dissertation, The University of Texas at Dallas, 2013 N. Rasmussen, Reactor Safety Study: An Assessment of Accident Risks in US Commercial Nuclear Power Plants, vol. 7 (NTIS, 1974) S. Shelemy, D. Swatek, Monte Carlo simulation of lightning strikes to the nelson river HVDC transmission lines, in International Conference on Power System Transients (2001), pp. 1–6 X. Shi, A. Bazzi, Fault tree reliability analysis of a micro-grid using Monte Carlo simulations, in Power and Energy Conference (IEEE, 2015), pp. 1–5 G. Sugumar, R. Selvamuthukumaran et al., Formal validation of supervisory energy management systems for microgrids, in Industrial Electronics Society (IEEE, 2017), pp. 1154–1159 G. Sugumar, R. Selvamuthukumaran, M. Novak, T. Dragicevic, Supervisory energy-management systems for microgrids: modeling and formal verification. IEEE Ind. Electron. Mag. 13(1), 26– 37 (2019) G. Vyzaite, S. Dunnett, J. Andrews, Cause-consequence analysis of non-repairable phased missions. Reliab. Eng. Syst. Saf. 91(4), 398–406 (2006)

Multi-objective Reliability-Based Design Optimization of Shell-and-Tube Heat Exchangers Using Combined Subset Simulation Method and Naive Bayes Algorithm Sima Ohadi, Jafar Jafari-Asl, Oscar D. Lara Montaño, and Naser Safaeian Hamzehkolaei

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Mathematical Models for Shell-and-Tube Heat Exchanger Design . . . . . . . . . . . . . . . . . . 2.1 STHE Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Reliability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Proposed Reliability Assessment Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Multi-Objective Model for STHE Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

166 167 168 171 172 173 175 179 181 181

Abstract

In today’s advanced world, optimization of energy systems to reduce energy consumption and costs in the industrial area has received much attention. Meanwhile, the optimal design of energy consuming components and equipment plays an influential role in this process. Shell and tube heat exchangers (STHE)

S. Ohadi · J. Jafari-Asl () Department of Civil Engineering, Faculty of Engineering, University of Sistan and Baluchestan, Zahedan, Iran e-mail: [email protected]; [email protected] O. D. Lara Montaño Departamento de Ingeniería Química, División de Ciencias Naturales y Exactas, Campus Guanajuato, Universidad de Guanajuato, Guanajuato, Mexico e-mail: [email protected] N. Safaeian Hamzehkolaei Department of Civil Engineering, Bozorgmehr University of Qaenat, Qaen, Iran e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_96

165

166

S. Ohadi et al.

are used as the most common type of heat exchangers in many industries, so optimal design of this equipment in order to save design and operation costs is important. Due to the nature of the problem, various environmental and structural parameters play a role in the final design, which causes many uncertainties in the design process of this equipment. Therefore, in this study, the optimal design of STHE with a multi-objective approach including minimizing costs and maximizing the reliability index was investigated. For this purpose, a new hybrid method including Naive Bayesian and subset simulation method is proposed. The redesign results show that the developed framework has the promising potential to enhance computational time efficiency with high accuracy for the multi-objective reliability-based design optimization of STHE. Moreover, the curve of Pareto front and a set of robust design values of STHE are obtained for the high-safety level design of STHE. According to the results, considering the uncertainties during the design lead to a safe and robust model. Since design uncertainty is unavoidable, the results are closer to reality and show greater accuracy. Keywords

Heat exchanger · Reliability-based design optimization · Subset simulation · Naïve baysian · Bell-Delaware

1

Introduction

In recent decades, the growth of industry and energy demand and rising energy costs have made optimizing energy systems as the major challenge for engineers and researchers around the world (Jamil et al. 2020; Gholizadeh et al. 2020). Heat exchangers are used as one of the main components in most industries such as power generation, oil and gas, petrochemical, automotive, and aerospace industries (Azarkish and Rashki 2019; Sahin et al. 2011). Among the various types, shell and tube heat exchangers (STHE) are of special importance due to their higher heat transfer capability and high pressure applications (Hadidi et al. 2013; Jamil et al. 2020; El Maakoul et al. 2016). Design optimization of a shell and tube heat exchangers is a very complex, time consuming, and costly operation. Optimization means finding the best answer from a set of options, which is an important and determinative activity in design process. Designers will be able to provide the best designs when they can save time and money with optimization methods (Sahin et al. 2011). Although several techniques have been proposed for the optimal design of shell and tube heat exchangers, this process is more complex and difficult than could be easily solved with traditional optimization methods such as mathematical programming methods and the like (Hadidi et al. 2013; Sahin et al. 2011). Nowadays, with the development and expansion of various optimization methods and intelligent algorithms, it can be seen the great changes in solving complex optimization problems like this.

Multi-objective Reliability-Based Design Optimization of Shell-and-Tube. . .

167

Among the efforts made around this area could mention the use of genetic algorithm (Wildi-Tremblay and Gosselin 2007; Caputo et al. 2008; Xie et al. 2008; Ponce-Ortega et al. 2009; Amini and Bazargan 2014; Sadeghzadeh et al. 2015; Jamil et al. 2020), particle swarm optimization algorithm (Patel and Rao 2010; Hajabdollahi et al. 2011; Mariani et al. 2012; Sadeghzadeh et al. 2015), artificial bee colony algorithm (Sahin et al. 2011), differential evolution algorithm (Singh and Pant 2014; Shaik and Babu 2016), harmony search algorithm (Fesanghary et al. 2009), imperialist competitive algorithm (Hadidi et al. 2013), Falcon optimization algorithm (de Vasconcelos Segundo et al. 2019), cuckoo search algorithm (Asadi et al. 2014; Khosravi et al. 2015), biogeography-based algorithm (Hadidi and Nazari 2013), and firefly algorithm (Khosravi et al. 2015; Mohanty 2016) to optimize the design of shell and tube heat exchangers. Most of the mentioned studies, however, neglect that the performance of STHE can be affected by uncertain design parameters. It is well known that the uncertainties of design parameters can affect the calculation of pressure drop with are mainly simulation-based (Azarkish and Rashki 2019). As a result, the solutions obtained from a deterministic optimization approach are also affected by uncertainty in the design parameters. Also, according to its nature, most of the engineering problems require the optimization of several goals simultaneously, which are often in conflict. So that by improving one goal, the other goal goes to the undesirable (Mirjalili et al. 2017a). The main goal in solving a multi-objective optimization problem is to find a set of optimal solutions called Pareto front that creates a relative balance between different objectives, according to development and emergence of diverse multi-objective optimization algorithms and based on No-Free-Lunch (NFL) theorem; it is necessary to examine the usefulness and efficiency of different algorithms in solving various engineering problems (Mirjalili et al. 2017b; Wolpert and Macready 1997). The objective of this work is to develop a novel structural reliability analysis method to improve computational time efficiency and accuracy of a traditional method to evaluate the safety level of engineering systems with non-linear performance functions. Then the developed method is coupled with the non-dominated sorting genetic algorithm (NSGA-II) to apply the multi-objective reliability-based design optimization (MORBDO) process of the STEH with the pressure drop failure.

2

Mathematical Models for Shell-and-Tube Heat Exchanger Design

In this study, a RBDO framework offers for optimizing the layout of shell-and-tube heat exchanger (STHE) under uncertainties. The main goal is to find the STHE design with the minimum total annual cost (TAC). The inlet and outlet temperatures for the hot and cold streams are known, also, the stream flow is known. To obtain the STHE design that presents the minimum TAC it is necessary to predict the thermohydraulic performance of the equipment, this is done by implementing the BellDelaware design methodology.

168

2.1

S. Ohadi et al.

STHE Modeling

The TAC is the sum of the operating cost that is highly dependent on the pressure drop of both sides of the heat exchanger, and the cost of the equipment that depends on the required heat transfer area. The heat transfer area is calculated with the Eq. (1), where Q is the heat duty, U is the overall heat transfer coefficient, FLM is the logarithmic mean temperature difference correction factor, and TLM is the temperature mean temperature difference computed with Eq. (2). The variable FLM is obtained with the Eq. (3) that depends on R and P, calculated with Eqs. (4) and (5), respectively. Thi and Tho are the inlet and outlet temperatures for hot stream, while Tci and tco are the inlet and outlet temperatures for the cold stream. A= TLM =

Q UTLM FT

(1)

(Thi − Tco ) − (Tho − Tci ) −Tco ln TThi ho −Tci

(2)

FLM =

+1 R−1

R2

ln

ln

1−P 1−P R

√

(3)

2 2−P R+1−√R +1 2−P R+1+ R 2 +1

R=

(Thi − Tho ) (Tco − Tci )

(4)

P =

(Tco − Tci ) (Thi − Tci )

(5)

The overall heat transfer coefficient is calculated according to the Eq. (6). hs and ht are the convective heat transfer coefficients for the shell side and tube side, respectively. Rsf and Rtf are the fouling resistance for the shell side and tube side, respectively. do is the outer diameter of tubes, di is the inner diameter of tubes, and k is the thermal conductivity of the tube wall material. U=

1 hs

+ Rsf +

do 2k ln

1 do di

+

do di Rtf

+

do 1 di ht

(6)

2.1.1 Shell Side The Bell-Delaware methodology calculates an ideal shell-side convective heat transfer coefficient, hid , that is employed to calculate the shell-side heat transfer coefficient, hs , using five correction parameters as shown in Eq. (7). Jc is the correction factor for baffle configuration, Jl is the correction factor for baffle leakage effects, Jb is the correction factor for bundle and pass partition bypass, and Js is the

Multi-objective Reliability-Based Design Optimization of Shell-and-Tube. . .

169

correction factor for baffle spacing that is larger at the inlet and outlet sections, and Jr is the correction factor for the adverse temperature gradient in laminar flow. hs = hid Jc Jl Jb Js Jr

(7)

The variable hid is calculated with the Eq. (8), where j is the Colburn factor, Ao, cr is the flow area at or near the shell center-line for one cross-flow section, Pr is the Prandlt number, and Cps is the heat capacity of the fluid placed in the shell side. The value of j is calculated with the Eq. (9). a1 , a, and a2 can be obtained from WildiTremblay and Gosselin (2007). Pt is the tube pitch and Res the Reynolds number in shell side. −2/3

hid = j

Cps P rs Ao,cr

j = a1

1.33 Pt /do

(8) a (Res )a2

(9)

The pressure drop in the shell side is obtained with the Eq. (10). pb, id and pw, id are the ideal pressure drop in the central section and the ideal window pressure drop, respectively. Nb is the number of baffles, Nr, cw is the number of effective tube rows in crossflow in the window section, and Nr, cc is the number of tube rows crossed during flow through one crossflow section between baffle tips. ζ b , ζ l , and ζ s are the correction factors for tube-to-baffle and baffle-to-shell leakage, bypass flow, and for the inlet and outlet sections having different spacing from the central section, respectively.

Nr,cw Ps = (Nb − 1) pb,id ζb + Nb pw,id ζl + 2pb,id 1 + ζb ζs Nr,cc

(10)

More details of the Bell-Delaware design methodology can be found in Shah and Sekulic (2003).

2.1.2 Tube Side The equation used to calculate de heat transfer coefficient in the tube side depends on the value of the Reynolds number, Ret , as shown in Eqs. (11), (12), and (13) for laminar, transient, and turbulent flow, respectively. Where ki is the thermal conductivity of the fluid placed in the tube side, Prt is the Prandtl number in the tube side, μt is the dynamic viscosity of the fluid in the tube side, and μw is the dynamic viscosity of the fluid at the temperature of the wall, and Nt is the number of tubes. L is the length of the tubes calculated with Eq. (14), and ft is the Darcy friction factor computed with Eq. (15).

170

S. Ohadi et al.

kt Ret P rt di (1/3) ht = 1.86 if Ret < 2300 di L ⎡ ht =

(11)

⎤

ft kt ⎢ 2 Ret P rt ⎣ 0.5 di 2/3 P rt 1.07 + 12.7 f2t

⎥ ⎦ if 2300 ≤ Ret ≤ 10000 −1

kt 0.8 1/3 μt 0.14 if Ret > 10000 ht = Ret P rt di μw L=

A πdo Nt

(12)

(13)

(14)

f = (1.82log (Ret ) − 1.64)−2

(15)

The pressure drop of tube side, Pt is caused due to the length of tubes, elbows, and inlet and outlet losses of the nozzles. Pt is computed with the Eq. (16). Np is the number of tube passes, ρt is the density of the fluid on the tube side, and vt is the velocity of the fluid in the tube side. Pt =

ρt vt2 2

L ft + 4 Np di

(16)

2.1.3 Cost Estimation The calculation of TAC is the summation of the capital cost or cost of the equipment, Cc , that must be annualized, and the operating cost, Cop , that refers to the cost of pumping per year. To calculate Cc , it is necessary to obtain the purchase cost of the equipment, CP, with Eq. (17) (Smith 2005). The purchase cost depends on the calculated heat transfer area. CP = 3.28 × 10

4

A 80

0.68 (17)

The purchase is multiplied for CM , CP , and CT , these are factors to consider the construction material, the operating pressure, and the operating temperature, respectively. Equation (18) shows the calculation of the setup cost. Ci = CM CP CT CP

(18)

The operating cost of the STHE is calculated using the pumping powers of the cold and hot streams, it is computed with Eq. (19). Es and Et are the pump power for shell and tube, respectively. Hr is the number of working hours per year and Ec is the cost of energy.

Multi-objective Reliability-Based Design Optimization of Shell-and-Tube. . .

Cop =

(Es + Et ) Ec Hr 1000

171

(19)

The TAC is calculated with Eq. (20), where the cost of the equipment is annualized. Where, n is the projected lifetime of the STHE in years and r is the interest rate. TAC = Ci

2.2

r(1 + r)n + Cop (1 + r)n − 1

(20)

Reliability Analysis

According to the reliability theory, the performance of a system under uncertainty of parameters is expression by a limit state function. Assume, X = [x1 , x2 , . . . , xn ]T is a vector of uncertainty parameters that affect the system resistance (R(X)) and loads (L(X)). The system fails when G(X) = R(X) − L(X) < 0, where G(X) is the limit state function of the system (see Fig. 1) (Jafari-Asl et al. 2021). Therefore, the failure probability (Pf ) of the system can be expressed as follows: Pf = P G R (X) , L (X) =

fX (x) dx

(21)

G(X)0

Failure domain

G(X) sl or “reject” otherwise. In the AD test code, a negligible number of 10−100 is added inside the Log[1 − p + 10−100 ] in the test statistics of the block because the size of some specimens might be equal to the complete sample size and thus p = 1 making Log[1 − p] undefined otherwise. In the following, the CVM and AD codes are presented for the two distributions. The users need to type them in a Mathematica sheet for implementation.

226

O. Kittaneh

Code I. The Cramér–von Mises Test Mathematica Code for Weibull distribution (*[1] The Inputs of the Code data: the observed data, t: the time of censoring, n: the complete sample size, M: number of sets of specimens, sl: the significance level*) data= {} t= n= M= Sl=

(*[2] computing the MLEs of

and

of the given data*)

r = Length[data]; α0 =

Log[Log[1 − 0.97]] − Log[Log[1 − 0.17]] ; Log[Quantile[data, 0.97]] − Log[Quantile[data, 0.17]]

1 β0 = ( ∗ Total[dataα0 ])1⁄α0 ; r A[c_, d_] = −(n − r)(t/c)d ∗ Log[t/c] + r + Total[Log[(data/c)d ]] − Total[(data/c)d ∗ Log[(data/c)d ]]; B[c_, d_] = (n − r) ∗ (t/c)d + Total[(data/c)d ] − r; α = FindRoot[{A[c, d] == 0, B[c, d] == 0}, {{d, α0}, {c, β0}}][[1]][[2]]; β = FindRoot[{A[c, d] == 0, B[c, d] == 0}, {{d, α0}, {c, β0}}][[2]][[2]];

(*[3] The Test Statistics of the Observed Data*) F[x_] = 1 − Exp[−(x/β)α]; dataSORT = Sort[data] ; z = F[dataSORT]; p = r/n; r

TS = ∑ (z[[i]] − (2i − 1)/(2 ∗ n))^2 − i=1

r ∗ (4 ∗ r^2 − 1) r^2 r 1 +n∗p∗( − p ∗ ( ) + ∗ p^2); 12 ∗ n^2 n^2 n 3

(*[4] The Simulated Test Statistics Distribution *) CVMtestW[α_, β_, n_, t_] ≔ Block[{UNIFdata, COMdata, COMdataSORT, r, CENdata, αcen, βcen, F1, z, p, CVM, A, B}, UNIFdata = RandomReal[{0,1}, n]; COMdata = β ∗ (−1 ∗ Log[1 − UNIFdata])1⁄α ; COMdataSORT = Sort[COMdata]; r = Count[COMdataSORT, u_ /; u ≤ t];

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data CENdata = Table[COMdataSORT[[i]], {i, 1, r}]; A[c_, d_] = −(n − r)(t/c)d ∗ Log[t/c] + r + Total [Log[(CENdata/c)d ]] − Total [(CENdata/c)d ∗ Log[(CENdata/c)d ]] ; B[c_, d_] = (n − r) ∗ (t/c)d + Total[(CENdata/c)d ] − r; αcen = FindRoot[{A[c, d] == 0, B[c, d] == 0}, {{d, α}, {c, β}}][[1]][[2]]; βcen = FindRoot[{A[c, d] == 0, B[c, d] == 0}, {{d, α}, {c, β}}][[2]][[2]]; F1[x_] = 1 − Exp[−(x⁄βcen)αcen ]; z = F1[CENdata]; p = r⁄n ; r

CVM = ∑ (z[[i]] − (2i − 1)/(2 ∗ n))^2 − i=1

r ∗ (4 ∗ r^2 − 1) r^2 r 1 +n∗p∗( − p ∗ ( ) + ∗ p^2); 12 ∗ n^2 n^2 n 3

{CVM}]

(*[5] The Simulated Critical and p Values*) = Quiet[Table[CVMtestW[α, β, n, t], M]]; b = Table[ [[i]][[1]], {i, 1, M}]; b = Complement[b, {Indeterminate}]; CV = Quantile[Re[b],1 − sl]; PV = N[Length[Select[b, # > TS &]]/M];

(*[6] The Outputs of the Code*) Print["The CVM Test Results for Weibull Distribution"] Print["The MLE of α = "] α Print["The MLE of β = "] β Print["The test statistics of the observed data = " ] TS Print["The simulated critical value of the test = " ] CV

227

228

O. Kittaneh

Code II. The Cramér–von Mises Test Mathematica Code for the Lognormal Distribution (*[1] The Inputs of the Code data: the observed data, t: the time of censoring, n: the complete sample size, M: number of sets of specimens, sl: the significance level*)data= {} t= n= M= Sl=

(*[2] computing the MLEs of

and

of the given data*)

r = Length[data]; θ0 = N [

τ0 = (

Total[Log[data]] ]; r

Total[(Log[data] − θ0)2 ] ) r

A[m_, s_] = (Log[t] − m) ∗ ( m −

B[m_, s_] =

0.5

;

Total[Log[data]] Total[(Log[data] − m)2 ] )+ − s2 ; r r

Total[Log[data]] n − r + ∗s∗ r r

Log[t] − m ] s − m; Log[t] − m ] 1 − CDF [NormalDistribution[0,1], s PDF [NormalDistribution[0,1],

τ = FindRoot[{A[m, s] == 0, B[m, s] == 0}, {{m, θ0}, {s, τ0}}][[2]][[2]]; θ = FindRoot[{A[m, s] == 0, B[m, s] == 0}, {{m, θ0}, {s, τ0}}][[1]][[2]];

(*[3] The Test Statistics of the Observed Data*) 1 −θ + Log[x] F[x_] = (1 + Erf[ ]); 2 √2τ dataSORT = Sort[data] ; z = F[dataSORT]; = / ; = ∑ ( [[ ]] − (2 − 1)/(2 ∗ ))^2 − =1

∗ (4 ∗ ^2 − 1) + 12 ∗ ^2

∗

∗(

^2 − ^2

1 ∗ ( ) + ∗ ^2); 3

(*[4] The Simulated Test Statistics Distribution *) CVMtestL[θ_, τ_, n_, t_] ≔ Block[{UNIFdata, COMdata, COMdataSORT, r, CENdata, θcen, τcen, F1, z, p, CVM},

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data COMdataSORT = Sort[COMdata]; r = Count[COMdataSORT, u_ /; u ≤ t]; CENdata = Table[COMdataSORT[[i]], {i, 1, r}]; A[m_, s_] = (Log[t] − m) ∗ ( m −

B[m_, s_] =

Total[Log[CENdata]] Total[(Log[CENdata] − m)2 ] )+ − s2 ; r r

Total[Log[CENdata]] n − r + ∗s∗ r r

Log[t] − m ] s − m; Log[t] − m ] 1 − CDF [NormalDistribution[0,1], s PDF [NormalDistribution[0,1],

θcen = FindRoot[{A[m, s] == 0, B[m, s] == 0}, {{m, θ}, {s, τ}}][[1]][[2]]; τcen = FindRoot[{A[m, s] == 0, B[m, s] == 0}, {{m, θ}, {s, τ}}][[2]][[2]]; 1 −θcen + Log[x] F1[x_] = ( 1 + Erf [ ]) ; 2 √2τcen z = F1[CENdata]; p = r⁄n ; r

CVMdata = ∑ (z[[i]] − (2i − 1)/(2 ∗ n))^2 − i=1

r ∗ (4 ∗ r^2 − 1) r^2 r 1 +n∗p∗( − p ∗ ( ) + ∗ p^2); 12 ∗ n^2 n^2 n 3

{CVM}]

(*[5] The Simulated Critical and p Values*) = Quiet[Table[CVMtestL[θ, τ, n, t], M]]; b = Table[ [[i]][[1]], {i, 1, M}]; b = Complement[b, {Indeterminate}]; CV = Quantile[Re[b],1 − sl]; PV = N[Length[Select[b, # > TS &]]/M];

(*[6] The Outputs of the Code*) Print["The CVM Test Results for Lognormal Distribution"] Print["The MLE of θ = "] θ Print["The MLE of τ = "] τ Print["The test statistics of the observed data = " ] TS Print["The simulated critical value of the test = " ] CV

229

230

O. Kittaneh

Code III. The Anderson-Darling Test Mathematica Code for Weibull distribution (*[1] The Inputs of the Code data: the observed data, t: the time of censoring, n: the complete sample size, M: number of sets of specimens, sl: the significance level*) data= {} t= n= M= Sl= The MLEs of

and

(*[2] computing the MLEs of

and

of the given data*)

r = Length[data]; α0 =

Log[Log[1 − 0.97]] − Log[Log[1 − 0.17]] ; Log[Quantile[data, 0.97]] − Log[Quantile[data, 0.17]]

1 β0 = ( ∗ Total[dataα0 ])1⁄α0 ; r A[c_, d_] = −(n − r)(t/c)d ∗ Log[t/c] + r + Total[Log[(data/c)d ]] − Total[(data/c)d ∗ Log[(data/c)d ]]; B[c_, d_] = (n − r) ∗ (t/c)d + Total[(data/c)d ] − r; α = FindRoot[{A[c, d] == 0, B[c, d] == 0}, {{d, α0}, {c, β0}}][[1]][[2]]; β = FindRoot[{A[c, d] == 0, B[c, d] == 0}, {{d, α0}, {c, β0}}][[2]][[2]];

(*[3] The Test Statistics of the Observed Data*) F[x_] = 1 − Exp[−(x/β)α]; dataSORT = Sort[data] ; z = F[dataSORT]; p = r/n; r

r

2∗r r r^2 TS = ∑ (2i − 1) ∗ (Log[1 − z[[i]]] − Log[z[[i]]]) /n − 2 ∗ ∑ Log[1 − z[[i]]] + n ∗ ( − ( )^2 − 1) ∗ Log[1 − p] + n n n i=1

i=1

∗ Log[p] − p ∗ n;

(*[4] The Simulated Test Statistics Distribution *) ADtestW[α_, β_, n_, t_] ≔ Block[{UNIFdata, COMdata, COMdataSORT, r, CENdata, αcen, βcen, F1, z, p, AD, A, B}, UNIFdata = RandomReal[{0,1}, n]; 1 ⁄α

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data CENdata = Table[COMdataSORT[[i]], {i, 1, r}]; A[c_, d_] = −(n − r)(t/c)d ∗ Log[t/c] + r + Total [Log[(CENdata/c)d ]] − Total [(CENdata/c)d ∗ Log[(CENdata/c)d ]] ; B[c_, d_] = (n − r) ∗ (t/c)d + Total[(CENdata/c)d ] − r; αcen = FindRoot[{A[c, d] == 0, B[c, d] == 0}, {{d, α}, {c, β}}][[1]][[2]]; βcen = FindRoot[{A[c, d] == 0, B[c, d] == 0}, {{d, α}, {c, β}}][[2]][[2]]; F1[x_] = 1 − Exp[−(x⁄βcen)αcen ]; z = F1[CENdata]; p = r⁄n ; r

r

AD = ∑ (2i − 1) ∗ (Log [1 − z[[i]]] − Log [z[[i]]]) /n − 2 ∗ ∑ Log [1 − z[[i]]] + n ∗ ( i=1

i=1

∗ Log[1 − p + 10−100 ] +

r2 ∗ Log[p] − p ∗ n; n

{AD}]

(*[5] The Simulated Critical and p Values*) = Quiet[Table[ADtestW[α, β, n, t], M]]; b = Table[ [[i]][[1]], {i, 1, M}]; b = Complement[b, {Indeterminate}]; CV = Quantile[Re[b],1 − sl]; PV = N[Length[Select[b, # > TS &]]/M];

(*[6] The Outputs of the Code*) Print["The AD Test Results for Weibull Distribution"] Print["The MLE of α = "] α Print["The MLE of β = "] β Print["The test statistics of the observed data = " ] TS

2∗r r 2 − ( ) − 1) n n

231

232

O. Kittaneh

Code IV. The Anderson-Darling Test Mathematica Code for the lognormal distribution (*[1] The Inputs of the Code data: the observed data, t: the time of censoring, n: the complete sample size, M: number of sets of specimens, sl: the significance level*) data= {} t= n= M= Sl=

(*[2] computing the MLEs of

and

of the given data*)

r = Length[data]; θ0 = N [

τ0 = (

Total[Log[data]] ]; r

Total[(Log[data] − θ0)2 ] ) r

A[m_, s_] = (Log[t] − m) ∗ ( m −

B[m_, s_] =

0.5

;

Total[Log[data]] Total[(Log[data] − m)2 ] )+ − s2 ; r r

Total[Log[data]] n − r + ∗s∗ r r

Log[t] − m ] s − m; Log[t] − m ] 1 − CDF [NormalDistribution[0,1], s PDF [NormalDistribution[0,1],

τ = FindRoot[{A[m, s] == 0, B[m, s] == 0}, {{m, θ0}, {s, τ0}}][[2]][[2]]; θ = FindRoot[{A[m, s] == 0, B[m, s] == 0}, {{m, θ0}, {s, τ0}}][[1]][[2]];

(*[3] The Test Statistics of the Observed Data*) 1 −θ + Log[x] F[x_] = (1 + Erf[ ]); 2 √2τ dataSORT = Sort[data] ; z = F[dataSORT]; p

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data (*[4] The Simulated Test Statistics Distribution *) ADtestL[θ_, τ_, n_, t_] ≔ Block[{UNIFdata, COMdata, COMdataSORT, r, CENdata, θcen, τcen, F1, z, p, AD}, UNIFdata = RandomReal[0,1, n]; COMdata = Exp[√2 ∗ τ ∗ InverseErf[2 ∗ UNIFdata − 1] + θ]; COMdataSORT = Sort[COMdata]; r = Count[COMdataSORT, u_ /; u ≤ t]; CENdata = Table[COMdataSORT[[i]], {i, 1, r}]; A[m_, s_] = (Log[t] − m) ∗ ( m −

B[m_, s_] =

Total[Log[CENdata]] Total[(Log[CENdata] − m)2 ] )+ − s2 ; r r

Total[Log[CENdata]] n − r + ∗s∗ r r

Log[t] − m ] s − m; Log[t] − m 1 − CDF [NormalDistribution[0,1], ] s PDF [NormalDistribution[0,1],

θcen = FindRoot[{A[m, s] == 0, B[m, s] == 0}, {{m, θ}, {s, τ}}][[1]][[2]]; τcen = FindRoot[{A[m, s] == 0, B[m, s] == 0}, {{m, θ}, {s, τ}}][[2]][[2]]; 1 −θcen + Log[x] F1[x_] = ( 1 + Erf [ ]) ; 2 √2τcen z = F1[CENdata]; p = r⁄n ; r

r

AD = ∑ (2i − 1) ∗ (Log [1 − z[[i]]] − Log [z[[i]]]) /n − 2 ∗ ∑ Log [1 − z[[i]]] + n ∗ ( i=1

i=1

∗ Log[1 − p + 10−100 ] +

r2 ∗ Log[p] − p ∗ n; n

{AD}]

(*[5] The Simulated Critical and p Values*) = Quiet[Table[ADtestL[θ, τ, n, t], M]]; b = Table[ [[i]][[1]], {i, 1, M}]; b = Complement[b, {Indeterminate}]; CV = Quantile[Re[b],1 − sl]; PV = N[Length[Select[b, # > TS &]]/M];

(*[6] The Outputs of the Code*) Print["The AD Test Results for Lognormal Distribution"] Print["The MLE of θ = "] θ

2∗r r 2 − ( ) − 1) n n

233

234

4

O. Kittaneh

Illustrative Examples

In this section, the codes presented in Sect. 3 are applied to several examples extracted from literature. The examples illustrate the goodness of fit of real data sets to Weibull and lognormal distributions. The data sets consist of lives of electrical devises that are used in energy systems. Example 4.1: (Solar Cells) Our first example is on the life of concentrated solar cells with experimental life data adapted from (Espinet-González et al. 2014). In (Espinet-González et al. 2014), a sample of 45 commercial concentrators latticematched GaInP/GaInAs/Ge cells were equally segregated and exposed to three temperature levels; T1 : 164 ◦ C (437 K), T2 : 126 ◦ C (399 K), and T3 : 119 ◦ C (392 K). The time a cell was able to endure that temperature was recorded, reflecting the cell lifetime. The operation was replicated by injecting current in the darkness, which is equivalent to the photogenerated current by cells subjected to the actual field irradiance of 820X. At the lowest temperature intensity T3 , the test took a long time to cause failures and the experiment was terminated after the failure of the ninth cell. This segment of cells is considered censored as some are still working, and thus their exact lifetimes are unknown, whereas the two higher temperature levels T1 and T2 resulted in the failure of all cells. The failure times of the cells are not explicitly reported in (Espinet-González et al. (2014), where the authors present them through Weibull probability plots to graphically demonstrate the compliance of the data to this distribution. However, graphical methods are considered subjective and analytical methods are needed in order to achieve a comprehensive analysis. Here we focus on the third stress T3 as the resulting data is censored. The failure times are extracted from Fig. 7 of (Espinet-González et al. 2014) and listed in Table 1. Since the maximum failure time is 3515 h, it might be reasonable to assume that the test was terminated after 3600 h. Now let us test the compliance of the data to the two models considered in this work using the codes of CVM and AD tests for censored data introduced in Sect. 3. In order to run the codes, users must insert the data (data), censoring time (t), the complete sample size (n), the number of simulations (M), and the significance level (sl). These inputs must look like this data={1300, 1500, 1850, 1950, 2490, 2490, 3280, 3510, 3515}; t=3600; n=15; M=10000; sl=0.05;

After running the two codes, the outputs will be as shown in Tables 2 and 3:

Table 1 The failure times, in hours, of the solar cells of Example 4.1 under stress level T3 1300 2490

1500 3280

1850 3510

1950 3515

2490

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data

235

Table 2 The CVM test results for Weibull and lognormal distributions of the solar cells life data provided in Table 1 Code I

Code II

The Cramer-von Mises

The Cramer-von Mises

Test Results for Weibull Distribution

Test Results for Lognormal Distribution

The MLE of α=

The MLE of θ=

2.60231

8.06732

The MLE of β=

The MLE of τ=

3745.84

0.517784

The test statistics of the observed

The test statistics of the observed

data=

data=

0.025052

0.0245141

The simulated critical value of the

The simulated critical value of the

test=

test=

0. 0697111

0.0674172

The simulated p-value=

The simulated p-value=

0. 4482

0.4511

Decision: 'Accept'

Decision: 'Accept'

Table 3 The AD test results for Weibull and lognormal distributions of the solar cells life data provided in Table 1 Code III

Code IV

The Anderson-Darling

The Anderson-Darling

Test Results for Weibull Distribution

Test Results for Lognormal Distribution

The MLE of α=

The MLE of θ=

2.60231

8.06732

The MLE of β=

The MLE of τ=

3745.84

0.517784

The test statistics of the observed

The test statistics of the observed

data=

data=

0.147656

0.136188

The simulated critical value of the

The simulated critical value of the

test=

test=

0. 420839

0. 382118

The simulated p-value=

The simulated p-value=

0. 5293

0. 5579

Decision: 'Accept'

Decision: 'Accept'

236

O. Kittaneh

It can be seen from the outputs of both tests that the two distributions well fit the experimental data but with slight preference to the lognormal distribution as it achieves higher p-values. The authors of (Espinet-González et al. 2014) adopted, based on a graphical approach, the Weibull distribution as the best fitting model for their data, whereas we find here from the calculations above that the lognormal is at least better than Weibull. Practitioners must be careful when using the codes or builtin functions available in the market when testing censored samples. For example, the MATLAB function of the CVM (cmtest) for Weibull distribution demands inserting the estimates of its parameters, the scale β, and shape α parameters. From the above output, β = 3745.84 and α = 2.60231. When implementing the CVM test of MATLAB as follows >> data={1300, 1500, 1850 , 1950, 2490, 2490, 3280, 3510, 3515}; >> Beta=3745.84; Alpha=2.60231; >>[H,PV,TS,CV]=cmtest(data’,’alpha’,0.05,’CDF’,[data’, wblcdf(data’,Beta,Alpha)])

the result will be as follows H=1

P=0.0420

TS=0.4811

CV=0.4534

H is 1 means that the test rejects the null hypothesis that the data follows the distribution under consideration, which is Weibull in this case, and 0 otherwise. PV, TS and CV are the p-value, the test statistic and the critical value of the test, respectively. This example shows the limitations of this command on MATLAB as it strongly rejects the data to follow the Weibull distribution though this distribution properly fits the data as clear from the probability plot in Fig. 1. A probability plots consists of a straight line that represents the theoretical distribution and points representing the experimental data. The data points fall on or close to the straight line of the Weibull distribution indicating its appropriateness to the data (Nelson 2009). On the other hand, if one tries to find the maximum likelihood estimates of the given data using the built-in command “wblfit” of MATLAB, the result will be β = 2716.8 and α = 3.4000, which are different but comparable to the correct estimates mentioned above, β = 3745.84 and α = 2.60231. The reason for having different results is that the wblfit considers the data as a complete sample and not censored. Now, when we run, using β = 2716.8 and α = 3.4000, the cmtest of MATLAB as follows >> data={1300, 1500, 1850 , 1950, 2490, 2490, 3280, 3510, 3515}; >> Beta=2716.8; Alpha=3.4000; >>[H,P,TS,CV]=cmtest(data’,’alpha’,0.05,’CDF’,[data’, wblcdf(data’,Beta,Alpha)])

we get H=0

P=0.8400

TS=0.0579

CV=0.4534

The new p-value is now p = 0.8400 (strongly accept), which is too optimistic, whereas the previous p-value is 0.0420 (strongly reject). This big difference between the two decisions demonstrates the shortcomings of the command and the mistake that practitioners would commit.

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data

237

Fig. 1 The Weibull probability plot of the data in Table 1

Fig. 2 The lognormal probability plot of the data in Table 1

The same wrong result can appear on Mathematica when using the builtin command CramerVonMisesTest[data, WeibullDistribution[2.60231, 3745.84]] producing a p-value of 0.0444 declaring the rejection of the Weibull distribution, and this is completely wrong. Figure 2 depicts the lognormal probability plot of the given data. The plot of lognormal distribution is very similar to that of Weibull, but the former has a clear better fit on the left tail of the data. The two figures show that the two distributions are suitable models to the given data with preference to lognormal distribution in complete agreement with the goodness-of-fit results presented in Tables 2 and 3.

238

O. Kittaneh

Example 4.2: (Lithium-Ion Batteries) In Mouais et al. (2021)), the authors consider the problem of choosing the best life model, Weibull or lognormal, of commercial lithium-ion batteries based on experimental data published in (Harris et al. 2017). After using four goodness-of-fit tests and the efficiency of censored sample indicator (Kittaneh et al. 2021b), they conclude, based on the usual asymptotic formula of CVM, that the lognormal outperforms Weibull. The experimental data consists of failure times, measured in cycles, of 20 batteries out of 24 in total after exposing them to high stress, where the batteries are cycled at 25 ◦ C with an Arbin BT2000. Each of them is charged in a constant current, constant voltage mode at 1C (4.4 A) constant current up to 4.35 V until the current decreases below C/40. Finally, the batteries are discharged at 10C (44 A) constant current until the terminal voltage dropped to 3 V. The test is terminated after 593 cycles of continuous charge and discharge, with only four survivors out of 24 batteries, and thus t = 593 can be considered as the point of censoring. Table 4 depicts the cycles by which the 20 failures happen. Figures 3 and 4 depicts the probability plots of the data shown in Table 4 using the two models considered in this chapter. It is clear from the two plots that the Weibull and lognormal properly fit the observed data. Let us now run the two codes and compare the graphical observation with the analytical results, where the analytical approaches depend on the p-value of

Table 4 The failure cycles of the 20 batteries of Example 4.2 255 341 449 518

301 379 475 537

326 408 497 541

Fig. 3 The Weibull probability plot of the data in Table 4

338 409 509 541

340 430 515 560

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data

239

Fig. 4 The lognormal probability plot of the data in Table 4

each distribution. As in the first example, the user must insert the following inputs to run the codes data={255, 301, 326, 338, 340, 341, 379, 408, 409, 430, 449, 475, 497, 509, 515, 518, 537, 541, 541, 560}; t=593; n=24; M=10000; sl=0.05;

and the outputs for the two distributions will be as shown in Tables 5 and 6: By looking at the p-values of both tests for the two distributions, we can see that CVM prefers the Weibull, whereas AD prefers the lognormal. This exactly agrees with the probability plots of the two distributions in Figs. 3 and 4. In fact, as mentioned in the introduction, AD test gives more weight to the tails, whereas CVM focuses more on the center. We can notice from the two figures that the lognormal distribution catches more points on its left tail, whereas the Weibull better controls the center of the data. In this case, more goodness-of-fit tests are needed to give the final decision, perhaps the Chi-square (Greenwood and Nikulin 1996) or Lilliefors (Lilliefors 1967) goodness-of-fit tests as discussed in (Mouais et al. 2021), which show that lognormal is a better model to the experimental data. In (Mouais et al. 2021), the authors use simulation to generate the critical and p values of the CVM. Their results are different from what is presented in Table 5 because they consider the data as a complete sample although they deal with the data as censored when estimating the two distributions’ parameters. Example 4.3: (Solid-state Lighting) Solid-state lighting is a promising technology to save energy, reduce cost, transmit data, connect lighting with other building systems, and essentially revolutionize our entire lighting infrastructure. The department of energy of the USA estimates that solid-state lighting luminaire

240

O. Kittaneh

Table 5 The CVM test results for Weibull and lognormal distributions of the Lithium-Ion Batteries life data provided in Table 4 Code I

Code II

The Cramer-von Mises

The Cramer-von Mises

Test Results for Weibull Distribution

Test Results for Lognormal Distribution

The MLE of α=

The MLE of θ=

5.21824

6.12912

The MLE of β=

The MLE of τ=

516.879

0.279582

The test statistics of the observed

The test statistics of the observed

data=

data=

0.0648943

0.0439397

The simulated critical value of the

The simulated critical value of the

test=

test=

0. 155219

0. 10126

The simulated p-value=

The simulated p-value=

0. 4646

0. 4263

Decision: 'Accept'

Decision: 'Accept'

is expected to save energy cost by $ 250 billion over next 20 years and avoid 1800 million metric tons of CO2 (LED Systems Reliability Consortium and U.S. Department of Energy 2013). Reliability remains one of the challenges, hindering further proliferation of this technology, and there is a crucial need for lighting industry and research centers to understand the durability and failure models of the solid-state lighting luminaires, and to develop precise probability distributions of the failures. In this example, we will apply the two codes in testing the compliance of the censored solid-state lighting (SSL) luminaire experimental data of the hammer test of (LED Systems Reliability Consortium and U.S. Department of Energy 2013). The hammer test described in LED Systems Reliability Consortium and U.S. Department of Energy (2013)) is an accelerated life test applied to 17 different commercial luminaires with test duration of 1470 h. 12 out of 17 luminaires failed before that time as five units still operating and their failure times are reported in Table 7. In (Kittaneh and Majid 2019), the authors give a brief but convenient description of the hammer test and conduct a comparison between the censored samples of Weibull and lognormal distributions based on the efficiency measure (Kittaneh et al. 2021b). However, none of these works conducted analytical or even graphical goodness-of-fit tests to choose the most preferable models among the two proposed. Just like Examples 4.1 and 4.2, the following are the inputs and the outputs of the two tests.

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data

241

Table 6 The AD test results for Weibull and lognormal distributions of the Lithium-Ion Batteries life data provided in Table 4 Code III

Code IV

The Anderson-Darling

The Anderson-Darling

Test Results for Weibull Distribution

Test Results for Lognormal Distribution

The MLE of α=

The MLE of θ=

5.21824

6.12912

The MLE of β=

The MLE of τ=

516.879

0.279582

The test statistics of the observed

The test statistics of the observed

data=

data=

0.418179

0.233322

The simulated critical value of the

The simulated critical value of the

test=

test=

0. 810849

0. 546034

The simulated p-value=

The simulated p-value=

0. 4666

0. 4852

Decision: 'Accept'

Decision: 'Accept'

Table 7 The failure times, in hours, of the 12 luminaires described in Example 4.3

293 586 969

293 754 1176

336 800

456 888

547 926

Inputs: data={293,293,336,456,547,586,754,800,888,926,969,1176}; t=1470; n=17; M=10000; sl=0.05;

Outputs (Tables 8 and 9): The two codes suggest that the best model is lognormal as it achieves the highest p-values for the two tests with significant difference. These results agree with the probability plots of the two distributions as illustrated in Figs. 5 and 6, and with the conclusion of (Kittaneh and Majid 2019).

5

Conclusions and Suggestions

This chapter has introduced Mathematica codes of two of the most powerful goodness-of-fit tests for censored data, the Cramer-von Mises (CVM) and Anderson-Darling (AD), for two of the most fundamental probability distributions

242

O. Kittaneh

Table 8 The CVM test results for Weibull and lognormal distributions of the Solid State life data provided in Table 7 Code I

Code II

The Cramer-von Mises

The Cramer-von Mises

Test Results for Weibull Distribution

Test Results for Lognormal Distribution

The MLE of α=

The MLE of θ=

1.68709

6.79907

The MLE of β=

The MLE of τ=

1198.55

0.753565

The test statistics of the observed

The test statistics of the observed

data=

data=

0.0418625

0.0196125

The simulated critical value of the

The simulated critical value of the

test=

test=

0. 0987321

0. 089893

The simulated p-value=

The simulated p-value=

0. 4274

0. 8195

Decision: 'Accept'

Decision: 'Accept'

Table 9 The AD test results for Weibull and lognormal distributions of the Solid State life data provided in Table 7 Code III

Code IV

The Anderson-Darling

The Anderson-Darling

Test Results for Weibull Distribution

Test Results for Lognormal Distribution

The MLE of α=

The MLE of θ=

1.68709

6.79907

The MLE of β=

The MLE of τ=

1198.55

0.753565

The test statistics of the observed

The test statistics of the observed

data=

data=

0.243018

0.141242

The simulated critical value of the

The simulated critical value of the

test=

test=

0. 555128

0. 494018

The simulated p-value=

The simulated p-value=

0. 4675

0. 7479

Decision: 'Accept'

Decision: 'Accept'

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data

243

Fig. 5 The Weibull probability plot of the data in Table 7

Fig. 6 The lognormal probability plot of the data in Table 7

in reliability studies, Weibull and lognormal. Such codes are crucial for practitioners and researchers in the field of reliability or related fields. The reason is that most of the experimental data are incomplete, and unfortunately, the available tests with built-in functions on most of the ready programs are only valid for complete data like MATLAB and Mathematica. The codes also provide the maximum likelihood estimates from censored data, which are not available on both packages. The type of censoring considered in this chapter is type I right censoring, the most commonly used type in applications. The codes used Monte Carlo simulations and designed on the basis of solid mathematical formulas published in reliable venues. The

244

O. Kittaneh

codes automatically provide the test statistics, critical value, and p-value of the test. The codes have been applied to test real data sets of lifetimes of energy and engineering items that were published before. The main disadvantage of the codes is that they are time consuming. When using a computer with normal speed, for Weibull distribution case, both CVM and AD codes need around 0.5 min, when the number of simulations is 5000 for samples of size 10, to 5 min, for 20,000 simulations with sample size 30, whereas for the lognormal distribution, the codes take less than these times.

References T. Anderson, D. Darling, A test of goodness of fit. J. Am. Stat. Assoc. 49(268), 765–769 (1954) A.C. Cohen, Truncated and Censored Samples (Dekker, New York, 1991) H. Cramér, On the composition of elementary errors. Scand. Actuar. J. 1928(1), 141–180 (1928) R.B. D’Agostino, M.A. Stephens, Goodness-of-Fit Techniques (Marcel Dekker, New York, 1986) A.K. Dey, D.K. Kundu, Discriminating between the Weibull and log-Normal distributions for typeII censored data. Statistics 46(2), 197–214 (2012) S.D. Dubey, Some percentile estimators for Weibull parameters. Technometrics 9, 119–129 (1967) P. Espinet-González, C. Algora, N. Núñez, V. Orlando, M. Vázquez, J. Bautista, K. Araki, Temperature accelerated life test on commercial concentrator III-V triple-junction solar cells and reliability analysis as a function of the operating temperature. Prog. Photovolt. Res. Appl. 23(5), 559–569 (2014) P.E. Greenwood, M.S. Nikulin, A Guide to chi-Squared Testing (John Wiley & Sons, New York, 1996) S.J. Harris, D.J. Harris, C. Li, Failure statistics for commercial lithium-ion batteries: A study of 24 pouch cells. J. Power Sources 342, 589–597 (2017) C.M. Jarque, A.K. Bera, A test for normality of observations and regression residuals. Int. Stat. Rev. 55, 163–172 (1987) J.S. Kim, B.J. Yum, Selection between Weibull and log-normal distributions: A comparative simulation study. Comput. Stat. Data Anal. 53, 477–485 (2008) O.A. Kittaneh, M.A. Majid, Comparison of two-lifetime models of solid-state lighting based on sup-entropy. Heliyon 5(10), e02551 (2019) O.A. Kittaneh, S. Helal, H. Almorad, H.A. Bayoud, G. Abufoudeh, M.A. Majid, Preferable parametric model for the lifetime of the organic light-emitting diode under accelerated current stress tests. IEEE Trans. Electron Dev. 68(9), 4478–4484 (2021a) O. Kittaneh, H. Almorad, S. Helal, M.A. Majid, On the efficiency of type I censored samples. IMA J. Math. Control. Inf. 38(2), 743–753 (2021b) LED Systems Reliability Consortium and U.S. Department of Energy, Hammer testing findings for solid-state lighting luminaires. RTI Project Number 0213159, 002 (2013) H.W. Lilliefors, On the Kolmogorov-Smirnov test for normality with mean and variance unknown. J. Am. Stat. Assoc. 62(318), 399–402 (1967) F. Massey, The Kolmogorov-Smirnov test for goodness of fit. J. Am. Stat. Assoc. 46(253), 68–78 (1951) T. Mouais, O.A. Kittaneh, M.A. Majid, Choosing the best lifetime model for commercial lithiumion batteries. J. Energy Storage 41, 102827 (2021) W. Nelson, Graphical analysis of accelerated life test data with the inverse power law model. IEEE Trans. Reliab. 21(1), 2–11 (1972) W. Nelson, Applied Life Data Analysis (Wiley, Hoboken, 2009) A.N. Pettitt, M.A. Stephens, Modified Cramer-von Mises statistics for censored data. Biometrika 63, 291–298 (1976)

Powerful Mathematica Codes for Goodness-of-Fit Tests for Censored Data

245

H. Schneider, Failure-censored variables-sampling plans for log-normal and Weibull distributions. Technometrics 31(2), 199–206 (1989) J. Zhang, F. Liu, Y. Liu, H. Wu, W. Wu, A. Zhou, A study of accelerated life test of white OLED based on maximum likelihood estimation using lognormal distribution. IEEE Trans. Electron Dev. 59(12), 3401–3404 (2012) J. Zhang et al., Constant-stress accelerated life test of white organic light-emitting diode based on least square method under Weibull distribution. J. Inform. Disp. 15(2), 71–75 (2014)

Bayesian Network for Composite Power Systems Using Hybrid Mutual Information Measure Tahereh Daemi, Mohammad Reza Salehizadeh, and Miadreza Shafie-khah

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Introducing the HMI Measure and Using it to Construct the BN Associated with Composite Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 The BN Construction Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Mutual Information and Normalized Mutual Information Measures . . . . . . . . . . . . 2.3 A New Hybrid Dependency Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Component Ranking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

248 250 250 253 255 256 257 263 264

Abstract

The development of effective reliability based evaluation approaches for assessment of power system component importance is very crucial in the planning and operational decision-making processes of power systems. Bayesian network (BN) is one of the most powerful tools that have been used for this purpose. Generally, a BN may be constructed based on expert beliefs, casual effect, or learning methods. In this chapter, as a contribution to the previous literature, a

T. Daemi () Department of Electrical Engineering, Yazd Branch, Islamic Azad University, Yazd, Iran e-mail: [email protected] M. R. Salehizadeh Department of Electrical Engineering, Marvdasht Branch, Islamic Azad University, Marvdasht, Iran e-mail: [email protected] M. Shafie-khah School of Technology and Innovations, University of Vaasa, Vaasa, Finland © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_138

247

248

T. Daemi et al.

new learning-based hybrid mutual information-oriented measure is developed for constructing the BN model for a composite power system (CPS) with emphasis on the involvement of the transmission components. In the previous literature, because of the lower failure probability of transmission components compared to generating units, transmission components have not been accurately involved in the BN model. The presented approach is implemented on IEEE 24-bus reliability test system. The analysis shows that the constructed BN of the case study based on the proposed hybrid mutual information measure provides the importance evaluation of transmission system components more precisely. Keywords

Composite power system · Reliability · Transmission · Bayesian network · Normalized mutual information

1

Introduction

Power system reliability issue has received more attention in the recent years (Varnavskiy et al. 2020; Xiang et al. 2020). From the reliability viewpoint, evaluation of the importance measure of power system components is essential in long- and short-term power system decision-making procedures such as, but not limited to, transmission expansion planning, generation expansion planning, maintenance scheduling, and network reinforcement. In recent years, the need for a fast and reliable evaluation technique is receiving more attention due to the higher mutual interactions of different components in the smart grid environment and in the case of cascading events that lead to blackouts (Li et al. 2020). Two approaches are used for assessing the importance of system components: analytical models and simulation-based approaches. Applying the analytical models such as reliability block diagram (RBD), structure function, fault tree analysis (FTA), and Markov modeling for complex systems, such as the CPS, is not effective nor practical. Among simulation-based approaches, Monte Carlo (MC) method has been frequently used in the previous literature (Wendai et al. 2004). For importance evaluation of each component via MC simulation, the MC procedure should be repeated for large number of iterations. Hence, the evaluation of the importance of components via MC simulation is time-consuming and difficult to be implemented. To tackle this problem, BN is developed. Generally, a BN is a graphical probabilistic model that has been used successfully to study the probabilistic relations between lots of variables in large-scale systems. A BN represents the conditional dependencies between a variable set via a directed acyclic graph (DAG) known as BN structure and conditional probability distributions assigned to each node of DAG known as BN parameters. In BN structure, if an edge from node X to node Y exists, X is called as the parent of Y (Neapolitan 2004). BNs have been applied for various areas of power system studies such as fault diagnosis of three-phase inverters (Cai et al. 2017), reliability assessment of grid-connected photovoltaic

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . .

249

systems with intermittent faults (Cai et al. 2015), incorporating of protection system failures in reliability studies (Eliassi et al. 2015a), and agent-based modeling of electrical energy markets (Dehghanpour et al. 2016). In Buchanan et al. (2019), the BN has been used for assessing system resilience and is applied to study an interdependent electrical infrastructure system. In Ren et al. (2020), a BN-based unified model has been presented for the performance and reliability study of the radial multi-microgrids. Two steps should be taken to design an appropriate BN: (a) constructing a proper structure for the BN and (b) assigning precise parameters to the corresponding nodes. In Huo et al. (2002), the BN of the CPS is constructed based on the physical topology of the system and the corresponding fault tree, or the minimal cut sets. This approach was applied to a small CPS having a simple fault tree. In Eliassi et al. (2015b), an approach is employed to determine the minimal cut sets, and then they are used to construct the BN. Since using cut sets for reliability analysis of systems with multi-state components is not a simple task, implementation of this method in CPS containing multi-state components such as derated generators is not practical for large power systems. Also, incorporation of protection system failures in reliability assessment of CPS, proposed in Eliassi et al. (2015a), suffers the same difficulty. Apart from time-consuming problem, in Eliassi et al. (2015a, b), the cut sets of the system are determined in condition of fixed load, while with varying the system load, the cut sets may be changed. In Ebrahimi and Daemi (2009), a BN is constructed for a CPS based on a learning approach. However, due to its generality of the employed learning algorithm, a relatively large burden of computation is required. Also in the final structure of the BN, the transmission components do not exist. In Daemi et al. (2012), to construct the BN, a novel learning-based approach has been adopted for a CPS. In Daemi et al. (2012), an initial structure is considered for the BN based on the causal relations, and then it is modified using the mutual information measures between variables. In Ebrahimi and Daemi (2010), it has been shown that due to the low failure probability of transmission system components, to reach the BN structure of the electric transmission system, a very large training data set is required that the structure learning becomes difficult. To overcome this problem, in Ebrahimi and Daemi (2010) the importance sampling (IS) has been used to provide a weighted training data set with lower number of data. Afterwards, to construct the BN, the MI-based approach presented in Daemi et al. (2012) with weighted training data is employed. In Ebrahimi and Daemi (2010), it is shown that by utilizing IS, the BN associated with the only transmission system can be obtained effectively. As the motivation of this research, it is shown in this study that the structure of BN of the power system consisting of both the generation and transmission systems based on MI measure is very sensitive to the threshold value selected for comparing MI values with it. Hence, the BN obtained based on MI measure may not be accurate enough. As it will be shown in this study, the cause of inaccuracy is due to considerable difference between forced outage rate (FOR) of generation and transmission components. Although transmission system components compared with generating units almost have low failure probabilities, their outages may have constructive effects on

250

T. Daemi et al.

reliability of system. In this regard, transmission components need to be accurately involved in the designed BN. In this study, as a major contribution, a new dependency measure named hybrid mutual information (HMI) measure is developed that is used in the learning-based approach of constructing the BN associated with a CPS which is consisting both transmission and generation sectors. Since transmission components’ FOR is low, the number of samples with transmission component outage is low. To have a richer data set containing the failure of transmission components and with lower number of data, IS is deployed in data generation of MC simulation. Generally, the structure learning of a BN is an NP-hard problem, and different algorithms are proposed for it. However, the special learning-based approach proposed in Daemi et al. (2012) simplify constructing the BN associated with power system, and it does not require much computational effort. It can be applicable for power systems with complex topologies, and it is not required for system cut sets be identified. Also, the multi-state components and variation of load can be considered easily, and it can be developed to consider other items in reliability studies of power systems. However, by introducing the HMI measure, while maintaining the advantages of constructing the BN from the learning-based approach proposed in Daemi et al. (2012), a precise analysis of transmission components simultaneous with generation system components is provided and the effectiveness of the proposed approach is shown in the modified IEEE 24-bus Reliability Test System. Table 1 presents a taxonomy of a few studies related to BN construction for a CPS’s reliability studies.

2

Introducing the HMI Measure and Using it to Construct the BN Associated with Composite Power System

2.1

The BN Construction Procedure

The procedure for constructing the BN is based on an initial structure. Afterward, it is modified using the BN’s possibility of learning from data. MC simulation is utilized for training data set production in a non-sequential way. The procedure to create the BN and use it for reliability analysis of the CPS is briefly presented in Fig. 1. In the following, at first the approach to produce training data is reviewed, and then the general approach to create the BN is briefly presented (Daemi et al. 2012; Ebrahimi and Daemi 2010).

2.1.1 Training Data Generation As mentioned in the previous sections, to create the BN, we need a training data set. The training data includes state vectors as S = G1 , . . . , Gng , L1 , . . . , Lnl , Li−j , Lk−n , . . . , Bus1 , . . . , Busm , SF

(Daemi et al. 2012) (Ebrahimi and Daemi 2010) Proposed approach

(Eliassi et al. 2015b) (Ebrahimi and Daemi 2009)

(Ren et al. 2020) (Huo et al. 2002)

Transmission sector ✗

✗

Generation sector

✗

Multi

Multi

Multi

Multi

Single

Single

Multi/single state components Single

Table 1 Taxonomy of studies related to BN construction for a CPS BNconstruction approach Minimal cut-set Minimal cut-set/fault tree Minimal cut-set Typical learning algorithms Learning with MI measure Learning with MI measure Learning with HMI measure

✗

Generation

✗

✗

Transmission

Possibility of component outage evaluation

✗

✗

✗

Scalability ✗

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . . 251

252 Fig. 1 The procedure to construct the BN with the application for reliability analysis

T. Daemi et al.

Generation of training data using MC simulation and Importance Sampling

Computation of HMI measures between variables based on initial structure considered for the BN

Removing the edges between nodes with low HMI measures from the initial structure of the BN to reach the final structure of BN

Inference from the BN to obtain different assessment

where Gi points to the ith set of generation units located on the same bus which are similar. The generating units related to Gi variable are considered as a derated generation unit, and the value is equal to the number of its related generation units in the failure state. Li is related to the state of the transmission line or power transformer i, and its value is equal to 1 if it is in failure state; otherwise, it is zero. Li − j denotes parallel transmission components i and j, having the same specification, and its value equals the number of corresponding components in outage state. SF variable is devoted to represent the system state in supplying load. Its value equals to zero, unless the loss of load occurs that it will be one. Variable Busk is similar to SF; however, it is devoted to the load point k. The value of Busk is equal to one, if the loss of load occurs in Busk ; otherwise, it is equal to zero. ng and nl are the numbers of generating unit sets and individual transmission system components, respectively. Also, m is the number of load buses. MC simulation is used to generate training data set. To evaluate and specify the sampled states and the values of Busk and SF variables in vector S, DC optimal power flow (OPF) is used. In the OPF model, the branch flow constraints and real power generation limitations are considered. As corrective actions to establish operational constraints, generation rescheduling and load shedding are applied. To perform a more accurate analysis of the transmission system, as suggested in Ebrahimi and Daemi (2010), the IS scheme is used in the data generation stage. Using IS results in a weighted training data set.

2.1.2 BN Construction The components outage might cause load loss in some load points. Load loss in each bus means the system load loss. So, the initial BN structure based on the independence assumption of the components’ outages and the casual relationships between variables is considered as shown in Fig. 2. However, it is clear that all of the components do not equally affect all of the load points. In this regard, to

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . .

253

Bus nodes G1

Bus1

G2

Bus2

System node

…

…

SF

L

…

Busn

Li-j

…

Components’ nodes

Fig. 2 The initial structure of the BN associated with the CPS

identify more critical relations between load points and components, a dependency measure should be used. After the BN structure is specified, the parameters of BN are determined as pointed in Daemi et al. (2012) and Ebrahimi and Daemi (2010). In the next section, a new dependency measure is proposed to identify the effective links between nodes.

2.2

Mutual Information and Normalized Mutual Information Measures

One of the simple indices for measuring the dependency between random variables is mutual information index. For random variables X and Y with marginal probabilities p(x) and p(y), and joint probabilities p(x, y), MI is defined as MI (X; Y) =

x,y

p (x, y) log

p (x, y) p (x) p (y)

(1)

MI measure is always non-negative. If X and Y are independent, p(X, Y) = p(X).p(Y), and therefore, MI measure is equal to zero. The higher value of MI indicates the stronger dependency between X and Y variables. The mutual information can be represented as the following equation in terms of variable entropies (Yao and Karmeshu 2003):

254

T. Daemi et al.

MI (X, Y) = H (X) − H (X|Y) = H (Y) − H (Y|X)

(2)

where H(X) is the entropy of variable X that measures the uncertainty on variable X and is defined as H (X) = − P (x) logP (x) (3) x

The mutual information on the basis of Eq. (2) indicates the uncertainty reduction about X (or Y) through observing Y (or X). The conditional entropy H(X|Y) is defined as H (X|Y) = P (y) H (X|y) = y

=−

P (y)

y

=−

P (x|y) logP (x|y)

(4)

x

P (x, y) logP (x|y)

x,y

Joint entropy of random variables X and Y is defined as H (X, Y) = −

P (x, y) logP (x, y)

(5)

x,y

Also, the MI measure is represented as MI (X, Y) = H (X) + H (Y) − H (X, Y)

(6)

The relation between MI measures and entropies is shown in Fig. 3. Eq. (6) implies that 0 ≤ MI (X < Y) ≤ min (H (X) , H (Y)) ≤

(7)

H(X) H(Y)

H(X|Y)

MI(X,Y)

Fig. 3 Relations between MI measure and entropies

1 (H (X) + H (Y)) 2

H(Y|X)

H(X,Y)

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . .

255

MI has been normalized in Kojadinovic (2005) in a way that the its maximum is 1. Four types of normalized NI are as follows: NImin

MI (X, Y) min (H (X) , H (Y))

(8)

2MI (X, Y) H (X) + H (Y)

(9)

NIx (X, Y) =

MI (X, Y) H (X)

(10)

NIy (X, Y) =

MI (X, Y) H (Y)

(11)

xy (X, Y)

=

NIxy (X, Y) =

2.3

A New Hybrid Dependency Measure

Although it is not expected that the BN structure consists of all components, it is expected that the BN comprises the most critical components. In the MI-based approach proposed for constructing the BN, MI measure is used to measure the dependency between components outage events and load loss in different load points. After the computation of MI measures, they are compared with a threshold value to decide about the existence of edges in the BN. Based on inequality (7), MI measure is always less than variable entropies. The variation curve of H(X) for variable X having a binomial distribution with parameter p, for different values of p is shown in Fig. 4.

0.35 0.3

H (X )

0.25 0.2 0.15 0.1 0.05 0

0

0.2

0.6

0.4

0.8

p

Fig. 4 The curve of H(X), X having binomial distribution with parameter p

1

256

T. Daemi et al.

It is observed that for p = 0.5, H(X) is maximum. Also, for values near to zero or one, H(X) is very small and near to zero. Therefore, for components such as transmission system components that their forced outage rate (FOR) values are very low, their entropies are very small, and so, even if their outages cause to certainly loss of load in some load points, the MI measure will be very small, and depending on the chosen threshold value, such components may or not appear in the BN. To ensure the existence of the edges between load points and components having very low FOR values that their outages affect the reliability of load points considerably, it seems that using the MI measure is not effective. Applying Eq. (10) can solve the problem, since in that equation the normalization is done based on entropy of components. On the other hand, for components with high FOR values, such as generating units, corresponding values of NIx are low. Therefore, a hybrid mutual information (HMI) to measure the dependency between components Xi and load point Busj is proposed as MI Xi , Busj NIx Xi , Busj , HMI Xi , Busj = max max MIj max NIxj

(12)

In Eq. (12), MIj and NIxj are respectively the vectors of MI and NI measures between Busj and all of the system components. Since the variation range of MI and NI measures is different, to improve their comparison, they are normalized on the basis of their maximum values on each bus. The value of HMI measure is limited to [0, 1], so the threshold value can be specified easily (e.g., 0.05).

2.4

Component Ranking

One of the main application of BN in power system is component ranking. A question that may be raised is why the BN is constructed, while the critical components can be identified using the MI, NI, or HMI. It should be noted that the BN provides the possibility of various probabilistic inferences, and the importance of components can be evaluated from different aspects as is done in Daemi et al. (2012), Ebrahimi and Daemi (2010), and Daemi and Ebrahimi (2012a, b). For example, in Daemi et al. (2012), it is shown that the BN can be used to components ranking on the basis of frequency and duration indices. So, having the BN for the power system can be useful. Although these evaluations can be done by other approaches such as MC simulation, they will be very time consuming. In the next section, Birnbaum’s measure (BM) is used for component ranking. BM indicates how system unavailability will change with changes in component unavailability and is defined as BMk =

∂PSF=1 ∂Pxk =1

(13)

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . .

257

where SF denotes system failure event and xk points to the number of component in failure state related to the kth component. So, PSF = 1 indicates the system failure probability and Pxk =1 indicates the failure probability of component k. Considering the conditional probability rule, Eq. (13) can be represented as BMk = P (SF = 1|xk = 1) − P (SF = 1|xk = 0)

(14)

The BM of component k represents the criticality state of system in view of probability with respect to that component. Based on Eq. (11), this measure indicates the importance of the component for system operation from the structural point of view and so sometimes is known as structural importance index (Anders 1983). In this study, with regard to considering the similar components in the same position with one multi-state variable and so with one multi-state node in the BN, the definition of BM will be different. For a multi-state system with multi-state components, the composite importance measures (CIM) are developed (RamirezMarquez and Coit 2005). On this base, for a two-state system with multi-state components, BM measure will be k 1 |P (SF = 1|xk = j) − P (SF = 1)| mk − 1

m

BMCIM = k

(15)

j=1

where mk is the number of state of xk .

2.5

Case Study

In this study to more obviously show the importance and efficiency of the new proposed dependency measure, a moderate size power system as IEEE 24-bus reliability test system (RTS) is selected for the case study. However, to put the transmission section in more stress, the load of the system is increased by 40%, and the capacity of transmission components is decreased to 10%, and then the approach is applied to the modified RTS (MRTS). The MRTS is depicted in Fig. 5 that has 24 buses, 32 generation units, 17 load points, 5 transformers, and 33 transmission lines (Reliability Test System Task Force 1999). As mentioned in section “A New Hybrid Dependency Measure,” similar generating units placed on the same bus are related to one variable Gi in vector S, and they are shown with one node in the BN. The variables Gi (∀i = 1 : ng) for the test system, with their specifications, can be referred to Table 1 of Ebrahimi and Daemi (2009). As an example, in that Table, variable G1 points to six generating units having the capacity of 50 MW and FOR value of 0.01, located on Bus22 . It should be mentioned that the approach to generate data and constructing the BN involving structure and parameter learning is done in MATLAB, and for inference from the BN, GeNIe and SMILE developed

258

T. Daemi et al.

Fig. 5 The single-line diagram of IEEE-RTS

at the University of Pittsburg (The Genie tool for Bayesian networks and influence diagrams) is employed.

2.5.1 Comparison of MI, NI, and HMI Measures Before showing the final structure of BN corresponding to the test system obtained using the HMI measure, it is appropriate to compare the MI, NI, and HMI measures between a load point and components. This is done for Bus8 in Figs. 6 and 7. From Fig. 5, it is observed that transmission lines 11, 12, and 13 connect Bus8 to the system. In this study, the used load shedding policy in contingency states is based on closeness to the component(s) on an outage. So, the outage of lines 11, 12, and 13 can affect the reliability of load point 8, and so it is expected that these lines are parents of node Bus8 in the BN.

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . .

259

0.08

0.0025 MI NI

0.0020

0.07 0.06

MI

0.04 0.0010

0.03 0.02

0.0005

0.01 G1 G3 G5 G7 G9 G11 G13 L1 L3 L5 L7 L9 L11 L13 L15 L17 L19

0.0000

NI

0.05

0.0015

0.00

Fig. 6 MI and NI measure between Bus8 and components

1.2 1.0

HMI

0.8 0.6 0.4 0.2 0.0

G1 G3 G5 G7 G9 G11 G13 L1 L3 L5 L7 L9 L11 L13 L15 L17 L19

Fig. 7 HMI measure between Bus8 and components

Figure 6 shows that the MI measures between these transmission lines and the load point 8 are low, and so the existence of edges between these lines and node Bus8 in the BN is highly affected by the chosen threshold value to compare with MI measures. However, the value of NI measures between transmission lines 11, 12, and 13 is high, and so based on this measure, these lines certainly are chosen as parents of node Bus8 . On the other hand, the NI measures between some variables that have large MI measures are relatively low due to their high FOR values and so their high entropies. The curve of HMI measures between load point 8 and components is shown in Fig. 7. It is observed that the effective components on the reliability of this load point are well identified using this measure. It should be mentioned that the value of MI, NI, and HMI measures of the other components that have not appeared in Figs. 6 and 7 is very low, and so they are not shown.

260

T. Daemi et al.

Fig. 8 MI measures between load point 1–6 with components

In another illustration, the values of MI and HMI measures between buses 1–6 and system components are, respectively, shown in Figs. 8 and 9. This can be done for all load points. However, for more clear view, here this is shown for only six load points. Also, some components wherein their related MI and HMI measures with all load points were very low are not considered in these figures. Based on Fig. 8, it is shown that most of MI measures between load points and transmission system components are very, low and so identification of effective transmission components on load points is highly dependent on the selected threshold value to compare them. However, it is shown that based on Fig. 9, the effective transmission components can be suitably identified based on HMI measures.

2.5.2 Final Structure of BN Associated with MRTS The final structure of the BN obtained using the HMI measure is shown in Fig. 10. Although the accuracy of the obtained BN can be evaluated by different quantitative probabilistic inferences from the BN, the BN structure also involves information, and so it can be evaluated intuitionally. In Table 2, the transmission system components that are parents of each loads point are shown. Referring to Fig. 5 and with regard to load shedding policy, it is observed that parents of transmission system components for load points are expected. It may be thought that it is not required to determine the parents of load points using the HMI measure, and they

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . .

261

Fig. 9 HMI measures between load point 1–6 with components

can be identified based on system topology. However, it should be noted that in this study, the closeness to the component(s) on outage is considered as load shedding policy. However, the other load shedding policies may be used in a power system, and so identification of effective components on load points will not be so easy, generally. The results in Table 2 verify the accuracy of the proposed dependency measure in constructing the BN.

2.5.3 Numerical Evaluation of Component Ranking The ranking of components based on the BM measure is performed. The results are shown in Fig. 11. For computation of BM of each component with mk states, mk MC simulation is run. MC should be repeated for each component which is very time-consuming. While these assessments can be easily done using the obtained BN. The results of MC simulation are also depicted in Fig. 11. As another assessment, the loss of load probability (LOLP) of Bus3 , given some components in outage states, is shown in Fig. 12. To show the accuracy of the results, this analysis is also done using the Monte Carlo simulation. For this purpose, it is required to set the state of each component

262

T. Daemi et al.

Fig. 10 The BN structure associated with MRTS

Table 2 Transmission system components which are parents of different load points Bus No. Transmission system components being buses’ parents

1 L1 L2

2 L1 L4 L5

3 L2 L6 L7

4 L4 L8

6 L3 L9

6 7 8 14 15 L5 L11 L11 L19 L24 L10 L12 L23 L13

16 19 L23 L29 L24 L34–35 L28 L29

20 L34–35 L36–37

individually to failure state and then run MC simulation to obtain the LOLP of buses or system. It is clear that it is time-consuming, but it can be easily performed by inference from the BN. However, the comparison of results obtained from the BN and MC verifies the accuracy of the BN model. In this study, the BN is applied to assess the CPS in view of loss of load probabilities and components ranking. It is noted that the approach is applicable for larger power system. A BN may be generally large. However, it is suitable that the number of parents for each node is not so numerous. For every load point, there are some components that have more effect on their reliability, and so the number of parents of the load point nodes in the BN will be limited. However, when the power system is of large size, the parents’ number of node SF will be large that it will not be suitable. For the large size power system, the nodes associated with load points can be divided to some groups, each group connect to one node and then the new nodes associated with groups connect to SF nodes, and so it can be employed for the large power system.

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . .

263

1

MC BN

BM

0.8 0.6 0.4

0

L11 L5 L10 L27 L7 G5 G13 G12 G11 L23 G9 G10 L18 G14 G3 G7 G4 L36-37 L16 G1 L17 L25-26 L28 L29

0.2

Fig. 11 Components ranking based in Birnbaum’s measure

LOLP

0.12 0.10

MC

0.08

BN

0.06 0.04 0.02 0.00 L6

L2

L7

G11

G13

G7

G6

G10

G12

Fig. 12 Loss of load probability of Bus3 given components outage

3

Conclusion

In this chapter, a new dependency measure was suggested to be used in constructing the BN associated with the power system considering both generation and transmission sections with the aim of achieving the more precise BN, and so more accurate analysis of transmission system components effects on the system reliability. The proposed measure was a combination of mutual information and a special normalized mutual information measure. Using the new proposed measure in the process of special BN structure learning provided the possibility of accurate analysis of the transmission system components while taking advantage of using the BN in detailed reliability evaluation of CPS. As a case study, the proposed measure was applied to the modified IEEE 24-bus test system, although it was applicable for larger CPS. The simulation results showed that using the new dependency measure in constructing the BN could provide a more precise evaluation of transmission system components on system and load points reliability in comparison with the BN constructed based on MI measure. As future work, it is suggested that the state of

264

T. Daemi et al.

load points in the BN is selected based on the measure of curtailed load or energy not supplied to provide more reliability analysis of CPS on the basis of energy indices. Also, the application of the proposed approach to renewable-based power systems, hybrid AC/DC systems, and micro-grids is suggested.

References G.J. Anders, Evaluation of importance and related reliability measures for electric power systems. IEEE Trans. Power Appar. Syst. PAS-102(3), 527–537 (1983) R. Buchanan, N.U. Ibne Hossain, S. Hosseini, R. Jaradat, M. Marufuzzaman, A framework for modeling and assessing system resilience using a Bayesian network: a case study of an interdependent electrical infrastructure system. Int. J. Crit. Infrastruct. Prot. 25, 62–83 (2019) B. Cai, Y. Liu, Y. Ma, L. Huang, Z. Liu, A framework for the reliability evaluation of gridconnected photovoltaic systems in the presence of intermittent faults. Energy 93, 1308–1320 (2015) B. Cai, Y. Zhao, H. Liu, M. Xie, A data-driven fault diagnosis methodology in three-phase inverters for PMSM drive systems. IEEE Trans. Power Electron. 32(7), 5590–5600 (2017) T. Daemi, A. Ebrahimi, Evaluation of components reliability importance measures of electric transmission systems using the Bayesian network. Electr. Power Compon. Syst. 40, 1377–1389 (2012a) T. Daemi, A. Ebrahimi, Detailed reliability assessment of composite power systems considering load variation and weather conditions using the Bayesian Network, International Transactions on Electrical Energy Systems, 2012b T. Daemi, A. Ebrahimi, M. Firuzabad, Constructing the Bayesian network for components reliability importance ranking in composite power systems. Int. J. Electr. Power Energy Syst. 43, 474–480 (2012) K. Dehghanpour, M.H. Nehrir, J.W. Sheppard, N.C. Kelly, Agent-based modeling in electrical energy markets using dynamic Bayesian networks. IEEE Trans. Power Syst. 31(6), 4744–4754 (2016) A. Ebrahimi, T. Daemi, Novel method for constructing the Bayesian network for detailed reliability assessment of power systems, in International Conference on Electric Power and Energy Conversion Systems, Sharjah, UAE, 2009 A. Ebrahimi, T. Daemi, A simple approach to construct the Bayesian network associated with electric transmission systems. Int. Rev. Electr. Eng. 5, 180–184 (2010) M. Eliassi, H. Seifi, M.R. Haghifam, Incorporation of protection system failures into bulk power system reliability assessment by Bayesian networks. IET Gener. Transm. Distrib. 9(11), 1226– 1234 (2015a) M. Eliassi, A.K. Dashtaki, H. Seifi, M.R. Haghifam, C. Singh, Application of Bayesian networks in composite power system reliability assessment and reliability-based analysis. IET Gener. Transm. Distrib. 9, 1755–1764 (2015b) L. Huo, Y. Zhu, G. Fan, Reliability assessment of power systems by Bayesian networks, in International Conference on Power System Technology, In proceeding of the PowerCon, 2002 I. Kojadinovic, On the use of mutual information in data analysis: an overview, in 11th International Symposium on Applied Stochastic Models and Data Analysis, 2005 L. Li, H. Wu, Y. Song, Y. Liu, A state-failure-network method to identify critical components in power systems. Electr. Power Syst. Res. 181, 106192 (2020) R.E. Neapolitan, Learning Bayesian Networks (Pearson Prentice Hall, Upper Saddle River, 2004) J.E. Ramirez-Marquez, D.W. Coit, Composite importance measures for multi-state systems with multi-state components. IEEE Trans. Reliab. 54(3), 517–529 (2005) Reliability Test System Task Force, The IEEE reliability test system-1996. IEEE Trans. Power Syst. 14, 1010–1020 (1999)

Bayesian Network for Composite Power Systems Using Hybrid Mutual. . .

265

Y. Ren et al., A reliability evaluation method for radial multi-microgrid systems considering distribution network transmission capacity. Comput. Ind. Eng. 139, 106145 (2020) The Genie tool for Bayesian networks and influence diagrams, http://www2.sis.pitt.edu/~genie/ K. Varnavskiy, Q. Chen, F. Nepsha, Structure orderliness assessment of grid development to improve the reliability of coal mine external electrical power supply. Electr. Power Syst. Res. 183, 106283 (2020) W. Wendai, J. Loman, P. Vassiliou, Reliability importance of components in a complex system, in Annual Reliability and Maintainability Symposium, 2004, pp. 6–11 Y. Xiang, Y. Wang, Y. Su, W. Sun, Y. Huang, J. Liu, Reliability correlated optimal planning of distribution network with distributed generation. Electr. Power Syst. Res. 186, 106391 (2020) Y.Y. Yao, P. Karmeshu, Information-theoretic measures for knowledge discovery and data mining, in Entropy Measures, Maximum Entropy Principle and Emerging Applications, (Springer, Berlin/Heidelberg, 2003)

Optimal Management of Smart Home Appliances Considering Stochastic Behavior of Wind Turbine Masoud Alilou, Hossein Shayeghi, and Behrouz Tousi

Contents 1 2 3 4

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Home Energy Management System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wind Turbine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Objective Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Electricity Bill . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Peak Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Energy Management Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Numerical Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1 Case 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Case 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

268 269 272 273 274 275 275 276 278 279 282 284 285

Abstract

In this paper, the operating schedule of home appliances and local wind turbine is optimized using a home energy management system. Price-based demand response program is utilized for operational scheduling. The uncertainty of wind speed is considered in the proposed energy management method. To achieve the best plan of devices, technical and economic formulation is developed. Multi-objective dragonfly algorithm is used for optimizing the techno-economic

M. Alilou () · B. Tousi Department of Electrical Engineering, Urmia University, Urmia, Iran e-mail: [email protected]; [email protected] H. Shayeghi Energy Management Research Centre, University of Mohaghegh Ardabili, Ardabil, Iran e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_140

267

268

M. Alilou et al.

objective functions, and then fuzzy setting method is utilized for choosing the best operational schedule of home devices from Pareto front. Numerical simulations illustrate the proper performance of the proposed method in finding the best schedule of home appliances and wind turbine and also improving the considered indices of the smart home.

Keywords

Electricity bill · Multi-objective dragonfly algorithm · Smart home appliances · Uncertainty · Wind turbine

1

Introduction

Home is one of the important parts of human’s life which has been affected by technology in the last years. Modern appliances such as refrigerator, dishwasher, and washing machine help residents of the home to perform household chores faster and easily. Residential consumers would like to control their home appliances automatically; this issue is applicable in smart homes (Li et al. 2021). The smart home is a combination of various home appliances which are related to each other through advanced technologies. Home energy management system (HEMS) has the responsibility for managing appliances based on consumer’s life, market price, and other important subjects of smart home and power network (Struckell et al. 2021). Although the consumption part of a smart home can be controlled automatically and therefore this is suitable for both consumer and operator of the network, the production part of the smart home is also important in becoming the home to a green home. Wind turbine (WT) is an attractive option for supplying energy of smart homes for reasons such as low cost, green energy, no pollutant emission, easy accessibility, renewability, and reduction of fossil fuel consumption (Pohl et al. 2021; Dejamkhooy et al. 2020). Although the tendency to use WT is increased day to day, one of the main disadvantages of these technologies is the uncertainty of their production power based on weather variation. Energy management in the power distribution system is a fascinating topic for researchers. Some articles related to energy management in smart grids and homes have been reviewed in (Gelazanskas and Gamage 2014). The multi-objective demand side management of a smart microgrid was studied in (Shayeghi and Alilou 2021). The economic and environmental indices of the microgrid were considered as the primary objective functions of the proposed demand side management method. The operator of the microgrid can provide the demand of the system using a local renewable and nonrenewable distributed generation units and also the upstream network. The authors of (Barbato et al. 2015) have proposed a method of DSM for reducing the peak demand of residential customers of the power network. The electricity tariff has been considered as a function of the total power demand of

Optimal Management of Smart Home Appliances Considering Stochastic. . .

269

residential users. In (Aghajani and Kalantar 2017), optimization of the capacity of the electric vehicle parking lot has been studied to improve the situation of energy and reserve market of the distribution system in the presence of wind turbines. The hybrid AC/DC microgrid architecture for smart homes has been proposed in (Wang et al. 2018) to improve the used penetration of DG units and to decrease the dependence of the microgrid on the upper network. In another study, the energy management was investigated in a smart microgrid in the presence of wind turbine, photovoltaic panel, and energy storage (Naderi et al. 2022). The considered grid was a practical network in Khalkhal. Iran. The formulated problem for optimal energy management and sizing was solved using a mixed integer linear programming. In (Rastegar et al. 2018), an energy management framework has been proposed for optimizing the operating time of home devices. This optimization has been done in two steps. Firstly, each customer minimizes his payment cost, and secondly, smart distribution company minimizes the deviation of distribution system load and the cost of modifying the desired scheduling of consumers. In (Rastegar et al. 2012), the load commitment framework has been studied to achieve the minimum electricity bill. Although, in the last years, various aspects of smart grids and smart homes have been studied by researchers, the novel point of this study is the simultaneous scheduling of the producer and the consumer devices of a smart home. A stochastic model of the wind turbine is also formulated in the proposed model. Moreover, energy management is applied to the smart home as a multi-objective optimization with considering a technical and an economic index. Therefore, the operating schedule of home appliances and local wind turbine is optimized in this paper. The combination of multi-objective dragonfly algorithm (MODA) and fuzzy setting method is used to solve the technical-economic problem and select the best schedule of home devices.

2

Home Energy Management System

Manually management of home appliances and energy sources based on variable market price and produced power of renewable units is difficult for residential consumers. Therefore, one of the important devices of a smart home is the home energy management system. HEMS can automatically manage the operational time of home appliances and the wind turbine. The HEMS has the capability for monitoring and arranging various devices of the smart home in real time, based on the user’s preference in order to improve the energy efficiency and conserve electricity cost. The home energy management system uses an intelligent algorithm to find the best schedule of home energy devices. This algorithm is the combination of the multi-objective dragonfly algorithm and the fuzzy decision-making method whose general procedure is demonstrated in Fig. 1. The output power of the wind turbine is uncertain because the wind speed has a probabilistic nature. Therefore, the home energy management system has to

270

M. Alilou et al.

User’s preferences Predetermined constraints Mean value and standard deviation of wind speed Market price

Input data

Evaluating the data and Managing the devices

Calculating the stochastic data of wind turbine by LHS and K-means clustering algorithms Applying MODA Evaluating constraints and preferences Applying the fuzzy method

Output data

Starting time of appliances Hourly power of wind turbine Hourly power transmission between smart home and grid Economics data

Fig. 1 The management procedure of the proposed home energy management system

utilize a stochastic model to compute the value of the uncertain parameter. The Weibull probability distribution function is adopted for modeling wind turbine uncertainty. The method for calculating the output power of wind turbine is explained completely in Sect. 3. In the home energy management system, the hybrid method of the Latin hypercube sampling algorithm and K-means clustering algorithm is used to find the stochastic data wind turbine. The Latin hypercube sampling method, which integrates stratified and random sampling, samples layers from the entire distributions of random variables. It can produce more stable and precise estimates in faster runtime than traditional sampling methods such as Monte Carlo sampling method. In the Latin hypercube sampling method, firstly, Ns samples are generated to demonstrate the stochastic nature of the uncertain parameter. Then, the cumulative distribution function of uncertain parameter is divided into Ns intervals with equal probability of 1/Ns . Afterward, a value is randomly selected from each interval. The sampled cumulative probability at interval ith is calculated by Eq. 1 (Mazidi et al. 2014). Pi =

1 Ns

ru +

(i − 1) Ns

(1)

Optimal Management of Smart Home Appliances Considering Stochastic. . .

271

Here, ru is a uniformly distributed random number (0, 1). Now, the sampled value is transformed into the inverse value using the inverse of the distribution function; mathematically, it is shown in Eq. 2. Xi = F −1 (Pi )

(2)

In stochastic programming like the Latin hypercube sampling method, the runtime raises when the number of scenarios increases. The operator of the power network has to decide as fast as possible; so it is better that a practical scenario reduction method is utilized to decrease the number of scenarios of stochastic problems. The reduction algorithms cause to reduce the computational time of simulating a large number of stochastic scenarios. The K-means clustering algorithm is used to reduce the number of scenarios. The main method of this scenario reduction algorithm is to arrange original scenarios of stochastic parameter into clusters; this arrangement is applied according to similarities of produced scenarios. It is worth mentioning that the output of K-mean clustering method is the cluster centroids and the number of scenarios allocated to each cluster. Totally, the flowchart of K-means algorithm is presented in Fig. 2 (Mazidi et al. 2014). Fig. 2 Flowchart of the K-means clustering method for reducing the number of scenarios

Select the proper number of required clusters according to the needs of the specific problem

Define the initial centroid of each cluster randomly

Compute the distance between each original scenario and each cluster centroid

Assign teach original scenario to the closest cluster based on calculated distances

Calculate new centroid of each cluster using the original scenario allocated to each cluster Are there changes in the cluster compositions between two consecutive iterations? No

Yes

272

3

M. Alilou et al.

Wind Turbine

In smart homes, the wind turbine is one of the useful energy sources especially in areas that the suitable level of wind speed is available during 24 hours. Wind turbines which are used in smart homes are usually smaller than wind turbines of grids in both size and capacity. The sample wind turbines that are used in smart homes are shown in Fig. 3. With regard to the stochastic behavior of wind, the wind turbine has also a stochastic performance. So a probabilistic model is considered for calculating the produced power of the wind turbine. Rayleigh probabilistic density function, which is a particular form of Weibull distribution, is utilized to model wind speed. In the Rayleigh function, the shape index is equal to 2. The probability function of wind speed can be calculated by Eq. 3 (Aghajani et al. 2017). f (v) =

K (K−1) −( v )K v e C CK

0≤v≤∞

Fig. 3 The sample wind turbines used in smart homes (Stathopoulos et al. 2018)

(3)

Optimal Management of Smart Home Appliances Considering Stochastic. . .

273

Fig. 4 The relationship between the output power of a wind turbine and the wind speed

Here, K and C demonstrate the shape index and the scale index of Weibull distribution, respectively. The relationship between the output power of a WT and the wind speed is formulated by Eq. 4. In this equation, V is the wind speed at the hub height of the wind unit; Vci , Vco , and Vr are the cut-in wind speed, the cut-out wind speed, and the rated wind speed, respectively; and Pw − rated is rated output power of wind unit. ⎧ ⎪ ⎨0 Pw = Pw−rated × ⎪ ⎩P w−rated

V −Vci Vr −Vci

0 ≤ V ≤ Vci , Vco ≤ V Vci ≤ V ≤ Vr Vr ≤ V ≤ Vco

(4)

This equation shows that electricity can be generated when there is minimum wind speed, and electricity generation continues until the rated wind speed is reached. At the rated wind speed, the electricity produced is equal to the rated power of wind generation unit. If the wind speed is less than the minimum or more than the maximum limit, the power generated by the wind turbine is zero. Therefore, the relationship between the output power of a wind unit and the wind speed at hub high could be shown as Fig. 4.

4

Objective Functions

In this paper, the energy management of the smart home is studied as a multiobjective optimization problem. Minimizing of the electricity bill and peak demand of the smart home in 24 hours is considered as objective functions. Mathematically, the main purpose of the optimization is presented by Eq. 5. Here, IEP and IPD show the electricity bill (EB) and the peak demand of the smart home, respectively. Objectivefunction = min {IEP , IPD }

(5)

274

4.1

M. Alilou et al.

Electricity Bill

The electricity bill of the SH is including the cost of energy purchased from the smart grid (CSG ), the cost of the produced energy of the wind turbine (CDG ), and the income from selling the energy to the smart grid (ISG ). Consequently, the EB of the SH is formulated based on costs and incomes in Eq. 6. IEP = CSG + CW T − ISG

(6)

The consumer purchases his demand from the distribution company with a variable price. The real-time pricing (RTP) day-ahead tariffs are determined by the independent system operator (ISO). As there is the wind turbine in the smart home, a part of electricity demand is supplied by home energy sources, and so the extra power is purchased from the smart grid. The total consumption of all appliances at period t is calculated by Eq. 7. Here, na is the number of appliances. pai is the energy consumption of each appliance at period t, while PApt is the total energy consumption of appliances at period t. PApt =

na i=1

(7)

pa i

Equation 8 is used for computing the hourly production power of the wind turbine. Here, nWT is the number of the local wind turbine. PDGt is the total produced power of the unit at period t. PW Tt =

nW T i=1

(8)

pd i

Consequently, Eq. 9 is utilized for calculating the hourly extra demand of the SH so that it is supplied by SG. PSHt = PApt − PW Tt

(9)

Finally, the daily cost of energy purchased from the smart grid is expressed by Eq. 9. CSG =

np t=1

PSHt × T r t

if

PSHt > 0

(10)

In this equation, Trt is the electricity tariff at period t which is predetermined by the independent system operator. It is worth mentioning that considering a scheduling period of one hour, the total number of periods for the day ahead would be 24. Equation 10 shows that SH purchases energy from the SG when the consumed power of appliances is more than the produced power of the wind turbine at the considered period. On the other hand, the smart home sells electricity to the smart grid at some periods when the generated power of the home energy source is more

Optimal Management of Smart Home Appliances Considering Stochastic. . .

275

than the consumed power of appliances. Therefore, the income from selling energy to the SG is formulated by Eq. 11. ISG =

np

− 1 × PSHt × pr t

t=1

if

PSHt < 0

(11)

Here, prt is the electricity price of sold back to the SG at period t. This price is predetermined between ISO and consumer. The cost of the produced energy of domestic WT units is presented in Eq. 12: CW T =

np nW T t=1

i=1

pd it × pr W T

(12)

where prWT is the cost of the produced 1 kWh energy by WT unit; the costs of investment, maintenance, and operation of the WT are merged in this parameter.

4.2

Peak Demand

Although economic parameters are more important than technical indices from the household consumer’s point of view, smoothing the load curve causes to improve the electricity efficiency of both SH and SG. For this reason, the minimization of the peak demand is also considered as one of the objective functions of the optimization. Equation 13 is utilized to calculate the peak of the power demand of the smart home. Here, PApit is the total consumption of appliances at period t.

IP D = maxt PApt

5

(13)

Constraints

The sample smart home is assumed to operate with the following constraints. Power balance constraint: The power balance constraint is presented at period t by Eq. 14. PtSG2SH + PtW T = PtAP P + PtSH 2SG

(14)

This equation states that the demand of appliances (PtAP P ) and the sold energy to the SG (PtSH 2SG ) are satisfied by the purchased energy from the SG (PtSG2SH ) and the produced power of the wind turbine (PtW T ). Appliances constraint: The home energy management system determines the best choice for the operational time vector of appliances. Therefore, the daily state vector of each appliance is defined as follow: Ii = [I1 , I2 , . . . , It , . . . IT −1 , IT ]

(15)

276

M. Alilou et al.

Here, It is the mode of appliance i at period t; the amount of this parameter is 1 or 0 that it is equal to on/off of appliance i at period t. All appliances should be executed during the day; this constraint is formulated in Eq. 16. In this equation, np is the number of the time interval during the day. np t=1

Iit ≥ 1

(16)

Moreover, each appliance has the number of operating periods. In other words, each device needs to run for consecutive periods without interruption when it is started. Equation 17 shows this constraint, where ps and pe are the starting and ending time interval of appliance i, respectively. OPi is the proper number of operating time intervals of appliance i. With regard to this constraint, the starting time of devices should be selected based on Eq. 18. pe t=ps

Iit = OP i

ps ≤ np − OP i + 1

(17) (18)

It is considered that the consumption of appliances is constant at each time interval of the operational period. In other words, we consider that each appliance has a fixed power demand profile during the operating time. This profile cannot be modified by the energy management program; otherwise, the device cannot correctly operate. Therefore, only the starting time of the appliances is optimized by the HEMS. Wind turbine constraint: Wind turbine should have the allowable size in the specific range that is shown in Eq. 19 PW T min ≤ PW T i ≤ PW T max

6

(19)

Energy Management Algorithm

As mentioned above, the combination of MODA and fuzzy method is utilized to multi-objectively optimize the daily schedule of home appliances and wind turbine. The main inspiration of the MODA develops from the swarming behaviors of dragonflies in nature. The MODA is able to improve the initial random population for a given problem, converge toward the global optimum, and provide very competitive results compared to other well-known swarm intelligent algorithm. In the multi-objective dragonfly algorithm, five behavior patterns of these insects are considered to improve the situation of artificial dragonflies. Separation, alignment, cohesion, attraction toward a food source, and distraction outward an enemy are the considered principles. In this intelligent algorithm, particles are improved using the following equations (Mirjalili 2016):

Optimal Management of Smart Home Appliances Considering Stochastic. . .

277

Xt+1 = (sSi + aAi + cCi + f Fi + eEi ) + W Xt

(20)

Xt+1 = Xt + Xt+1

(21)

N Si = − X − Xj

(22)

Where, j =1

Ai = Ci =

N j =1

N j =1

(23)

Vj /N

Xj /N − X

(24)

Fi = X + − X

(25)

Ei = X− –X

(26)

After multi-objective optimization of the technical-economic issue of energy management of the smart home, the fuzzy satisfying method is applied to find the best compromise solution which represents the best amount of each objective function equal to the best schedule of smart home energy sources and appliances. In the fuzzy method, the maximum value of the membership μk , which is calculated using Eqs. 27 and 28, can be chosen as the best compromise solution (Alilou et al. 2020a).

μi = k

⎧ ⎪ ⎨ ⎪ ⎩

Fi k ≤ Fi min Fi min < Fi k < Fi max

1 Fi max −Fi k Fi max −Fi min

0

Fi NO k μi μk = NK i=1 NO k=1

max

i=1 μi

k

≤ Fi

(27)

k

(28)

Consequently, the complete method for smart energy managing the smart home is shown in Fig. 5. Figure 5 indicates that the details of home appliances, wind turbine, and economic parameters are initially inputted into the home energy management system. Then, the home management system produces initial random particles for utilizing the intelligent algorithm. Of course, the home energy management system uses also the hybrid method of the Latin hypercube sampling algorithm, and K-means clustering algorithm is used to find the stochastic data wind turbine. Particles are improved by applying multi-objective dragonfly algorithm while considering objective functions and constraints. Finally, home energy management system utilizes the fuzzy decision-making method for selecting the best compromise particle equal to the optimal schedule of home appliances and wind turbine.

278

M. Alilou et al.

Start

Input initial details of home appliances

Input initial details of wind turbine

Input economic parameters

Create random particles

Calculate amount of objective functions

Calculate amount of objective functions

Evaluate constraints

Improve hourly schedule of wind turbine

Improve starting time of home appliances

Update particles by MODA

Have reached maximum iterations?

No

Yes Select the best particle by Fuzzy method

Determine the hourly schedule of home appliances and wind turbine

Calculate amount of considered indices

End

Fig. 5 Flowchart of the proposed method

7

Numerical Results and Discussion

In this section, the proposed method for managing the operational schedule of the smart home’s devices is applied to a sample home. It is considered that this SH has 11 appliances which should be used each day. Home appliances are divided

Optimal Management of Smart Home Appliances Considering Stochastic. . . Table 1 The required time intervals of home appliances

Fixed appliance Refrigerator Purifier Lights Microwave oven Oven TV Iron Shiftable appliance Washing machine Dishwasher Boiler Vacuum cleaner

279

The number of time intervals 24 6 5 1 1 4 2 The number of time intervals 2 2 6 1

into shiftable and fixed devices. Washing machine, dishwasher, boiler, and vacuum cleaner are considered as shiftable appliances, while refrigerator, purifier, lights, microwave oven, oven, TV, and iron are fixed devices of SH. The number of required time intervals of each appliance is presented in Table 1. The hourly consumption of both shiftable and fixed appliances is shown in Fig. 6 (Alilou et al. 2020b). In this paper, the wind turbine is the energy sources of the smart home. The WT unit used in this study is of WINDMILL type with the capacity of 1500 W. Moreover, the cut-in, normal, and cut-out speeds are 4, 13, and 20 m/s. The shape and scale indices for the WT are 2 and 6.5, respectively. Hourly wind speed data are shown in Fig. 7 (Alilou et al. 2021). The price of the produced power of the wind turbine is determined as 0.075 $/kWh. Moreover, the price of sold power of SH to the grid is 0.129 $/KWh; in other words, the smart grid distribution company purchases the power of the SH with this price. The market price of the grid is shown in Fig. 8 (Alilou et al. 2020c). For better evaluating the results, the proposed scheduling method is analyzed through two case studies. The following case studies are conducted: • Case 1: A home with smart appliances • Case 2: A home with smart appliances and WT

7.1

Case 1

In this case, the smart home doesn’t have the wind turbine. So the HEMS has to manage the operational time of appliances based on the market price. The starting time of fixed devices is the same in all cases. Table 2 presents the starting time of fixed devices of the smart home. So in this case, the operational time of shiftable appliances is optimized by the proposed method. As mentioned above, firstly, the multi-objective dragonfly algorithm is applied to optimize the objective functions

280

M. Alilou et al. Washing machine

Boiler

0.6

Power (kW)

Power (kW)

1.5 1 0.5 0

0.4 0.2 0

1

3

5

7

9

11 13 15 17 19 21 23

1

3

5

7

Time-intervals

Power (kW)

Power (kW)

Vacuum Cleaner

0.6

0.4 0.2 0

0.4 0.2 0

1

3

5

7

9 11 13 15 17 19 21 23

1

3

5

7

TIme-intervals

Power (kW)

Power (kW)

11 13 15 17 19 21 23

Refrigerator

0.06 0.04 0.02

0.04 0.02 0

0 1

3

5

7

1

9 11 13 15 17 19 21 23

3

5

7

Time-intervals

9 11 13 15 17 19 21 23

Time-intervals

Lights

0.3

Oven 0.8

Power (kW)

Power (kW)

9

Time-intervals

Purifier

0.08

0.2 0.1

0.6 0.4 0.2 0

0 1

3

5

7

1

9 11 13 15 17 19 21 23

3

5

7

Time-intervals

9

11 13 15 17 19 21 23

Time-intervals

TV

Iron

0.4

Power (kW)

0.15

Power (kW)

11 13 15 17 19 21 23

Time-intervals

Dishwasher

0.6

9

0.1 0.05

0.3 0.2 0.1

0

0 1

3

5

7

9 11 13 15 17 19 21 23

1

3

5

Power (kW)

Time-intervals

7

9 11 13 15 17 19 21 23

Time-intervals

Microwave oven

0.06 0.05 0.04 0.03 0.02 0.01 0 1

3

5

7

9 11 13 15 17 19 21 23

Time-intervals

Fig. 6 Load profiles of home appliances

Optimal Management of Smart Home Appliances Considering Stochastic. . .

281

Wind speed (m/s)

9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 101112131415161718192021222324 Fig. 7 Hourly wind speed forecast

Market price ($/Kwh)

0.5 0.4 0.3 0.2 0.1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Fig. 8 The market price of the smart grid (RTP tariff) Table 2 The starting time of fixed appliances

Appliance Refrigerator Purifier Lights Microwave oven Oven TV Iron

Starting time 1 17 18 7 14 21 8

and create the Pareto front. Figure 9 shows optimized objective functions. Then, fuzzy setting method is utilized to select the best schedule of home devices. Therefore, the starting time of shiftable appliances is selected by applying the fuzzy method to Pareto optimal solutions. The starting time of these appliances is shown in Table 3. According to this table, most of the shiftable appliances are

282

M. Alilou et al.

Fig. 9 The Pareto front after applying MODA

Table 3 The starting time of shiftable appliances in Case 1

Appliance Washing machine Dishwasher Boiler Vacuum cleaner

Starting time 15 4 3 5

utilized when the cost of electrical energy is the lowest. Moreover, the peak demand of the SH has also effect on the operational time of devices. With utilizing the results of Tables 2 and 3 (the starting time of appliances) and the consumption power of devices at each time interval of their operational period, the hourly power of the smart home is calculated. Figure 10 shows the hourly consumption of appliances of the SH in Case 1. The peak demand of the SH is 1.325 KW which occurs at 15th interval. As the SH doesn’t have generation unit, so all demand of the home is supplied by the smart grid. The hourly electricity bill of the SH is demonstrated in Fig. 11. The daily electricity bill of the smart home is $2.0598.

7.2

Case 2

In this case, the home energy management system can provide part of the demand of the home by a personal wind turbine. Of course, the owner of the SH can sell energy to the smart grid when the produced power of the wind turbine is more than the demand of appliances. So the hourly produced power of the wind turbine and the shiftable appliances of the SH are optimized by the proposed method. The starting time of shiftable appliances is shown in Table 4. The hourly energy detail of the smart home including the demand of appliance, the produced power of the WT, and the transferred power between SH and SG is shown in Fig. 12. According to this figure, the wind turbine can supply most part of

Optimal Management of Smart Home Appliances Considering Stochastic. . .

283

1.4 1.2

KW

1 0.8 0.6 0.4 0.2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Hour Fig. 10 Hourly consumption of appliances of the smart home in Case 1

0.5 0.4 $

0.3 0.2 0.1 0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324 Hour Fig. 11 Hourly electricity bill of the smart home in Case 1 Table 4 The starting time of shiftable appliances in Case 2

Appliance Washing machine Dishwasher Boiler Vacuum cleaner

Starting time 16 5 5 4

the home’s demand, and it also causes the owner of the SH to sell extra produced power by the WT to the SG. Therefore, the peak demand of the SH which should be supplied by the SG is 0.619 KW. Its amount is reduced about 54% than Case 1. Figure 13 demonstrates the hourly economic detail of the smart home including the cost of the wind turbine, the energy purchase fee, the energy sales revenue, and the electricity bill. As shown in this figure, the daily electricity bill of the smart home is 0.1261 $ after utilizing the wind turbine. Therefore, the electricity cost of the SH is reduced about 94%.

284

M. Alilou et al. Demand of Apps

2

Power of WT

Transfer data of SH

KW

1.5 1 0.5 0 1

-0.5

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

-1 Hour

Fig. 12 Hourly energy detail of the smart home in Case 2

0.3

Energy purchase fee

Energy sales revenue

Cost of DG

Electricity bill

$

0.2 0.1 0 1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

-0.1 -0.2

Hour

Fig. 13 Hourly economic detail of the smart home in Case 2

8

Conclusion

The operational schedule of smart homes, which including appliances and wind turbine, was optimized in this paper. The combination of MODA and fuzzy setting method was utilized to multi-objectively optimize the technical-economic objective functions. Numerical results, obtained from applying the proposed method in the sample smart home with controllable appliances and wind turbine, demonstrate that the HEMS of the SH can properly optimize the operational time of home’s devices in order to reduce the peak demand and electricity bill of the smart home. The wind turbine has a high effect on the performance of the SH so that the electricity bill is decreased about 95% after utilizing the WT unit. The peak demand of the smart home is also reduced about 54% after utilizing the proposed energy management method. Totally, the SH has the proper performance when all consumer and producer devices of the home are operated by the proposed method so that the owner of the smart home reduces his costs and also earns money using the sold energy to the smart grid.

Optimal Management of Smart Home Appliances Considering Stochastic. . .

285

References S. Aghajani, M. Kalantar, Operational scheduling of electric vehicles parking lot integrated with renewable generation based on bilevel programming approach. Energy 139, 422–432 (2017) G.R. Aghajani, H.A. Shayanfar, H. Shayeghi, Demand side management in a smart micro-grid in the presence of renewable generation and demand response. Energy 126, 622–637 (2017) M. Alilou, V. Talavat, H. Shayeghi, Simultaneous placement of renewable DGs and protective devices for improving the loss, reliability and economic indices of distribution system with nonlinear load model. Int. J. Ambient Energy 41, 871–881 (2020a) M. Alilou, B. Tousi, H. Shayeghi, Multi-objective unit and load commitment in smart homes considering uncertainties. Int. Trans. Electr. Energy Syst. 30, Article No. e12614 (2020b) M. Alilou, B. Tousi, H. Shayeghi, Home energy management in a residential smart micro grid under stochastic penetration of solar panels and electric vehicles. Sol. Energy 212, 6–18 (2020c) M. Alilou, B. Tousi, H. Shayeghi, Multi-objective energy management of smart homes considering uncertainty in wind power forecasting. Electr. Eng.., In press (2021) B. Barbato, A. Capone, L. Chen, F. Martignon, S. Paris, A distributed demand-side management framework for the smart grid. Comput. Commun. 57, 13–24 (2015) A. Dejamkhooy, M. Hamedi, H. Shayeghi, S. SeyedShenava, Fuel consumption reduction and energy management in stand-alone hybrid microgrid under load uncertainty and demand response by linear programming. J. Oper. Autom. Power Eng. 8, 273–281 (2020) L. Gelazanskas, K. Gamage, Demand side management in smart grid: a review and proposals for future direction. Sustain. Cities Soc. 11, 22–30 (2014) W. Li, T. Yigitcanlar, I. Erol, A. Liu, Motivations, barriers and risks of smart home adoption: from systematic literature review to conceptual framework. Energy Res. Soc. Sci. 80, 102211 (2021) M. Mazidi, A. Zakariazadeh, S. Jadid, P. Siano, Integrated scheduling of renewable generation and demand response programs in a microgrid. Energy Convers. Manag. 86, 1118–1127 (2014) S. Mirjalili, Dragonfly algorithm: a new meta-heuristic optimization technique for solving singleobjective, discrete, and multi-objective problems. Neural Comput. & Applic. 27, 1053–1073 (2016) E. Naderi, A. Dejamkhooy, S. SeyedShenava, H. Shayeghi, MILP based optimal design of hybrid microgrid by considering statistical wind estimation and demand response. J. Oper. Autom. Power Eng. 10, 54–65 (2022) J. Pohl et al., Environmental saving potentials of a smart home system from a life cycle perspective: how green is the smart home? J. Clean. Prod. 312, 127845 (2021) M. Rastegar, M. Fotuhi, F. Aminifar, Load commitment in a smart home. Appl. Energy 96, 45–54 (2012) M. Rastegar, M. Fotuhi, M. MoeiniAghtaie, Developing a two-level framework for residential energy management. IEEE Trans Smart Grid 9, 1707–1717 (2018) H. Shayeghi, M. Alilou, Multi-objective demand side management to improve economic and environmental issues of a smart microgrid. J. Oper. Autom. Power Eng. 9, 182–192 (2021) T. Stathopoulos et al., Urban wind energy: some views on potential and challenges. J. Wind Eng. Ind. Aerodyn. 179, 146–157 (2018) E. Struckell, D. Ojha, P. Patel, A. Dhir, Ecological determinants of smart home ecosystems: a coopetition framework. Technol. Forecast. Soc. Chang. 173, 121147 (2021) Y. Wang, Y. Li, Y. Cao, Y. Tan, L. He, J. Han, Hybrid AC/DC microgrid architecture with comprehensive control strategy for energy management of smart building. Electr. Power Energy Syst. 101, 151–161 (2018)

Novel Data-Driven Methods for Evaluating Demand Response Programs in a Smart Grid Lihui Bai and Arnab Roy

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Project Background and Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Two Data-Driven Demand Response Evaluation Methods . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 A DR Evaluation Method for Coincident Load Reduction . . . . . . . . . . . . . . . . . . . . 3.3 Evaluating Annual Energy Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Computational Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Effect on Reducing Coincident Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Effect on Annual Energy Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Effect of Critical Peak Pricing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

288 291 292 292 292 294 296 297 300 302 304 305 305

Abstract

This work aims to provide an accurate assessment of a demand response program that encourages cyber-physical interactions in residential power distribution systems. The advanced technologies considered in this chapter include Wi-Fienabled programmable smart thermostats, high-efficiency and connected water heaters, residential battery storage systems, improved weatherproofing, and advanced metering infrastructure (AMI). We present the design of a field demonstration study, the data collection method, and the data-driven comparative

L. Bai () Department of Industrial Engineering, University of Louisville, Louisville, KY, USA e-mail: [email protected] A. Roy () Procter and Gamble Co., Cincinnati, Ohio, USA e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_152

287

288

L. Bai and A. Roy

evaluation methodology. In analyzing the impacts of various technologies on a home’s coincident load, we propose a novel day-matching algorithm combined with a paired t-test. In analyzing annual energy savings and efficiency, we propose a two-stage algorithm considering three seasons (shoulder, winter, and summer) and degree-day adjustment factors. The new evaluation methods are implemented on a demand response pilot program with 330 participating homes in a mid-western US municipality, each installed with various technologies. Computational results on the impacts of each technology on the coincident load and annual energy savings show the proposed data-driven methods are effective and scalable. Keywords

Demand response · Energy savings · Coincident load · Internet of things · Power distribution

1

Introduction

The Annual Energy Outlook 2015 report from the US Energy Information Administration (EIA) indicates that residential customers contributed about 21.53% of the total energy used in 2014. With rapid advancement of technologies and cloud computing, many utility companies nowadays are exploring the Internet of Things (IoT) and cyber-physical interactions in residential power distribution (Ullah et al. 2021). The ultimate goal is to increase energy efficiency through successful deployment of advanced technologies such as advanced metering infrastructure, novel variable pricing schemes, and energy storage. Furthermore, it is well known that when consumers’ use of electricity is mainly driven by convenience, coincident demand occurs and results in electric load peaks that require ancillary services at power plants with increased costs. In the past decade, researchers and practitioners have come to agree that carefully designed and implemented demand response (DR) programs provide significant opportunities for reducing system peak load, thus leading to a more efficient power distribution system. To date, many research laboratories as well as utilities have implemented DR pilot programs (Chen et al. 2021; Torriti et al. 2010) thanks to the development of connected technologies such as programmable thermostats, advanced metering infrastructure (AMI), and Wi-Fi-enabled smart appliances. These pilot programs seek to demonstrate the effectiveness of their respective deployed technologies, with two common objectives, namely, improving energy efficiency and improving system load factor (by reducing system peak). Proper program evaluation on whether these objectives are met is as crucial as the DR project itself, because the power industry relies on these evaluations to improve future design and deployment of technologies and infrastructure. While the literature on devising optimal control and pricing algorithms for DR program is quite comprehensive (Tsui and Chan 2012), research on empirical

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . .

289

methods for assessing the effectiveness of such DR programs is scarce. Many of these works are conducted by government agencies to assess the viability of the program. At the macro-level, for example, the Federal Energy Regulatory Commission has long been publishing its annual report (Federal Energy Regulator Commission 2021) on the state of such programs across the USA and their impact on savings on peak load. Di Maio et al. (2021) present a framework to assess the adequacy of the integrated energy systems by studying various uncertainties in climate conditions. As another example, the US Department of Energy Office of Scientific and Technical Information have conducted multiple studies (Shen et al. 2012; Hale et al. 2018) in order to gauge the effect of demand response programs in electricity markets. Particularly, Shen et al. (2012) study the effect of DR in high potential electricity markets with distributed resources. The authors conclude that DR resources, on average, can save 10% compared to grid resources as it can avoid line losses. Hale et al. (2018), on the other hand, look to address the increasing energy demand through demand response. In addition, a study conducted by the energy firm Brattle Group Dunham-Jones (2000) states that DR programs have achieved approximately 5% reduction on peak demand, which amounts to about $300 million savings. Finally, Stoll et al. (2017) simulate 14 different DR scenarios in the Florida market and conclude that such programs would reduce the cost of production by reducing the low-load hours of generators. Another stream of research examines, at a micro-level, the effectiveness of DR programs on individual consumers. Cappers et al. (2010) conduct a comprehensive study to assess the contribution of DR programs in the US. Furthermore, Li et al. (2017) propose an unsupervised learning algorithm to estimate the baseline load of consumers in a DR program. The authors suggest an adaptive density-based spatial clustering algorithm to assess the typical load patterns (TLP) of individual customers. Additionally, Sun et al. (2019) propose a data-driven probabilistic peak demand estimation framework using AMI data while including the sociodemographic data of the consumers for a smart-meter trial project in London. Using sampling via copulas and correlation-based grouping, their evaluation finds that participants in organized wholesale markets become better in load curtailment and cost savings with experience. For micro-level assessment of DR programs, many have used simulation to study and estimate residents’ energy consumption. For example, Conte et al. (2007) develop a simulation environment for washing machines and dishwashers with the ability to change the control parameters of the appliances. The efficiency of different combination of the parameters is assessed using a global delay index (an index to measure the delay of operation of the devices from the user preferred time) and number of overloads (excess load from the baseline). In Stavrakas and Flamos (2020), time shift of the use of appliances is quantified to determine the efficiency of different energy sources such as conventional and renewable energy. Similarly, Gruber and Prodanovic (2012) use time shift of appliances and employ a probabilistic approach in a residential energy consumption simulation. Energy efficiency is calculated under the influence of the time shift along with the assumption of price elasticity of the consumers. Finally, Pedersen et al. (2017)

290

L. Bai and A. Roy

develop two, centralized and decentralized, economic model predictive control schemes and demonstrate energy efficiency delivered by them, by comparing with traditional PI controllers. The authors show that the proposed schemes lead to cost savings of 3–6% over traditional controllers. This paper develops two novel data-driven methods to evaluate a demand response program and tests these methods with a DR pilot program implemented by a multiplicity utility in the Midwest US in 2015–2017. The objective of the DR pilot program is two-fold. First, through the deployment of four technologies, i.e., Wi-Fi-enabled programmable smart thermostats, high-efficiency and connected water heaters, residential battery storage systems, and improved weatherproofing, the program will reduce the total energy consumption. Second, through the implementation of a critical peak pricing (CPP), an innovative residential electricity rate featuring demand charge during monthly system peak hour, and supporting technologies (programmable thermostats and battery systems), the project will reduce the system peak, thus improve the system load factor. The advanced metering infrastructure (AMI) is in place for the project to monitor the load in real time. Utilizing these user-level data, this paper proposes two methods to evaluate two objectives often associated with a DR program, i.e., improving annual energy savings and system load factor. In addition, these methods are applied to examine the effect of one critical variable – the CPP rate – on the total energy consumption as well as the coincident load for the Midwest US municipality. The contribution of the paper is three-fold. First, most existing evaluation methods used by a DR project do not address coincident load; this paper considers proper evaluation of the reduction in coincident load in the presence of a CPP rate structure and proposes a novel day-matching algorithm to assist with such evaluation. Second, in assessing the annual energy savings, most literature applies monthly or yearly degree days to normalize weather-related factors; this paper takes a data-driven approach to identify weeks that belong to three seasons, i.e., “shoulder season,” winter, and summer. Weekly consumption is either kept the same if belonging to the “shoulder season” or adjusted using current as well as 10-year historical heating/cooling degree days in winter/summer season. Third, to date existing DR evaluation methods are ad hoc and specific to the particular pilot DR project in study. The two proposed evaluation methods in this paper are not only data-driven but more importantly systematic and scalable. The remainder of the paper is organized as follows. Section 2 provides the background of the DR field demonstration project in the Midwest US municipality and presents the formal problem statement. Section 3 proposes the evaluation methodologies for assessing the improvement in coincident load and total energy consumption of participant homes due to a DR pilot project. Section 4 applies the two proposed evaluation methods to the above mentioned DR pilot project and discusses the results, and Section 5 concludes the paper.

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . .

2

291

Project Background and Problem Statement

As mentioned previously, our paper aims to develop a general-purpose data-driven framework for assessing the effectiveness of a demand response program. This research intent is inspired by our participation in a DR pilot program with a Midwest US community. The pilot project had 330 participating households during the period of 2015–2017. One overarching innovation applied to all 330 participant home is a critical peak pricing (CPP) rate. Under this CPP rate, the utility company will issue a much higher electricity rate with 24-hour notice based on their predicted communitywide monthly peak. This CPP rate is designed to encourage customers to reduce consumption during the peak period and thus improve the system load factor. The latter is one of the two objectives of the pilot program. The other key objective of the pilot program is to increase energy efficiency or decrease annual electricity consumption by installing one or multiple of the four smart technologies in the 330 participant homes. These include smart thermostats (Wi-Fi connected and remotely controlled), advanced heat pump water heaters, ultra-efficiency home envelope, and residential battery systems. Depending on these households’ historical energy consumption pattern and their household heating sources (gas vs. electric), they are divided into five groups, each receiving distinct set of combined technologies. This will enable us to properly evaluate the effect on consumption of different technologies. The five groups and their associated technologies are presented in Table 1. There are three central questions to be addressed by this paper. First, how do we estimate the effect of the installed smart technologies on coincident load, thereby on system load factor? Second, how do we determine the change in annual energy consumption due to the four technologies? Third, how do we evaluate the effect of the CPP rate with supporting technologies? The two evaluation methods proposed in the next section will address the first two questions directly and will be used to address the third question as well.

Table 1 Home categorization Home category Ultra Advanced electric Advanced gas Basic electric Basic gas

Description All advanced technologies installed Electric HPWH; excludes advanced enveloping, battery Gas HPWH; excludes advanced enveloping, battery Electric HPWH; excludes battery Gas HPWH; excludes battery

# of homes 50 95 18 37 126

292

L. Bai and A. Roy

3

Two Data-Driven Demand Response Evaluation Methods

3.1

Notation

The notations used in subsequent models in this paper are provided in the Table 2.

3.2

A DR Evaluation Method for Coincident Load Reduction

Coincident load (CL) is the amount of power used by a given household that occurs during the 1-hour time where the entire community’s total load is at the peak for the month. In other words, CL is the household’s electricity load that coincides with the system monthly peak hour. Monthly peak loads are significant for utility companies as it has a direct effect on the reliability of the power system, as well as the cost of purchasing power from the electricity market on behalf of the community they serve. The latter is because, on the power generation side, peak demand required by utilities often forces the power generation company to use their reserve capacity, which are significantly more expensive than the scheduled capacity due to ramp-up and ramp-down operations. In the power distribution system, coincident load provides a direct measure of a home’s contribution towards the system peak, therefore a proper piece of data towards CPP billing. In this section, we focus on examining the effects of the various installed technologies on coincident load reduction for the 330 participant homes in five groups. Comparison of a home’s coincident load between the test and control years can be accomplished through paired t-tests. In order to conduct a paired t-test by controlling weather conditions, we propose a two-step procedure. First, identify a set of day pairs: one in the control year without CPP or installed technologies and one in the test year with CPP and installed technologies, so that weather conditions are similar on the two days. Weather conditions are represented by an array factors such as temperature, humidity, and wind speed. Second, homes’ coincident loads on those paired days in the two years are collected and subject to comparison. To identify the day pairs between the test and control years, we acknowledge that, in DR programs, utility companies can announce multiple so-called event days on which they predict high probability of system peak. For example, in the case of the Midwest municipality DR pilot project, there were approximately 4–6 event days for a given month. Assuming users respond to an “event day alert” the same way they do to an actual monthly peak day, we designate all alerted event days in the test year as target days. We then select multiple candidate days according to the method explained below and subsequently pair the target day with one of the candidate days. i Algorithm 1 below determines a matching day for a target day d, where fd,t is the value of the weather factor i for day d during time interval t. Examples of these weather factors include temperature, humidity, pressure, dew point, and wind speed. The algorithm calculates the mean squared error between the target day and a candidate matching day as the weighted sum of the squared error of each weather

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . .

293

Table 2 Notation Sets D T I Y W N otation ly By SHy Sy Wy dy lˆyS , lˆyW

Description Set of target days d Set of time intervals t Set of weather factors i Set of years y Set of weeks w Description Baseload for year y Baseline load for year y Shoulder season weeks for year y Summer weeks for year y Winter weeks for year y Degree days for year y Variable summer and winter loads

Lˆ Sy , Lˆ W y

Weather-adjusted variable summer and winter loads

i fd,t

Weather factor i for day d and interval t

Algorithm 1 1: procedure DAY MATCHING ALGORITHM 2: Set a target day d 3: for each candidate day d within a pre-specified window (± days) in the previous year do 4: for each time interval t do 5: Calculate theweighted sum of squared error of weather factors: i − f i )2 SEd ,t = i wi (fd,t d ,t 6: end for SE 7: Calculate the mean squared error MSEd = t |T | d,t 8: end for 9: Choose a set of candidate matching days D such as, + ρ}, MSEd ≤ Min {MSE ∀d ∈ D d d

10:

Select the matching day d ∗ from D such as, i |} d ∗ = argmin{|fˆdi −1 − fˆd−1 d

i over all t for day d where fˆdi is the average of fd,t 11: Return d ∗ as the matched day for target day d 12: end procedure

factor for each time interval t ∈ T . It is worth mentioning that since weather factors have different scales, we normalize their values into an interval of [0, 1] by applying the feature scaling method in Zheng and Casari (2018). Note that in Step 9, we use a set of candidate matching days in Algorithm 1 for the final selection because research of historical weather data revealed that previous day weather has a significant effect on the weather of a particular day. Of all the weather factors analyzed, temperature is the most significant factor to electricity

294

L. Bai and A. Roy

consumption, and hence only temperature is taken as a selection criterion in Step 10. Thus, the final selected day d ∗ not only has a weather profile close to the target day but also is close in terms of previous day temperature profile. The values of the parameters wi , ρ, and vary according to application. Once we have a matched day d ∗ for the target day d, we can find the coincident load for each such day. There are multiple event days reported in a month, especially during extreme seasons. As mentioned previously, each event day constitutes a target day in Algorithm 1. Coincident load for each target day and that for the corresponding matched day are then identified. Collecting such paired coincident load data throughout the year among all homes within a homogeneous group, a paired t-test is then conducted to determine the statistical significance of the difference. This will reveal whether the DR technologies and pricing scheme have motivated the customers to change their consumption pattern. We present the numerical results of applying Algorithm 1 and paired t-test to the DR pilot project in Sect. 4.

3.3

Evaluating Annual Energy Savings

Our second objective is to develop a data-driven evaluation framework with regard to annual energy savings across several seasons due to a DR program. Weather adjustment is crucial for such analysis to be valid. For example, the energy savings in an extremely hot month in the test year may appear less pronounced as it should if such weather condition is not incorporated or normalized properly in the analysis. In the literature, both academic and industrial (e.g., McMaster and Wilhelm 1997; Thom 1954), the degree day method has been widely used for “normalizing” energy use of buildings. For example, the Climate Prediction Center under the National Weather Service at the National Oceanic and Atmospheric Administration (NOAA) for the US Department of Commerce maintains a database of Degree Days Statistics (e.g., Council et al. 2006) to facilitate weather-adjusted energy studies. There are heating and cooling degree days, where heating degree days (HDD) calculate the sum of the positive differences between a base temperature and the average daily outdoor temperature over the study period. Formally, HDD =

(base temperature − avg. daily temperature),

(1)

days in study period

where base temperature is set to be 65°F in the US and average daily temperature is simply (maximum daily temperature − minimum daily temperature)/2.

(2)

The study period is typically on a month, or seasonal, or yearly basis. Similarly, the cooling degree days (CDD) calculate the sum of the negative differences between a base temperature and the average daily outdoor temperature.

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . .

295

For the purpose of evaluating a DR program with installed smart technologies, we propose that the above degree day adjustments only apply to a home’s HVAC load instead of its total load. Consequently, this requires us to separate a home’s HVAC loads from its total load. Therefore, careful annual load analysis for 52 weeks during a year is needed. For example, our case study with the Midwest US municipality DR pilot project validates the notion of breaking a home’s total load into two: HVAC and base load. Figure 1 depicts the average consumption during each of the 52 weeks for homes with all four technologies installed, i.e., “ultra-homes” as per Table 1. The figure confirms that there exists a so-called shoulder season during which home’s energy consumption is lowest and rather flat. In contrast, during the summer and winter, loads exhibit higher magnitudes and variability primarily due to weather. Similar seasonal consumption patterns are demonstrated in similar figures (not reported here) for other four types of homes. Given the three-season pattern exhibited in home’s annual consumption profiles from the project, we employ a two-stage approach to estimate the difference between a home’s total energy consumption during the two years: one before and one after implementing the DR program. The first stage utilizes weekly energy usage data to identify weeks in the shoulder (SH), winter (W), and summer (S) seasons for any home. The second stage applies degree day adjustment to energy usage data in each season and calculates the total season-adjusted energy consumption and compares the consumption between the two years. Algorithm 2 below formally describes the two-stage approach in nine steps. Several concepts are key to Algorithm 2. First, the weekly baseline load (By ) sets the foundation for the weather-independent energy consumption for a home. For each of the two years of interest, we identify four consecutive weeks that have the lowest total consumption as the “shoulder season,” and the total consumption is the total baseline load. The weekly baseline load is simply the total baseline load divided by four. Second, the variable load (lˆyS and lˆyW ) represents the weatherdependent energy consumption due to the HVAC system. It is calculated by deducting the baseline load from the seasonal (summer or winter) load. This variable load is used as a surrogate for the HVAC load for the home during summer and winter. Third, the weather-adjusted variable load (Lˆ Sy and Lˆ W y ) integrates degree days to account for variation in weather between the two years. In Step 8 of the

Fig. 1 Average weekly consumption by ultra-homes for 52 weeks

296

L. Bai and A. Roy

Algorithm 2 1: 2: 3: 4: 5: 6: 7: 8: 9: 10: 11:

procedure ANNUAL ENERGY SAVING ESTIMATION ALGORITHM Set 2 years of interest for comparison, and let year index y ∈ {1, 2}. Identify the baseload ly for y = 1, 2. Calculate the weekly baseline load: By = ly /4, for y = 1, 2 Identify shoulder, summer, and winter weeks for each of the 2 years; let SH y , Sy , and Wy be the number of shoulder, summer, and winter weeks for y = 1, 2. Obtain the total energy consumption for winter and summer for year y, denoted as lyW and lyS . Calculate the variable load (due to HVAC) as lˆyS = lyS − Sy · By and lˆyW = lyW − Wy · By for y = 1, 2. ˆW ˆW ADD Calculate the weather-adjusted variable load as Lˆ Sy = lˆyS · ADD DDy and Ly = ly · DDy Calculate the weather-adjusted total annual consumption: Ey = SH y · By + Lˆ Sy + Lˆ W y . Calculate and return the difference between E1 and E2 end procedure

above algorithm, DDy is the total degree days for that season and ADD is the normalization factor based on the most recent 10-year average degree days.

4

Computational Results

In this section we test the two proposed data-driven DR evaluation methods in Sects. 3.2 and 3.3 with the case study of the DR pilot program implemented in a Midwest US municipality in 2016–2017. Recall that four smart technologies were installed in 330 homes in 5 categories (Table 1) in 2016–2017 along with the deployment of the CPP rate structure. Therefore, our evaluation of the DR program will compare relevant quantities (either annual energy consumption or coincident load) during this year with the benchmark/control year 2014–2015 when no technologies and CPP were in place. However, AMI was in place in the entire community for both test and control years. Two major sets of data were requested for the analysis. The first set contains the AMI data with 1-hour resolution in 2016–2017, as well as the AMI data in the benchmark year 2014–2015. In addition, we request weather data containing temperature, humidity, pressure, dew point, and wind speed from the local weather station for the same time periods. The time interval in the weather data we have obtained is usually inconsistent; therefore, we have rounded the time to nearest 5-minute intervals. A sample AMI data for a particular home is provided in Table 3, while Table 4 includes a snippet of the weather data. Next we apply Algorithm 1 in Sect. 3.2 to evaluate the effects of the DR pilot program on reducing coincident load. Alternately, this reveals if the DR program is effective in motivating electricity users to curtail their load when given a “system peak alert” 24 hours in advance.

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . .

297

Table 3 Sample AMI data Home ID 001 001 001 001 001 001 001 001 001 001 001 001

Time 8/1/2016 0:00 8/1/2016 1:00 8/1/2016 2:00 8/1/2016 3:00 8/1/2016 4:00 8/1/2016 5:00 8/1/2016 6:00 8/1/2016 7:00 8/1/2016 8:00 8/1/2016 9:00 8/1/2016 10:00 8/1/2016 11:00

kWh 1.512 1.341 1.266 0.609 0.561 0.444 0.579 0.591 0.339 0.966 1.803 2.121

Table 4 Sample weather data Date 8/1/2016 8/1/2016 8/1/2016 8/1/2016 8/1/2016 8/1/2016 8/1/2016 8/1/2016 8/1/2016 8/1/2016 8/1/2016 8/1/2016

4.1

Time 0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00

Temperature (°F) 74.8 74.8 74.7 74.6 74.6 74.5 74.4 74.3 74.1 74.0 73.9 73.9

Humidity (%) 95 95 95 95 96 95 96 95 95 96 96 96

Pressure (in) 30.00 29.99 29.99 29.99 29.99 29.99 29.99 29.99 29.99 29.99 29.99 29.99

Dew point (°F) 73.3 73.3 73.2 73.1 73.4 73.0 73.2 72.8 72.6 72.8 72.7 72.7

Wind speed (mph) 0 0 0 0 0 0 0 0 0 0 0 0

Effect on Reducing Coincident Load

To compare homes’ coincident loads in 2014–2015 and 2016–2017 years, we select the event days called and reported by the utility company for the period of September 2016 to September 2017. Each such target day is matched with day in 2015 having similar weather conditions using Algorithm 1. In Fig. 2, we observe the temperature and humidity profiles of target day 8/9/2016 and paired/matched day 7/29/2015 following the same pattern and hence make them suitable for conducting the paired t-test. The difference between the coincident load on each target and the matched day throughout the study period among all homes (within a particular group) are the samples subject to the paired t-test.

298

L. Bai and A. Roy

Fig. 2 Comparison of weather conditions between two matching days Table 5 Sample paired t-test results

Target Dates 9/7/2016 9/8/2016 9/9/2016 9/7/2016 9/8/2016 9/9/2016 9/7/2016 9/8/2016 9/9/2016 9/7/2016 9/8/2016 9/9/2016

Home category Ultra

Advanced electric

Advanced gas

Basic gas

p-value 1.83e−13 1.33e−13 1.45e−12 2.74e−11 3.37e−24 4.21e−21 2.77e−06 2.68e−06 6.86e−06 0.0030 0.0012 0.0007

Average load reduction 2.407 2.557 2.359 1.665 2.782 2.380 2.771 2.755 2.393 1.092 1.447 1.513

An illustration of such an analysis is provided in Table 5 with column “Target dates” indicating the event days in 2016. “Average load reduction” reports the coincident load reduction averaged over each category of home for each event day. A p-value less than or equal to 0.05 indicates this average coincident load reduction from 2015 to 2016–2017 is statistically significant. There are 48 event days from September 2016 to August 2017; among them, 1 for ultra-group, 5 for advanced electric group, 18 for advanced gas group, 10 for basic electric group, and 2 for basic gas group have yielded p-value larger than 0.05 in the paired t-tests. This shows that the statistical power for analyzing advanced gas and basic electric groups is not great due to smaller sample sizes for the two groups. We pool winter and summer event days separately and study the coincident load reductions broken down by seasons and by home types in order to better understand the impacts of technologies on coincident load reduction. Event days which yield an insignificant p-value have been removed from this seasonal analysis. Table 6 shows the average, min, and max coincident load reductions over event days in winter (and summer) seasons during the study period. Several observations can be made from Table 6. First, all electric homes show consistent performances between

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . . Table 6 Summary of coincident load comparison in seasons

299

Seasonal coincident load reduction per home per event day Summer Average Minimum Maximum Ultra 2.5 1.4 3.2 Advanced electric 2.3 0.5 3.0 Advanced gas 2.8 1.0 3.4 Basic electric 1.4 1.0 2.1 Basic gas 1.8 1.1 2.6 Winter Average Minimum Maximum Ultra 2.5 0.1 6.0 Advanced electric 2.4 0.01 4.4 Advanced gas 1.2 0.4 1.9 Basic electric 1.5 0.7 2.6 Basic gas 0.9 0.4 1.6

two seasons. Ultra and advanced electric homes post similar higher reduction of 2.3–2.5 kW per event day per home, while basic electric homes post approximately 1.4–1.5 kW per event day per home. Second, on a given event day, all participant homes are estimated to collectively help reduce the system peak load by 667 kW in the summer and 542 kW in the winter. Third, the coincident load reduction for basic electric is consistently less than ultra and advanced electric homes due to the non-existence of the residential battery system in these homes. Finally, as expected, gas-heated homes show approximately 50% less coincident reduction during winter than summer (e.g., for basic gas homes, 1.8 kW coincident load reduction per event day per home in the summer compared to 0.9 kW reduction in the winter). We also attempt to examine the variability for coincident load reduction for three groups with sufficient sample size, i.e., ultra, advanced electric, and basic gas, for summer and winter, respectively. Particularly, we choose 8/22/2017, the system peak day August 2017, to represent summer, and 2/3/2017, the system peak day in February 2017, to represent winter. Figure 3a and b use home-level coincident load reduction data on these two representative summer and winter days as sample to draw box plots for the three groups. Collectively, these two figures indicate that (1) ultra-homes have smallest variations on coincident load reduction among the three and (2) the median coincident load reductions for three groups are similar in the summer while basic gas homes clearly have lower reduction in the winter when compared to the other two groups. Note that certain homes have demonstrated a negative reduction between the two years; however, they all lie in the bottom 25th quartile of each subgroup. We offer some remarks to summarize the results presented above. Our analysis revealed that ultra-homes, which in addition to the advanced technologies had efficient weatherproofing installed, had the highest energy savings among all the categories. The savings translate to approximately 4300 kWh per home when extrapolated to a whole year. Homes with electric HVAC systems have shown greater savings in terms of coincident load. Ultra and advanced electric homes post

300

L. Bai and A. Roy

Fig. 3 Coincident load comparison by seasons and by home type

similar higher reduction of 2.3–2.5 kW per event day per home in both seasons of summer and winter, while basic electric homes post approximately 1.4–1.5 kW per event day per home. The better performance of such homes can be attributed to the battery systems. Ultra-homes show the most consistent behavior throughout the two seasons. The homes together are estimated to reduce 667 and 542 kWh of energy from the peak. It is noteworthy that homes having gas powered HVACs significantly shave less load from the peak.

4.2

Effect on Annual Energy Savings

We apply Algorithm 2 in Sect. 3.3 to evaluate the effects of the pilot program with four installed technologies and CPP on annual energy savings for all participant homes. First, before comparing energy consumption between 2016–2017 and 2014–2015 for participating homes, we identify the characteristics of the “shoulder season,” summer, and winter weeks for those two years. Table 7 displays the results using the weather-adjustment method described in Algorithm 2. Note that winter weeks span

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . . Table 7 Seasonal characteristics

Summer weeks # of summer weeks 10-year summer ADD Winter weeks

301

Wk #22–35 14 963 Wk # 44–52 in 2014 (or 2016) & Wk #1–9 in 2015 (or 2017) 18 3,163

# of winter weeks 10-year winter ADD Summer DDy 2014–2015 826 2015–2016 1,129 Winter DDy 2014–2015 3,628 2015–2016 2,590 Weekly baseline load By (in kwhs) 2014–2015 65.2 2015–2016 45.9

two consecutive years. For example, in the 2014–2015 cycle, “winter” is defined as week 44 through 52 in 2014 and then week 1 through 9 in 2015. After analyzing weather conditions in the two study years, we collect AMI data from all five groups of participant homes (as per Table 1). During the preliminary analysis, we observed “outlier homes” with extremely high or low energy savings for various reasons. First, a home can have extremely low energy consumption during a period of time, if the home is vacant due to vacation during one of the two years. Second, change of ownership between the two comparison years can also contribute to either extremely high positive or extremely high negative energy savings for the same home location. Third, change of household size due to newborns and college students moving back, among others, can also cause drastic change in energy consumption. In order to establish a uniform criterion of identifying outlier sites, we study the annual savings percentages for the two largest subgroups: advanced electric (93 homes) and basic gas (116 homes). Figure 4 shows the box plots along with outliers identified, and it has led us to use −50% and +50% as the lower and upper limits for identifying outlier homes. In other words, if a home’s savings is either below −50% or above 50%, we remove them from the subsequent analysis. Finally, it’s worth noting that because the advanced gas group has the smallest sample size of 17 among all subgroups, and because variability would be more pronounced for small samples, we have chosen a more stringent criterion (−45%, 45%) to remove outliers for this subgroup. Using these criteria, Table 8 displays the number of outliers removed from each group in the final analysis. Once outliers are removed, comparisons on annual energy savings are conducted as the last step. We obtain the average energy savings between 2016–2017 and

302

L. Bai and A. Roy

Fig. 4 Outlier identification in annual energy savings analysis

Table 8 Sample sizes before/after outlier removal Description Original # of homes with complete data # of homes within ± 50% savings # of outliers % of outliers

Ultra 50 48

Advanced electric 93 74

Advanced gas 17 17

Basic electric 33 20

Basic gas 116 90

48

67

13

18

87

0 0

7 9.5

4 23.5

2 10

3 3.3

2014–2015 for all five subgroups of participating homes. Table 9 summarizes the results. Several observations can be made. First, ultra-homes pose the largest energy savings percentage of 27.1% with approximately an average savings of 4300 kWh per home per year. Second, only from the 233 included in the final analysis, the DR project has successfully achieved the annual energy savings of 745,588 kWh.

4.3

Effect of Critical Peak Pricing

Effect on annual energy consumption. To evaluate the effect of the critical peak pricing (CPP) alone on reducing total energy consumption, we choose the homes in the community that are not participant of the DR project as study subjects. While they are also subject to the same CPP rate effective January 1, 2016, they do not have any technologies introduced by the DR project. Thus, comparing monthly consumption between 2015 and 2016, i.e., before and after implementing CPP, for these homes is a reasonable measure for the effect of CPP on residential energy consumption.

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . .

303

Table 9 Energy savings results Annual energy savings Description Ultra Advanced electric Savings (%) 27.1 14 Savings (kWh) 4343.5 2859.4 Subgroups sample size 48 67 Annual projection 745,588 for all homes (kWh) Effect of CPP on annual load Description Low Savings (%) 6 Savings (kWh) 804.7 Effect of CPP on coincident load Description Low Coincident load 0.45 reduction (kWh)

Advanced gas 9.4 1309.8 13

Basic electric 17.9 3467.2 18

Basic gas 19.5 3058.5 87

Medium 14 2276.9

High −8 −507.9

Medium 0.72

High 0.13

We identified 259 non-participant homes’ and collected their hourly energy consumption data each month during 2016 and 2015. After removing outliers, 207 non-participant homes remained in the study. These 207 homes were then subjected to k-means clustering analysis and were grouped into three clusters: low-, medium-, and high-usage homes. Figure 5 shows the distribution of the three groups with 78 low-usage, 34 medium-usage, and 95 high-usage homes. Finally, for each group, we applied Algorithm 2 to estimate their weather-adjusted monthly energy consumption thus the difference between the two years. As indicated in Table 9, CPP has positive impact in incentivizing residents to save energy for low- and medium-usage homes. Particularly, low-usage homes recorded 6% annual savings and medium-usage homes 14% savings. Interestingly, high-usage homes posted negative response to the CPP tariff.

Effect on reducing coincident load. Recall that at the center of the coincident load analysis is the 1 hour at which the entire community’s total load reaches its peak for the month. Each home’s coincident load during that hour (one of the 720 hours of the month), i.e., their contribution to the community-wide monthly peak, is subject to the much higher rate as per the CPP tariff. In addition, during each month in 2016, the utility announces multiple DR event days alerting the possibility of a monthly peak with a 24-hour notice. Therefore, our research treats all these “alerted peak hours” as the true peak hours, as we anticipate residents will respond the same manner as if they were all true peak hours. The clustering of homes and the comparison between 2015 and 2016 were done in a similar fashion as for the annual energy savings in Section 4.3.1. The main

304

L. Bai and A. Roy

Fig. 5 Cluster of non-participant homes

DD

y,m difference is the application of the weather adjustment factor ADD (similar to Algorithm 2) at the hourly level for a given month m and year y. Particularly, the DDy,m ˆ · lh,m,y , where lˆh,m,y is the original adjusted hourly consumption Lˆ h,m,y = ADD load during hour h in month m and year y. In assessing the effect of the CPP tariff on reducing the coincident load, out of the 259 non-participant homes, 17 were identified as outliers. The remaining 242 were then clustered into three groups based on their historical coincident load (CL) level, and there were 94, 41, and 102 in low-CL, medium-CL, and high-CL groups, respectively. Table 9 indicates all three groups show a positive response to the CPP tariff by reducing their coincident load during the alerted and actual monthly peak hours. This in turn benefits the utility company by reducing the system peak, hence improving the system load factor.

5

Conclusions

The goal of this paper is to develop two data-driven approaches to evaluate the effectiveness of demand response (DR) programs which are increasingly popular towards efficient power distribution systems. One evaluation framework is designed to assess the effect on annual energy savings due to a DR project by integrating an advanced “season identification” method at the weekly level and the widely used degree day analysis. The second evaluation framework is to assess the effect on coincident load due to a DR project by a carefully designed day matching algorithm. Existing evaluation methods in the DR projects mostly do not address the coincident load and are ad hoc and specific to the particular DR project in study. The two proposed approaches in this paper are not only data-driven but more importantly systematic.

Novel Data-Driven Methods for Evaluating Demand Response Programs in a. . .

305

A field demonstration DR project is used as a case study to test the efficacy of the proposed evaluation frameworks. The DR project features a community-wide critical peak pricing (CPP) tariff and a set of technologies available to participant homes including Wi-Fi-enabled thermostats, efficient heat pump water heater, ultra-treatment of home envelop, and residential battery systems. The case study demonstrates the proposed evaluation methods are effective and computationally efficient. Furthermore, this evaluation helps the DR project to conclude that (1) homes with electric HVAC systems reduce more coincident load consistently than gas powered ones; (2) homes having battery systems for energy storage perform better than the ones without such equipment; (3) the installed technologies along with the CPP tariff can significantly reduce the system peak load; (4) the DR project was successful in reducing participant homes’ annual energy consumption; and (5) the DR project achieves the goal of incentivizing residents to reduce their peak-hour consumption by implementing the CPP tariff.

6

Cross-References

A Simulation-Based Framework for the Adequacy Assessment of Integrated

Energy Systems Exposed to Climate Change Processing Smart Meter Data Using IoT, Edge Computing, and Big Data Analyt-

ics

References P. Cappers, C. Goldman, D. Kathan, Demand response in us electricity markets: empirical evidence. Energy 35(4), 1526–1535 (2010) Y. Chen, L. Zhang, P. Xu, A. Di Gangi, Electricity demand response schemes in china: pilot study and future outlook. Energy 224, 120042 (2021) Federal Energy Regulator Commission, 12 December 2022, Electric Quarterly Reports (EQR), 2021. https://www.ferc.gov/power-sales-and-markets/electric-quarterly-reports-eqr G. Conte, D. Scaradozzi, A. Perdon, M. Cesaretti, G. Morganti, A simulation environment for the analysis of home automation systems, in 2007 Mediterranean Conference on Control & Automation (2007), pp. 1–8 N.R. Council et al., Completing the Forecast: Characterizing and Communicating Uncertainty for Better Decisions Using Weather and Climate Forecasts (National Academies Press, 2006) E. Dunham-Jones, Seventy-Five Percent (2000) F. Di Maio, S. Morelli, E. Zio A simulation-based framework for the adequacy assessment of integrated energy systems exposed to climate change. Springer International Publishing, pp. 1–35 (2021). https://doi.org/10.1007/978-3-030-72322-4_125-1 J.K. Gruber, M. Prodanovic, Residential energy load profile generation using a probabilistic approach, in 2012 Sixth Uksim/Amss European Symposium on Computer Modeling and Simulation (2012), pp. 317–322 E.T. Hale, L.A. Bird, R. Padmanabhan, C.M. Volpi, Potential Roles for Demand Response in HighGrowth Electric Systems with Increasing Shares of Renewable Generation. Technical Report, National Renewable Energy Lab. (NREL), Golden, 2018 K. Li, B. Wang, Z. Wang, F. Wang, Z. Mi, Z. Zhen, A baseline load estimation approach for residential customer based on load pattern clustering. Energy Proc. 142, 2042–2049 (2017)

306

L. Bai and A. Roy

G.S. McMaster, W. Wilhelm, Growing degree-days: one equation, two interpretations. Agric. For. Meteorol. 87(4), 291–300 (1997) T.H. Pedersen, R.E. Hedegaard, M.D. Knudsen, S. Petersen, Comparison of centralized and decentralized model predictive control in a building retrofit scenario. Energy Proc. 122, 979– 984 (2017) B. Shen, G. Ghatikar, C.C. Ni, J. Dudley, P. Martin, G. Wikler, Addressing Energy Demand Through Demand Response. International Experiences and Practices. Technical Report, Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, 2012 V. Stavrakas, A. Flamos, A modular high-resolution demand-side management model to quantify benefits of demand-flexibility in the residential sector. Energy Convers. Manag. 205, 112339 (2020) B. Stoll, E. Buechler, E. Hale, The value of demand response in Florida. Electr. J. 30(9), 57–64 (2017) M. Sun, Y. Wang, G. Strbac, C. Kang, Probabilistic peak load estimation in smart cities using smart meter data. IEEE Trans. Ind. Electron. 66(2), 1608–1618 (2019) H.C.S. Thom, The rational relationship between heating degree days and temperature. Mon. Weather Rev. 82(1), 1–6 (1954) J. Torriti, M.G. Hassan, M. Leach, Demand response experience in Europe: policies, programmes and implementation. Energy 35(4), 1575–1583 (2010) K.M. Tsui, S.-C. Chan, Demand response optimization for smart home scheduling under real-time pricing. IEEE Trans. Smart Grid 3(4), 1812–1821 (2012) M. Ullah, A. Wolff, P. Nardelli Processing Smart Meter Data Using IoT, Edge Computing, and Big Data Analytics. Springer International Publishing pp. 1–15 (2021). https://doi.org/10.1007/ 978-3-030-72322-4_124-1 A. Zheng, A. Casari, Feature Engineering for Machine Learning: Principles and Techniques for Data Scientists (O’Reilly Media, Inc., 2018)

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis Marios Tsioufis and Thomas A. Alexopoulos

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Features of Electricity Markets and the Evolution of Thermal Plants . . . . . . . . . . . . . . . . 3 Market Structure Versus Market Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Revisiting the Convergence Issue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Club Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

308 309 310 316 319 324

Abstract

Electricity markets are constantly evolving in several countries, at least in the last three decades, with Western countries leading the way. In Europe, this development happened at a different pace for all its member states, resulting in different operating frameworks and convergence issues in a single market. The European directives have helped in this direction, especially after 2000. Nevertheless, due to the technical characteristics of the electricity market and its forced dependence on the rest of the energy markets, the question of the effectiveness of these measures is raised. Given these, this chapter analyzes the current situation in the EU countries, exploiting the increased pressure for demand that appeared mainly in the second half of 2021, due to the post-Covid activity, and examines the behavior of electricity markets, in clubs of countries,

M. Tsioufis Department of Economics, University of Peloponnese, Tripoli, Greece T. A. Alexopoulos () University of Peloponnese, Tripoli, Greece International Centre for Economic Analysis, Waterloo, Canada e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_154

307

308

M. Tsioufis and T. A. Alexopoulos

in terms of convergence. The results show that the operating framework is not strong enough to cope with strong pressures, and additional improvements are needed. Keywords

Club convergence · Electricity market structures · Merit order effect · Operating reserve demand curve

1

Introduction

Humanity today enjoys unprecedented levels of development and technology compared to previous centuries. It would not be surprising to say that there is already a long way to go from the current standard of living compared to the 1990s. Progress is evident in almost every area of human activity and life. Information today is done almost instantly, with distance no longer being a parameter. If we exclude harsh times for humanity, such as pandemics or wars, the life expectancy is constantly increasing. It is characteristic that in 1960 the average life expectancy worldwide was only 52 years, while in 2019, it had exceeded 72 years (The World Bank 2022). If we are limited to developed countries, it exceeds 80 years with optimistic forecasts for even more years. Beyond that, it has only been 50 years since the Internet came into our lives, and we can no longer think of our world without it. Since the first communication attempt in October 1969 through the ARPANET network, significant steps disproportionate to the time required have been achieved. In addition, the current level of technology allows us to enjoy the maximum benefit from what we use and what we do, such as our stay in big cities, the use of transport, our employment. For example, agricultural work has become more accessible and efficient in the primary sector. Transportation is currently at the crossroads of electrification, with all that positive implies for comfort and the environment. Of course, one would wonder if developments have only a positive impact on our societies. Not, but every action or new situation has positive and negative consequences, and by extension, it is something we usually accept from the beginning. A critical factor in achieving all of the above was electricity, whereas its use involves generating, transmitting, and distributing it through appropriate infrastructure and networks. Electricity is a relatively recent product or service, as one of the first commercial power stations was the Pearl Street Station in Manhattan in 1882 for street lighting (Sulzberger 2013). Electricity, however, was not limited to this use but extended to many other industrial and residential services, based on the final form of energy we wanted from it, such as mechanical, kinetic, thermal, and others. Economists have turned their attention to so-called General Purpose Technologies (GPTs). Such technologies preassume, first to be widely used in many sectors of the economy and secondly be characterized by technological dynamism, while their diffusion is said to improve “innovational complementarities.” Electricity is

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis

309

the most representative example and the critical element for the second Industrial Revolution at the beginning of the twentieth century. Its increasing use had a positive impact on reducing energy intensity (Rosenberg 1998). Electricity is one of the secondary forms of energy, which in technical terms, presupposes the conversion of other primary forms of energy, that is, additional energy losses. Nevertheless, its use was so widespread in the industrial revolution that, on the whole, it managed to reduce the energy intensity of modern economies significantly. An example here is in the steel industry, replacing the practice of open hearths with two electrically dependent furnace technologies. The impact of electricity on modern economies came through two different paths. The first was the reconstruction (electrification) of all productive activities, making full use of electricity possibilities. Electric motors are the child of electrification. The second way was the widespread use of electrical products, which significantly improved our quality of life. Of course, both of these paths took a long time to implement as they required technological advances in other fields, such as materials science. Apart from technological progress, the operation of a competitive electricity market through its liberalization or reconstruction took even more time due to several factors (Environmental performance, concerns the physical operation of power markets as well as the financial related products (Thomakos and Alexopoulos 2016; Alexopoulos 2018)). In addition, the deregulation of electricity markets did not coincide in the large western economies, as there were significant technical and financial obstacles, while the issue of their convergence will be investigated in this chapter below.

2

Features of Electricity Markets and the Evolution of Thermal Plants

Excluding the first attempt to operate small electricity networks, mainly for lighting, the first large-scale electricity providers were under state control. Until the 1970s for North America and the 1990s for Europe, electricity markets were fully integrated, with electricity generation, transmission, and distribution being carried out primarily by the only state-owned electricity company; in other words, there was a natural monopoly. The features of the electricity itself mainly shape the physical characteristics of the electrical network. Electricity is one of the most widely used energy products such as oil, coal, gas, nuclear energy, etc. Beyond that, however, it has no other resemblance to them. Its most characteristic difference is that it is a nonstorable, on a large scale, product compared to oil, natural gas, etc. In addition, electricity cannot be separated but follows the path with the lowest resistance. These two elements have shaped the way it is produced and used. In production, technological progress has been crucial in improving the efficiency of converting primary energy sources into electricity and installing higher power plants. For example, the first thermal power plants operated under low pressure, meaning a thermodynamic cycle of low pressure and low thermal efficiencies of about 10–20%. The generated power was in the order of 10 MW. Over time technological advances, such as the combustion of

310

M. Tsioufis and T. A. Alexopoulos

pulverized coal, the application of reheating technologies, the use of improved steam turbine designs, the development of new iron and steel alloys, the operation of gas stations, and combined-cycle natural gas, led to the development of power plants with a rated power of 1000 MW and efficiency significantly over 50% (Termuehlen and Emsperger 2003). This improvement is the result of the improvement of all the individual factors of the overall efficiency of a thermal power plant, as shown in the following equation. ntotal = nRankine Cycle × nTurbine × nBoiler × nAuxiliaries × nGenerator Hence, from the original low pressure and temperature stations, we are today to the super-critical and ultra-supercritical stations, while we also have a range of technologies available for improved performance and more environmentally friendly (Agraniotis et al. 2017; Kimura et al. 2011). Typically, we can mention the case of combined cycle stations. There are two cases in this category, coal gasification and gas turbine combined cycle power plants. Coal gasification combined cycle stations today achieve efficiencies of close to 50%, 15% less CO2 emissions, 30% less water use for cooling compared to conventional ultra-supercritical thermal power plants, while they are also quite flexible, allowing the use of low-quality carbon (Mitsubishi Power 2020a; Szima et al. 2021). The second category of gas turbine combined cycle stations are power stations, which reach efficiencies of 65%, have a wide range of rated power (from a few MW to over 1000 MW), reduced emissions by 50% compared to conventional stations, and fast start-up times and wide range of rated power, from a few tens to over 1000 MW (Mitsubishi Power 2020b). Overall, although there are significant variations per region globally, on average only in the last 25 years the total efficiency of thermal power plants has increased from 32.5% to 35.5% (Fraunhofer ISI 2015).

3

Market Structure Versus Market Forces

However, in addition to the evolution of thermal power plants, whether they operate as base stations or as intermediate stations, the current design of electricity markets has been shaped by the widespread use and integration of renewable power generation units in the electricity grids. RES units are now an integral part of the structure of the electricity markets. Leading the European Union and later the United States of America and Asia, renewables now cover a growing share of the electricity generation mix. In the late 1990s and early 2000s, RES as immature technologies and economically incomparable to thermal power plants entered the electricity markets under a protected regime (Banja et al. 2017). More specifically, there was no risk of selling the generated electricity, as RES covered by priority the country’s electricity load, thus shifting the entire supply curve in the wholesale electricity market to the right, leaving reduced room for all other producers. In addition, various financial compensation mechanisms were created, with the most important being Feed-in Tarrifs and Feed-in Premiums (among others there were also the

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis

311

so-called green certificates). Today the price they enjoy results from auctions of specific quantities of power per RES technology (Ragwitz et al. 2014; IRENA and CEM 2015). For example, the German Federal Network Agency recently auctioned 1319 GW of wind farms with a maximum price of 0.0588 Eur/kWh, and it aims to auction a total of 4 GW for 2022 (Ivanova 2022). The European Union promotes a common RES financing mechanism; through it, each member state will participate by financing a joint RES project account or by agreeing to host European RES projects on their territory (Blucher et al. 2020). The penetration of RES in the electricity networks changed their modus operanti and, at the same time, created counterbalancing forces. More specifically, the entry of RES initially resulted in the so-called merit order effect. This changed the classic curve of the daily electric load. Thus, from the traditional electricity demand behavior, that is, the peak at noon and the sinking of the evening hours (the bell curve), we went to a leveling of this as a result of the merit-order effect (Sensfuß et al. 2008). This change occurred precisely because of the scattered production of RES, which peaks at the same noon hours when there is a high demand for electricity by human societies. The aggregation of the electricity supply curve in the wholesale market of the next day also called the merit order stack is essentially the result of ordering the offered power by ascending marginal cost. In the literature, the marginal operating costs of power plants, from the smallest to the largest, are as follows:

MCCRES > MCNUCLEAR > MCLIGNITE > MCCOAL > MCNAT

GAS

> MCOIL

Figure 1 shows a stylized example of the merit order stack. Setting aside the structure of the electricity market, that is, the technical constraints imposed by the very nature of the market, we can argue that economic

Fig. 1 Illustrative example of a merit order Stack. (Source: Own elaboration)

312

M. Tsioufis and T. A. Alexopoulos

forces will reduce the daily average marginal price of the system, as the entry of RES, with almost zero marginal cost, will displace the more expensive facilities. In the short term, this has indeed happened to a large extent. At the beginning of the last decade, there has been a reduction in the system’s marginal value in several European electricity markets. In literature, depending on the degree of penetration of RES, the short-term merit order effect has been estimated from 1.7 Euro/MWh to 7 Euro/MWh (Sensfuß et al. 2008; Saenz de Miera et al. 2008). On the long term though, that is, when all the available power has been adapted to the supply shock from the input of RES, the merit order effect is limited, if it exists at all. The distinction between short-term and long-term analysis is necessary as conventional power plants take a long to adapt. Thus, according to Lamont (2008), the longterm merit order effect fails to reduce the system’s price, while Weber and Woll (2007) conclude that the long-term merit order effect is zero or even slightly positive. In addition, Green and Vasilakos (2011), in their analysis of an energy system configured for Great Britain, find that “the changes in conventional capacity effectively offset the impact of wind generation on prices, and so the time-weighted average price is similar in the equilibria we find with and without wind generation.” Finally, the above conclusions are reached by more recent research by Werner and Muesgens (2021), which characterizes the merit order effect as a temporary phenomenon created by the slow adjustment rate of electrical power. Their research also concludes that concerns about power plants facing the risk of being considered a stranded asset are short term, and market forces will eventually counteract the redundant capacity. The following figure shows the marginal price for the Greek electricity market from the beginning of 2009 until the middle of 2019. We select this period, as in these years, we had the majority of RES penetration in the system, allowing us to observe both the short-term and the long-term merit order effect. In addition, to make this easier to observe, we have created 24 time series (Fig. 2), one for each hour of the day, and we have calculated the corresponding monthly moving averages of these series as a smoothing filter.

Fig. 2 Historical time series of the Greek Day-Ahead Market per hour. (Source: Own elaboration)

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis Table 1 Production fuel mix of Greece

Type of producer Lignite Natural gas Renewables Hydropower Interconnections

2009 60% 12.96% 3.36% 11.6% 8.47%

2014 49.27% 11.87% 20.55% 6.35% 11.97%

2016 21.6% 19.8% 23.15% 11.52% 24%

313 2020 10% 30% 36% 6% 18%

Source data: ipto (2022)

We observe a decrease in the average daily marginal price, as the difference between the cheapest and most expensive time decreases significantly (curve flattening) from 2013 onwards. In other words, in the short term, when we have no capacity adjustments, the input of RES works beneficially in lowering the system’s price. However, from 2016 onwards, there has been an increasing trend in the system’s price. This could be explained by the end of the short term, transitional period and the shift to the new permanent operation of the system. Table 1 illustrates, to some extent, the change from the transition period to the permanent state of power capacities. Although we should include the corresponding capacity factors for each source of electricity production, we can nevertheless conclude the produced energy mix in our analysis. Thus, we see that the energy production mix in 2016 is closer to the current production mix, in contrast to the production mix of 2014 which is more similar to that of 2009. In addition, we note that while RES already in 2014 had increased their share in the electricity production mix, the other sources, except for hydroelectric production, generally maintained their percentages. On the contrary, from 2016, an adjustment in power capacity (based on the produced energy) is apparent, resembling more the permanent state of 2020 than of 2014. Regarding the electrical interconnections, we also observe an increase in the long run, taking advantage of the flexibility they offer in the system. At the same time, as the production of RES is intermittent and has a stochastic in nature production, the system’s requirements for flexibility increased. Therefore, the more RES power enters the system, the more flexibility is needed to address the added uncertainty and the variability. By flexibility, we mean the system’s response to the abrupt changes in the generated electricity. The International Energy AgencyInternational Smart Grids Action Network (IEA-ISGAN) considers flexibility as the ability of the power system to respond to changes (Hillberg et al. 2019). According to ela et al. (2014), on a resource scale, flexibility is “the ability of a resource, whether any component or collection of components of the power system, to respond to the known and unknown changes of power system conditions at various operational timescales.” Standard features of thermal power plants that have to be considered for balanced cooperation with RES are the minimum generation levels (technical minimum), ramp rates, startup/shutdown time, and minimum up/down time. Combined cycle gas turbines offer the best performance on this effort and are crucial for higher levels of Res penetration in the system. Besides this, other alternatives include the development of large scale storage facilities (pumped hydroelectric storage reservoirs, spinning flywheels, and compressed air),

314

M. Tsioufis and T. A. Alexopoulos

Fig. 3 The duck curve effect in CAISO. (Source: Own elaboration)

active demand with demand response actions with smart meters, smart tariffs and smart connected devices, and last but not least, more in numbers and higher in power capacity, electrical interconnections. While the above-mentioned actions are to deal with variable renewable generation in real-time, an additional need for increased reliability for the whole system has emerged. At first, this new technical requirement was addressed with long-term capacity reserve mechanisms or forward markets. Given low levels of RES penetration, participants in reserve markets had encountered no serious problems. The increased penetration of RES in the system altered the usual net daily load curve, created the so-called “duck curve” (Fig. 3), and started a struggle among what we call market forces versus market structure. Roughly speaking, the low marginal cost of RES, on the one hand, results in excluding flexible CCGT thermal units from the system (market forces), but on the other side, RES penetration requires the inclusion of more flexible units like CCGT (market structure). This contradictory situation disincentivizes further natural gas power plants installations, as it creates no clear revenue streams, creating the “missing money problem.” Furthermore, EU countries do not have a unique capacity remuneration mechanism to counteract this risk. Among the different price and volume-based approaches are capacity markets, capacity payments, reliability auctions, capacity obligations, strategic reserves, combinations of them, or no capacity remuneration (energy only market). Recent literature has demonstrated this problem as a core issue for further RES penetration. Towards that direction, MISO (2013) has

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis

315

theoretically introduced the idea of the operating reserve demand curve (ORDC), which enables the theory of scarcity pricing in electricity procurement. The concept of ORDC was first developed by Stoft (2002) and Hogan (2005). They suggested that its incorporation would consolidate reserve and energy markets into an energyonly market operating on co-optimization. The first actual implementation came from the Electric Reliability Council of Texas (ERCOT) in 2014. ORDC’s mechanism is based on the Loss of Load Probability (PLoss _ Load ), on the price value of lost load (PLL) and the probability of shortage event (Pshortage _ event ). It is a function of the minimum contingency reserve levels (ML) and the available reserve levels (RL), and it uses normal probability distributions. Pshortage _ event becomes a certainty when RL is below ML. Hence, we have the following: PLoss Load (RL − ML) → ℵ (μ, σ ) Pshortageevent (RL) =

PLossLoad (RL − ML) , RL − ML ≥ 0 1, RL − ML < 0

P rice_adder ORDC = Pshortageevent (RL) ∗ P LL − mcsecurityconstrained mcsecurityconstrained is the marginal cost paid by electricity consumers, also called the security-constrained economic dispatch. The above mechanism results in a higher price adder when reserve levels decrease but stay above the minimum reserve threshold and a lower price adder in the opposite case. ORDC improves price markups and ensures, to a certain extent, the necessary remuneration of the realized investment costs of flexible units like CCGT thermal units, since in their presence, prices reflect actual scarcity conditions. Nevertheless, a recent study by Bajo-Buenestado (2021), using an ARDL model with seasonal dummies, finds that ORDC prices are significantly negatively affected by wind generation levels when wind generation is relatively low. Also, more wind capacity installations increase the probability of zero ORDC prices in the ERCOT market. If combined with further price distortions coming from emergency policies, like ERCOT’s Reliability Must Run (RMR) policy, ORDC prices became less effective, additional capacity mechanisms and/or forward electricity markets might be needed for corrective adjustments. Cooperation of these mechanisms with an ORDC system is realistic (Hogan 2015), while Zhou and Botterud (2014) has proposed modifications to reduce the impact of further wind capacity installations. Another case is acknowledging relative linear marginal step-up costs of all the different electricity producers, forming the merit order stack (supply curve). This hypothesis was relatively low in interest until the last events in the Russian-Ukraine war. Regardless of the true origin, we witness significant increases and short-term

316

M. Tsioufis and T. A. Alexopoulos

variations in gas prices, resulting in nonlinearities in the merit order curve. This should not stay unattended by the regulatory authorities. It creates severe inframarginal rents for the rest of the producers, above even and the relatively accepted ones needed to overcome fixed costs. As with natural gas prices decreasing in the past, today’s high prices constitute a stressor for wholesale and retail electricity markets, creating adverse conditions as with excess market power. Further regulatory actions to decouple gas prices with electricity ones need to be taken. National level subsidies would distort price signals and should be the last surviving mechanism. Instead, the classic arbitrage mechanisms should stay on the table as they remain the least invasive tool. Any inframarginal rent could be used to cover the predicament situation of the most stressed producers who clear the system’s marginal price each time. Usually, it is the gas-based producers. Of course, further consideration is needed for a mature stabilization mechanism having the least invasive effect on price signals both in energy and reserve markets. Given all the above-mentioned mechanisms, the question arises if electricity markets are evolving in a converging fashion or if there are persisting differences depending potentially on the initial market conditions, the inherited fuel mixes, and/or the applied regulatory policies. It is of essence to identify the evolving paths of electricity markets, a question we attempt to answer in this chapter with a club convergence analysis, because electricity, as a secondary energy source, is equally essential for modern economies as the production factors of labor or capital. Irrespective of the shifts and changes in electricity markets, it has to be mentioned that RES has made a positive contribution against energy poverty, especially in areas where there is no or very restricted electricity transmission network. Such areas are the sub-Saharan countries or the underdeveloped countries. With the widespread use of RES, it is now possible to create microgrids, which do not necessarily have to be connected to a central transmission network, giving access to electricity in societies in dire need.

4

Revisiting the Convergence Issue

In the last two decades, important policies have been implemented for the liberalization (deregulation) of the electricity market to maximize the social surplus, render market operation in full competition, and technological improvement with more efficient electricity production. European Directives 96/92 / EC, 2003/54 / EC, 2009/72/ EC and most recently 2019/944/EC have contributed to this. Before these directives, most European electricity markets followed their evolutionary path. Typically we can mention the United Kingdom (before Brexit), which with its privatization in 1990, adopted the concept of a pool Market and central dispatch mechanism. Unfortunately, this created a duopoly with subsequent high markups on marginal cost and substantial deadweight losses, which, in turn, created dissatisfaction among the UK regulator and the Parliament. As a response, The Electricity Pool of England and Wales was replaced in 2001 by the New Electricity

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis

317

Trading Agreement (NETA), an energy-only market. The change in the electricity generation mix, which was already taking place since the first legislative act, became even more prominent after NETA, replacing coal with natural gas (dash for gas) and eliminating the use of oil. However, it failed to deliver adequate RES installations and provide incentives for these investments. Thus, the market was restructured again in 2013 with the Energy act of 2013, whose part was the Electricity Market Reform (EMR). After the NETA, the margin between coal or gas prices and electricity prices was reduced and eventually became coupled. The lessons from UK’s reform are that long-lasting investments, like nuclear and combined capture and storage facilities, need contractual support, and RES support works well now. At the same time, current regulation sends good investment signals and is innovative in supporting the move to the smart network economy but is less good at agile tariff adjustments. Another important electricity market for its convergence path is the Nordic market. The Nordic market has three dimensions-characteristics; first, it is the first institutionalized international electricity market, with Baltic states being the last entry in 2010. Second, it is the market where the zonal pricing model was first introduced (15 price zones), and last, it relies low on fossil fuels, with hydro generating half the electricity needed. Still, it faces issues like the future planning of nuclear power or the efficiency of the Norway-Sweden green certificate scheme. Many share the opinion that the integration of EU electricity markets in a common energy market has been achieved to a satisfactory level. Nevertheless, what we are still missing of is a single approach for a capacity remuneration mechanism and a RES support scheme. The question that arises is if these differences do not allow for real convergence of electricity prices across the EU. Given the concurrent strong turbulences in electricity price and the liquid overall regulatory and policy context, this issue becomes more relevant. Further speaking, the early 2000s context was targeting on competition and market integration with a focus on day-ahead wholesale markets, while the current framework interests more in decarburization investments and focuses more on intraday and real-time markets to manage variable res growth. Today, we are concerned more on fixed costs of decentralized technologies rather than variable costs technologies and new network investments instead of the optimized use of existing infrastructure. Last, active demand participation is prevalent among energy policymakers, and the importance of the concept of prosumers is recognized as a key element for further RES penetration. While the EU has performed steps to what is called competition “in” the market, it still needs additional acts and measures for competition “for” the market. In literature, the argument of electricity market convergence examines the Law of One Price (LOP), suggesting a perfectly competitive market. Theoretically, we expect this to happen as competition increases. Nevertheless, due to the distinctive characteristics of electricity markets, as explained above, convergence might not be achieved for all EU markets to the same degree and at the same time. It is expected that neighboring markets with significant power flow interconnections will reach a sort of convergence sooner. Zonal pricing

318

M. Tsioufis and T. A. Alexopoulos

helps in this direction as it balances significant differences that exist in the case of a single national system price. Earlier studies use a variety of methodologies to investigate the convergence issue, including cointegration analysis, principal component analysis, unit roots tests, localized autocorrelation functions, machine learning techniques such as the random forest algorithm, and others. These analyses concern either the EU countries as a whole or bilateral tests between countries in pairwise mode. Zachmann (2008) tests the hypothesis of convergence due to the ongoing restructuring process in the EU electricity sector. He finds no full market integration but pairwise convergence among the EU countries; in addition, he shows that convergence exists in the majority of off-peak hours, while the opposite is true for peak hours. Saez et al. (2019) examine market integration in central west Europe after implementing the Flow-Based Market coupling (FMBC). FMBC is the best approach for integration (European Union 2015), as it allows electricity to flow from a cheaper to a more expensive market until an equilibrium is reached, resembling the communication vessels. They conclude that RES penetration can contradict market integration unless transmission capabilities are further adapted to facilitate cross-border exchanges and work with renewable sources. In another approach, Menezes et al. (2016) assess the time-varying dynamics of electricity prices for the British, French, and NordPool markets. Their results reveal both stationary and nonstationary periods. When a trend exists, this might reflect a trend in fuel prices; similar results are found in Alexopoulos (2017) for retail electricity prices. A recent analysis (Cassetta et al. 2022a, b) investigates the residential and industrial end-users convergence, adopting the notion of club convergence. They find multiple clubs and different convergence paths between domestic and nondomestic users. They relate the existence of various clubs to differences in public intervention and regulation. In a second study, Casetta et al. (2022a, b) address the case of simultaneous retail electricity and gas convergence. Their three-stage analysis identifies four clusters of EU countries and shows that a slow beta-convergence is not related to sigma and club convergence. This rationale lies in country-specific factors such as national energy and climate policies blocking the convergence process at the retail price level. In general, literature examines the convergence issue from different perspectives and empirical methods. In this chapter, we attempt to explore price convergence among the wholesale EU electricity markets, considering the impact of the price shock of natural gas on electricity markets. Though it is given low attention, the aggregate electricity supply function of the day-ahead market is, in essence, a patch of other markets of primary energy sources. This was not a problem, but after the latest unprecedented gas price escalations, it is crystal clear that electricity prices are more vulnerable as though before, setting a new perspective when studying the convergence of EU electricity markets. To that end, we apply a club convergence analysis separating the examined period in two states, one before gas price shock and one while in shock.

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis

5

319

Club Convergence

To avoid assumption issues related to trend stationarity or stochastic nonstationarity but also to allow heterogeneous agent behavior, we select the Phillips and Sul (2007) approach of convergence. Cointegration tests are suffering from the existence of unit roots in the differential series, leading to a false true divergence result. In addition, due to their idiosyncratic features, it is a highly improbable event, for European electricity markets to reach to a single convergence point, rendering the concept of club convergence the best fit. Phillips and Sul, instead of dividing the examined cross-sections in subgroups based on some prior variable, like location, economic status, etc., identify potential clusters, or clubs as they call them, within their proposed convergence algorithm. Following their approach, we use a nonlinear panel model for the Day-ahead prices, DHit , and decompose it, to a common path κ t and an individual component δ it , which measures the distance of power market i from the common trend behavior, as follows: DH it = δit κt If the lim δit → δ for all power markets, convergence exists. To test this, the t→∞ following ratio is considered: hit =

DHit 1 N i=1 DH it N

=

δit 1 N i=1 δi N

hit is the relative transition path, featuring the distance of the ith power market from κt. Phillips and Sul implement the following OLS model to test the convergence hypothesis log (H1 /Ht ) − 2log (log(t)) = α + βlogt + εt , t = rT , rT + 1, . . . , T log(t) has been selected based on the least distortion and the best test power, r equals to 0.2 as T = 4380 > 100. The null hypothesis of convergence is the one-sided t-test Ho : β ≥ 0 against the Ha : β < 0, based on heteroscedasticity and autocorrelation consistent standard errors εt . The log(t) test is then used as in the data-driven algorithm of Phillips and Sul. The steps are: Cross − section sorting → Core group F ormation → Club membership preparation → Recursion of the log(t) test → Club merging

320

M. Tsioufis and T. A. Alexopoulos

Table 2 Descriptive statistics for the European day-ahead prices Country AUS BEL BG CH CRO CZE DE-LU ESP FRA GR HOL HU IRE ITA LAT LITH POL POR ROM SER SLK SLV SWE

First half 2021 Mean Std. Dev. Min Max 58.80 20.44 −66.18 138.05 56.94 23.24 −66.18 139.48 113.56 37.51 0.10 273.68 61.21 17.87 −79.40 137.48 60.85 22.26 −263.31 139.07 59.53 19.08 −36.26 139.72 55.27 23.37 −69.00 139.72 58.89 2.58 0.01 121.24 58.75 21.17 −66.18 163.77 62.29 19.74 −0.01 144.64 56.68 21.39 −66.18 136.71 60.86 19.55 −35.00 153.47 81.80 40.17 −20.43 500.00 66.86 0.00 3.00 139.07 54.27 25.91 −1.41 255.00 55.83 25.33 −1.41 255.00 62.60 13.48 12.20 119.12 58.91 25.74 0.00 121.24 54.00 19.77 0.02 122.12 59.85 19.42 1.03 145.15 59.70 19.26 −36.26 139.72 61.47 20.53 −66.18 139.07 47.23 21.02 −1.41 249.98

Second half 2021 Mean Std. Dev. 155.04 83.80 151.43 87.43 311.91 150.89 168.85 89.47 168.70 8.07 141.89 80.89 138.53 82.87 165.14 69.82 159.75 93.27 170.73 72.34 149.38 80.19 167.01 8.24 173.41 65.66 183.04 76.80 123.34 80.30 125.13 82.76 111.51 55.25 165.28 69.72 33.15 16.80 168.35 79.66 145.91 8.24 168.77 80.69 99.35 74.17

Min −63.03 −70.00 18.97 −53.85 1.04 −1.22 −63.03 0.90 −63.03 9.70 −63.03 3.77 0.00 50.06 0.04 0.04 24.12 0.90 0.77 8.09 −1.22 1.04 −1.97

Max 620.00 620.00 1061.04 532.21 533.19 620.00 620.00 409.00 620.00 542.50 620.00 620.00 500.00 533.19 1000.07 1000.07 529.20 409.00 110.24 539.91 620.00 533.19 626.06

Because 2021 was a mixed year, with power markets operating both in normal clearing prices and under price shocks, we implement the club convergence analysis independently in the first and second half of this year. Doing so enables us to extract useful policy conclusions from the convergence behavior under different conditions. The different conditions, for all the European power markets, between the first and the second half of 2021 are apparent in Table 2 descriptive statistics. Not only are prices higher overall, but they vary significantly more, adding more risk in the markets. Furthermore, in Table 3, we decompose day-ahead prices into a between has been included for (the overall mean DH i , and a within reasons of comparison) component. Again, the results verify the general increase of day-ahead prices and the higher variation compared to the first half of 2021. It is worth mentioning that the variance between countries is half of the within a country variance. In other words, day-ahead prices in EU markets have a closer behavior when they are looked at a country level, compared to a country-hour overall level or a within a country level. Also, we notice that the proportion of variations is nearly equal, regardless of the studied period. The “within” a country variation is double

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis

321

Table 3 The “within” and “between” components of the day – ahead prices

Overall Between Within Overall Between Within

First half 2021 Mean Std. Dev. 61.94 26.37 12.82 23.28 Second half 2021 153.74 94.29 47.38 82.04

Min −263.31 47.23 −262.23

Max 500. 113.56 480.14

Observations N = 100,607 n = 23 T = 4374

−70. 33.15 −139.19

1061.04 311.9 1030.47

N = 98,854 n = 23 T = 4298

Fig. 4 EU’s power markets day-ahead prices in the second half of 2021. (Source: Own elaboration)

in magnitude compared to “the between” counties variation both in the first and the second half of 2021. Same conclusions are reached in the Figs. 4 and 5 that follow. Club convergence results are on the same page. Following Du’s (2017) empirical work, we first remove the cyclical component and investigate club convergence on the trend component, using the band pass filter of Lawrence and Fitzgerlad (2003). Similar time series are obtained with other filtering methods like the Baxter King

322

M. Tsioufis and T. A. Alexopoulos

Fig. 5 EU’s power markets day-ahead prices in the first half of 2021. (Source: Own elaboration)

(BK), the Hodrick Prescott (HP) or the Butterworth (BW). We then run the log t convergence test for the first and the second half. The results in Table 4 report the coefficient, standard error and t-stat for log(t). Since the value of t-statistic is more than −1.65 in the first half and less in the second half, we reject at the 5% level, the null hypothesis of convergence only for the second half year of 2021. Nonconvergence in the second half dictates us to examine possible converging clubs among EU countries. As shown, in Table 5. EU power markets seem to converge in four clubs and there are four counties that are not converging to any of these clubs, forming a fifth nonconverging group. Overall, convergence results show that the European deregulation directives, in an attempt to couple all the wholesale electricity energy markets operating under the same competition playbook, have been achieved. Nevertheless, a shock in a single agent in the merit order stack (supply curve), which happened in the gasbased power producers in the second half of 2021, weakens aggregate convergence to converge clubs and nonconvergent groups. As shown in Alexopoulos (2017),

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis

323

Table 4 Log t convergence testq results log t convergence test First Half of 2021 Second Half of 2021

Variable Log(t) Variable Log(t)

Coeff. −0.1389 Coeff. −0.5233

SE 0.2246 SE 0.2194

T-stat −0.6186 T-stat −2.386

Table 5 Club convergence for EU power markets for the second half of 2021 Club classifications Club 1 Club 2 Club 3 Club 4 Not convergent group 5 a Germany

#countries = 7 AUS, FRA, GR, HU, IRE, ITA, SER #countries = 6 BEL, BG, CH, CRO, HOLL, SLV #countries = 4 ESP, POL, POR, SWE #countries = 2 DE-LUa , ROM #countries = 4 CZE, LAT, LITH, SLK

and Luxemburg share the same day-ahead prices

natural gas as a primary energy source for electricity production seems to have a growing importance for electricity prices, as gas-based producers usually clear dayahead prices in wholesale markets. Physical interconnections or neighboring conditions could explain, to some extent, these converging clubs. We understand why Italy, Greece, and perhaps Austria belong to the same club. Still, there is no apparent reason why the Lithuanian and Latvian power markets do not converge to Club 3, which includes Sweden, as they all participate in the Nordic market. A first rationale could be that the market structure, expressed by the specific fuel mix each country has for producing electricity, still has the upper hand in price formation over the known market forces of competition. Also, there is no evidence of how the different European capacity mechanisms respond to similar shocks, in terms of price signals, in the wholesale power markets. Capacity remuneration mechanisms with different exposure to market conditions, such as strategic reserves, will have different spillover effects on wholesale energy markets, compared, for instance, to the ORDC mechanism, in case of a shock in primary energy sources needed for electricity production. Future research is needed to examine the exact effects of these mechanisms and how they produce unpredictable and eventually unwanted divergence behaviors among the EU power markets.

324

M. Tsioufis and T. A. Alexopoulos

References M. Agraniotis, C. Bergins, M. Stein-Cichoszewska, E. Kakaras, High-efficiency pulverized coal power generation using low-rank coals, in Low-Rank Coals for Power Generation, Fuel and Chemical Production, (Elsevier Ltd., 2017). https://doi.org/10.1016/B978-0-08-1008959.00005-X T.A. Alexopoulos, The growing importance of natural gas as a predictor for retail electricity prices in US. Energy 137, 219–233 (2017). https://doi.org/10.1016/j.energy.2017.07.002 T. Alexopoulos, To trust or not to trust? A comparative study of conventional and clean energy exchange-traded funds. Energy Econ. 72, 97–107 (2018) W. Antweiler, F. Muesgens, On the long-term merit order effect of renewable energies. Energy Economics 99, 105275 (2021). https://doi.org/10.1016/j.eneco.2021.105275 R. Bajo-Buenestado, Operating reserve demand curve, scarcity pricing and intermittent generation: Lessons from the Texas ERCOT experience. Energy Policy 149, 112057 (2021). https://doi.org/ 10.1016/j.enpol.2020.112057 M. Banja, M. Jégard, F. Monforti-Ferrario, J. Dallemand, Renewables in the EU: The support framework towards a single energy market EU countries reporting under article 22(1) b, e and f of renewable energy directive. 22(1) (2017). https://doi.org/10.2760/69943 F. Blucher, et al., The new renewable energy financing mechanism of the EU in practice. Aures II, no. 817619. (2020) E. Cassetta, C.R. Nava, M.G. Zoia, EU electricity market integration and cross-country convergence in residential and industrial end-user prices. Energy Policy 165, 112934 (2022a). https:// doi.org/10.1016/j.enpol.2022.112934 E. Cassetta, C.R. Nava, M.G. Zoia, A three-step procedure to investigate the convergence of electricity and natural gas prices in the European Union. Energy Econ. 105 (2022b). https:// doi.org/10.1016/j.eneco.2021.105697 L.J. Christiano, T.J. Fitzgerald, The band pass filter. Int. Econ. Rev. 44, 435–465 (2003) L.M. de Menezes, M.A. Houllier, M. Tamvakis, Time-varying convergence in European electricity spot markets and their association with carbon and fuel prices. Energy Policy 88, 613–627 (2016). https://doi.org/10.1016/j.enpol.2015.09.008 K. Du, Econometric convergence test and Club clustering using Stata. Stata J. 17(4), 882–900 (2017). https://doi.org/10.1177/1536867X1801700407 E. Ela, M. Milligan, A. Bloom, A. Botterud, A. Townsend, T. Levin, Evolution of wholesale electricity market design with increasing levels of renewable generation. NREL/TP-5D0061765. Golden (2014) European Union, Establishing a guideline on capacity allocation and congestion management. Off. J. Eur. Union (2015). https://doi.org/10.3000/17252555.L.2009.200.eng Fraunhofer ISI. (2015). How Energy Efficiency Cuts Costs for a 2-Degreee Future. November, 100 R. Green, N.V. Vasilakos, The Long-Term Impact of Wind Power on Electricity Prices and Generating Power (ESRC Centre for Competition Policy Working Paper Series). (2011) E. Hillberg, B. Herndler, S. Wong, J. Pompee, J.-Y. Bourmaud, S. Lehnhoff, et al., Flexibility Needs in the Future Power System (IEA Energy Technology Network, 2019) https://doi.org/10.13140/ RG.2.2.22580.71047 W.W. Hogan, On an “Energy Only” Electricity Market Design for Resource Adequacy (Technical report, Center for Business and Government, John F. Kennedy School of Government, Harvard University, 2005) W.W. Hogan, Electricity market design energy and capacity markets and resource adequacy, in Proceedings of EUCI conference on capacity markets: Gauging their real impact on resource development and reliability, 2015 Ipto, Monthly Energy Reports (2022) [online]. Available at: https://www.admie.gr/en/market/ reports/monthly-energy-balance?since=01.04.2014&until=31.05.2020&op=Yoβoη&since1= ´ &until1=&page=3. Cited 02 March 2022

On the Coupling of the European Day-Ahead Power Markets: A Convergence Analysis

325

IRENA and CEM, Renewable Energy Auctions – A Guide to Design (IRENA and CEM, 2015), pp. 123–156. www.irena.org A. Ivanova, Germany Launches 1.32-GW onshore wind tender. RenewablesNow (2022) [online]. Available at: https://renewablesnow.com/news/germany-launches-132-gw-onshorewind-tender-777960/ H. Kimura, T. Sato, C. Bergins, S. Imano, E. Saito, Development of technologies for improving efficiency of large coal-fired thermal power plants. Hitachi Rev. 60(7), 365–371 (2011) Lamont, D. Alan, Assessing the long-term system value of intermittent electric generation eration technologies. Energy Econ. 30(3), 1208–1231 (2008) MISO, FERC Electric Tariff, Schedule 28, Demand curves for operating reserve, regulating and spinning, 4.0.0. Technical report. Midwest Independent Transmission 2013 Mitsubishi Power, Integrated Coal Gasification Combined Cycle (IGCC) Power Plants (Mitsubishi Power, Ltd, 2020a) https://power.mhi.com/products/igcc P.C.B. Phillips, D. Sul, Transition modeling and econometric convergence tests. Econometrica 75, 1771–1855 (2007) Mitsubishi Power, Gas Turbine Combined Cycle Power Plants (GTCC) Power Plants (Mitsubishi Power, Ltd, 2020b) https://power.mhi.com/products/gtcc M. Ragwitz, A. Held, J. Winkler, C. Maurer, G. Resch, M. Welisch, S. Busch, Auctions for Renewable Energy in the European Union Auctions for Renewable Energy in the European Union. Study on Behalf of Agora Energiewende. (2014) N. Rosenberg, The role of electricity in industrial development. Energy J. 19(2), 7–24 (1998) http://www.jstor.org/stable/41322772 Y. Saez, A. Mochon, L. Corona, P. Isasi, Integration in the European electricity market: A machine learning-based convergence analysis for the Central Western Europe region. Energy Policy 132, 549–566 (2019). https://doi.org/10.1016/j.enpol.2019.06.004 G. Sáenz de Miera, P. del Río González, I. Vizcaíno, Analysing the impact of renewable electricity support schemes on power prices: the case of wind electricity in Spain. Energy Policy 36(9), 3345–3359 (2008) F. Sensfuß, M. Ragwitz, M. Genoese, The merit-order effect: A detailed analysis of the price effect of renewable electricity generation on spot market prices in Germany. Energy Policy 36(8), 3086–3094 (2008) S. Stoft, Power System Economics: Designing Markets for Electricity (Wiley, 2002) Sulzberger, The Pearl Street Generating Station, 1882. IEEE Global History Network, January 1881, (2013) [online]. Available from: http://www.ieeeghn.org/wiki/index.php/ Milestones:Pearl_Street_Station. Cited 2 Mar 2022 S. Szima, C. Arnaiz del Pozo, S. Cloete, S. Fogarasi, Á. Jiménez Álvaro, A.-M. Cormos, C.-C. Cormos, S. Amini, Techno-economic assessment of IGCC power plants using gas switching technology to minimize the energy penalty of CO2 capture. Clean Technol. 3, 594–617 (2021) https://doi.org/10.3390/cleantechnol3030036 H. Termuehlen, W. Emsperger, Clean and Efficient Coal-Fired Power Plants Clean and Efficient Coal-Fired Power Plants Development Toward Advanced Technologies (ASME Press, New York, 2003) The World Bank. Life expectancy at birth [online]. Available from: https://data.worldbank.org/ indicator/SP.DYN.LE00.MA.IN?most_recent_value_desc=true. Cited 2 Mar 2022 D.D. Thomakos, T. Alexopoulos, Carbon intensity as a proxy for environmental performance and the informational content of the EPI carbon. Energy Policy 94, 179–190 (2016) C. Weber, O. Woll, Merit-Order-Effekte von Erneuerbaren Energien – Zu schön um wahr zu sein? EWL Working Papers, https://EconPapers.repec.org/RePEc:dui:wpaper:0701 (2007) G. Zachmann, Electricity wholesale market prices in Europe: Convergence? Energy Econ. 30(4), 1659–1671 (2008). https://doi.org/10.1016/j.eneco.2007.07.002 Z. Zhou, A. Botterud, Dynamic scheduling of operating reserves in co-optimized electricity markets with wind power. IEEE Trans. Power Syst. 29(1), 160–171 (2014)

Recent Developments in the Smart Energy Systems Adil Wazeer and Apurba Das

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Concept of Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Key Technological Developments for Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . 3.1 Intelligent Novel Energy Power Generation Forecast Technology . . . . . . . . . . . . . . 3.2 Advanced Energy Storage Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Digital-Energy Integrated Market and Service Mechanism . . . . . . . . . . . . . . . . . . . . 4 Application of IoT in Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Application in Policies for Energy Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Application in Energy Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Application in Grid Control and Power Generation . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Application in Actuation and Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Application in Data Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Application in Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7 Application in Real-Time Environment Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . 4.8 Recent Advancement in IoT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Case Study: Smart Energy Towns in China . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Constraints to Build China’s SET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Driving Force for China’s SET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Recommendations to Build China’s SET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Challenges in Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Future Prospects in Smart Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

328 329 330 330 331 333 333 334 334 335 335 336 336 337 337 338 339 339 340 340 341 342 342

A. Wazeer () School of Laser Science and Engineering, Jadavpur University, Kolkata, West Bengal, India A. Das Aerospace Engineering and Applied Mechanics Department, IIEST-Shibpur, Howrah, India © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_173

327

328

A. Wazeer and A. Das

Abstract

The energy crisis and pollution have accelerated the energy system’s change. To speed the design of smart cities and accomplish the sustainable objective, smart energy systems have gained major funding and expansion. Recently, terms Smart Energy and Smart Energy Systems are utilized for describing a comprehensive perspective than “Smart grid.” Smart Energy Systems, unlike Smart Grids, concentrate on much more humanistic approach that includes more areas (heating, electricity, cooling, buildings, industry, and transportation), allowing for the recognition of much feasible and inexpensive solutions for the transition to potential sustainable power solutions. Smart energy systems, which incorporate several energy industries, are seen as a possible framework for delivering a complete and optimum solution for a feasible, economical, and long-term power system. Despite much research into the concept, execution, and efficiency of these systems, designing and managing a smart energy system continues to be an issue. This chapter addresses Smart Energy Systems in light of recent advancements in the sector. Keywords

Smart Energy · Smart Energy Systems · Smart Grid · Energy Storage · Smart Cities · Clean Energy

1

Introduction

The universal energy consumption is quickly increasing, and the primary source is fossil fuels. Because fossil fuels are unsustainable and have negative environmental consequences, new techniques and mindsets are critically needed to establish an efficient system in the coming years (Su 2020). In the year 2012, the term “smart energy systems” was coined for characterizing anticipated novel approach for energy systems that will integrate diverse energy domains (Lund et al. 2012). The popularization of renewable energy, improved understanding of energy conservation, and the predicted rise in electricity bills all contribute toward usage of smart energy systems. By synchronizing demand along production and incorporating diverse energy domains, smart energy systems hope of having effective management as well as usage of power (Reynolds et al. 2017). Several energy-related features are regarded to be highly significant for smart energy systems engineering, notably conserving energy, smart utility metering, green architecture, local cooling mechanisms, adaptive monitoring, plus automating (Koutra et al. 2018). Majority of renewable power supplies (like wind and solar energy) are inconsistent, also power consumption fluctuates greatly as well (Aghamolaei et al. 2018). Both of these issues make an energy system’s equilibrium and durability more difficult to maintain. As a consequence, a superior smart energy system is more than just a collection of subsystems that communicate in sophisticated ways. There are several scientific papers accessible regarding concept, strategy, and execution of smart energy systems. Numerous current review studies on smart

Recent Developments in the Smart Energy Systems

329

energy systems cover concerns such as related ideas, subsystem configuration, and energy management control. Shao and team investigated components and subsystems for smart energy systems, as well as the methodologies for calculating and analyzing flow of energy in characteristic smart energy systems (Shao et al. 2016). Gayathri examined the energy system concerning standpoint of smart energy storage plus described energy storage system control methodologies (Venkataramani et al. 2016). Lund and co-researchers examined current state of analysis on the major smart energy subsystems (Lund et al. 2017). Among the most difficult aspects of developing smart energy system is figuring out how to build and run it. As a result, works on smart energy system design and operational management have garnered considerable attention. Various optimization aims, models, and methods are used in different investigations. Poul et al. analyzed the EnergyPLAN model’s applicability for smart energy systems as well as explained enhanced operational metrics. Mohammad and team developed the notion of electricity network concerning smart energy system optimization and looked at how it may be used in variation of contexts, comprising residences, business, as well as industrial (Mohammadi et al. 2018). Wazeer discussed the comprehensive strategy to mini/micro grid development, which illustrates why India’s remote regions require smart grid capacity (Wazeer and Singh 2018). Bahramara discussed how the HOMER model may be used to optimize renewable energy systems, including the optimization technique and apparatus model (Bahramara et al. 2016). Suganthi examined the use of fuzzy logic for optimization of renewable energy systems and suggested appropriate fuzzy logic in various renewable energy subsystems (Suganthi et al. 2015). Despite substantial research into the design, deployment, and management of smart energy systems, determining whether or not a system is smart remains challenging. Furthermore, from both a conceptual and practical standpoint, how to build and administer smart energy systems continues to be a significant difficulty. The idea of smart energy systems is presented initially, accompanied by major advancements in technology in smart energy systems. Also it has consideration on the application of IoT into smart energy systems, as well as a case study of China. Finally, the challenges and future prospects in the area of smart energy systems are discussed. The goal of this chapter is to extend the application of smart energy systems and the long-term sustainability of community in two distinct manner firstly smart energy system research must start, having mix of technical advancement as well as practical implementation and secondly the techniques in smart energy systems must contemplate the necessities of people’s livings to develop in a much smart and diversified track.

2

Concept of Smart Energy Systems

Since, phrase “smart energy systems” has been coined in the year 2012, there have been several energy-related technologies that are also known as “smart energy“or “smart energy systems.” Examining and evaluating the measurements, composition, aims, and administration of diverse smart energy systems yields numerous

330

A. Wazeer and A. Das

alternative conceptions and interpretations of smart energy systems. According to Lund, smart energy systems are “an approach in which smart electricity, thermal, and gas grids are combined with storage technologies and coordinated to identify synergies between them in order to achieve an optimal solution for each individual sector as well as for the overall energy system” (Lund et al. 2017). According to papers, the smart energy development area had a definite pattern of transitioning from smart grid to smart energy systems. Smart grid was an initial notion in smart energy, although opinions over its basic objective and focus remained divided. According to Orecchini and Antiangeli, the main role of smart grid is to resolve the power grid’s instabilities for combining significant percentage of renewable energy sources (Orecchini and Antiangeli 2011). According to Ahn et al., the smart grid must concentrate about how to encourage consumers to fully involve in regulated energy flows, such as through the usage of electric cars, heat pumps, and further ways. Though different perspectives could be combined at the system level, providing a single description remains a challenge (Ahn et al. 2011). The notions of smart energy as well as smart energy systems progressively emerged well after introduction of smart grid. Most researchers agreed that smart energy systems were a good idea, therefore they went on to extend and apply the notion. The majority of the research that follows, concentrates on the technological elements of smart energy systems. Mathiesen, for instance, put the notion of smart energy systems to modeling in order to demonstrate that smart energy systems are indeed the ideal method for achieving 100% renewable energy production sustainability (Mathiesen et al. 2015). Nastasi and Lo have added to the diversity of smart energy systems by proposing use of hydropower to link heat and electricity (Nastasi and Lo 2016). Simultaneously, several academics have emphasized the importance of market and regulatory concerns in the implementation of smart energy systems. Like, Shi et al. analyzed contemporary smart energy systems’ demand-side management and demand-side response tools, pointing out that customers in most countries won’t engage into power market like planned. Developing smart energy systems is far from straightforward combination for physical energy system upgrades besides information technologies, and the role of the market and policy must not be overlooked (Shi et al. 2016).

3

Key Technological Developments for Smart Energy Systems

3.1

Intelligent Novel Energy Power Generation Forecast Technology

Newer energy has evolved in a manner of crucial driver in speeding energy transition as well as reaching worldwide zero-carbon discharge because to its benefits of being clean and low carbon. Policy assistance and considerable cost reductions in solar and wind energy production have accelerated development in new power sources in latest years. Even by end of year 2019, the worldwide installed capacity of

Recent Developments in the Smart Energy Systems

331

wind turbines had surpassed 651 GW, with over 6000 GW of newly built capacity (Pryor and Barthelmie 2021). With fast growth of newer energy power production, there still are new issues for the power grid’s dependable functioning (Gao et al. 2019). As a result, original energy power generation forecasting technique has already been extensively explored and implemented in order to address this issue. Statistics for grid planning and operations may be obtained from reliable predictions of new energy production that could significantly handle this situation of largescale novel energy production instability and anti-peaking properties (Hossain et al. 2021). This technique enables advanced intelligent technique for gathering plus analyzing efficient statistics for forecasting overall energy production during next period, ensuring stable functioning of numerous system links including such energy production, transportation, distribution, and demand. Based on the models utilized, traditional forecasting approaches may be categorized into three categories: physical approach, statistical approach, and AI methodology (Wang et al. 2019). Physical technique relies upon weather forecasting that is mostly utilized for predicting atmospheric dynamics. Furthermore, because it requires a lot of computing power, it’s not suited for short-term forecast (Wu et al. 2018). A functional model of past performance and predictive items is established using the quantitative approach. Its goal (Hu et al. 2018) is to disclose the mathematical link involving energy online time-series data, however its linear approach limits its capacity to address extended forecasting challenges. Prospective AI systems have more possibility in data mining and abstraction of features (Krawczyk et al. 2017) that expands its potential uses. It includes both shallow model and deep learning predictions. Shallow model’s attribute assortment method is costly, and the generality ability is restricted, rendering complex datasets intricate. As a consequence, deep learningbased forecasting technique has already been investigated beyond to overcome these issues. Regarding photovoltaics, wind power, as well as other industries, clean energy production prediction technique is now frequently employed. The technique is capable of hourly, minute-, and second-level projection, and it plays an essential supplementary role in the secure and reliable functioning of modern energy production.

3.2

Advanced Energy Storage Technologies

Innovative ES (Energy Storage) system integrates comprehensive cross-discipline study, such as, kinetics, thermodynamics, fluid mechanics, microstructure, physical evolution, chemical processes, and others, to successfully realize the effective storage of a variety of energy sources (Liu and Du 2020). It enhances energy supply dependability and adaptability by addressing the unpredictability of dispersed generation and load in energy demand. Physical and Chemical Energy Systems are two types of ES technologies. Pumped storage, compressed air ES, flywheel ES, and superconducting magnetic ES are examples of physical ES. Liion batteries, Na-ion batteries, lead-acid batteries, flow batteries, sodium-sulfur batteries, and other innovative battery Energy Systems are examples of chemical

332

A. Wazeer and A. Das

energy storage. Pumped storage is among most significant large-scale ES techniques worldwide, with more than 127 GW of installed capacity (Joseph and Chelliah 2017). Pumped storage achieves reciprocal transformation of electricity, potential energy, and mechanical power via numerous linked reservoirs at varying heights, thus addressing the instability of renewable energy output and the power grid’s frequency control issues (Ahmadi et al. 2020). Pumped storage, on the other hand, has certain drawbacks such as the building of pumped storage power plants is constrained by geography, as the higher and lower reservoirs must be close together and have a reasonable height variation. The energy density , which a pumped storage power plant could obtain is modest when there is little elevation variation. Additionally, pumped storage power facilities have a low economic rate of return, large capital costs, and lengthy recovery times. Compressed air ES stores energy at high-pressure air and uses it to power turbines that burn higher pressure air and perhaps different gas fuels to create electricity (Tong et al. 2021). This has been investigated and deployed on a vast scale, second only to pumped storage; it includes benefit of extended charge/discharge durations. For instance, the process and safety concerns of several compressed air ES system kinds were investigated, allowing for an accurate assessment of efficiency and applicability potential (Olabi et al. 2020). In adiabatic compressed air ES systems, a novel heat storage approach was investigated, namely, the application of multiphase materials that easily managed issue of heat loss that occurs (Li et al. 2021). Flywheel ES is now one of the longest-lasting energy storage technologies, making it ideal for limited and short-term purposes (Lee et al. 2010). Flywheel ES has moved into an era of fast advancements, resulting in an increase in research and innovation of single technology and assimilated usage innovation in flywheel ES. For instance, the relevance of flywheel ES technique in functioning of larger scale electric cars was investigated, giving technological assistance for fixing the challenges of significant backup power losses (Thormann et al. 2021). Magnetic superconductivity ES stores electromagnetic energy inside a superconducting energy storage coil with quick reaction time, higher transformation efficiency, and higher dynamic power transfer efficiency (Mukherjee and Rao 2019). It has been a study focus in academia because it could effectually enhance the dependability of power supply and dynamic control of power grid. Researchers are now investigating the instant response of superconducting magnetic energy storage during short-term power system shocks (Kouache et al. 2020). Chemical ES has garnered considerable interest because of its ease of use and great efficiency, and it has evolved as the main defining path in a lot of ES uses (Miller et al. 2021). Chemical ES technology, as symbolized by batteries, offers the most diversity, the quickest advancement, and most acceptable financial viability when compared to other kinds of electric ES (Mao et al. 2015). Furthermore, its numerous properties may be tailored to satisfy a variety of power system requirements. Chemical ES, as the core and bottleneck technology of electric cars (Mao et al. 2019), is critical to the widespread adoption of modern electric vehicles, as well as energy conservation and pollution control.

Recent Developments in the Smart Energy Systems

3.3

333

Digital-Energy Integrated Market and Service Mechanism

Creating a complete digital market and service framework is crucial for furthering the construction of smart grids and smart cities as a result of digital revolution. The conventional model of distributing electricity is changing into complete service with diverse energy types and varied price modes, thanks to the reforms of the electricity system and the growing opening of the multi-energy industry (Aghaei et al. 2009) (Naughton et al. 2020). To investigate the significance of size of the market, digital readiness, and good governance in the market, Hiteva and Foxon (2021) and DuchBrown and Rossetti (2020) investigated 217 digital energy platforms in the EU market and the social economic valuation generated by energy services for users and power systems, respectively. Furthermore, the electronic energy market trading system acts as a conduit for trades and operations between market participants. In the transactions, the confidentiality, integrity, and openness of many pieces of data are critical. It also determines if the transaction processing procedure will run smoothly. As a result, a reliable power trade technique employing homomorphic encryption technology to overcome the anonymity leak in energy transactions was investigated to better understand the features of transmitting information in the digitized energy market transactions (Yi et al. 2021). Furthermore, because the marketing system incorporates a wide range of topics, its execution must take into account the functionality and desire to participate in numerous entities. As a result, running an original market mechanism while considering different stakeholders and attaining broad active involvement would be a critical obstacle to conquer in the adoption of this system.

4

Application of IoT in Smart Energy Systems

For smart energy systems, the Internet of Things represents a new framework. Newer IoT-connected equipments provide information that are utilized to invest in technology, boost productivity and effectiveness, solve crucial themes, enhance realtime decision-making, and generate creative and unique encounters. As even more devices connect, meanwhile, electricity utilities will confront more connectivity, segmentation, and security issues. The Internet of Things (IoT) supports current technologies. IoT technology connects all power production and trade elements, transformation transparency, and provides actual impact at each phase of the energy cycle, including usage to supply to end user. The economics of both hydrocarbon and renewable energy generation benefit from this partnership. By human labor and an advanced processing approach, IoT techniques for the power sector assist organizations reduce operational expenses and service. Wind farm functions can be optimized, processes can be maximized, and expenses can be greatly reduced thanks to IoT and energy technologies. Transmission lines, monitoring systems, mobility employee management, fieldwork tracking, and IoT in power management are all examples of IoT applications in the power industry. The fuel industry (oil, coal),

334

A. Wazeer and A. Das

the power sector, and IoT in gas and oil industries are all examples of IoT in the manufacturing sector. IoT system enables providers to examine relevant information on a regular basis, resulting in considerable budget savings and efficiency. Customer service and corporate operations will benefit from automation and simplicity.

4.1

Application in Policies for Energy Operations

Lawmakers must finance research and development programs as well as testing activities in order to bring IoT technology to market, and utilize building codes to promote the use of these devices. Naser and colleagues (Motlagh et al. 2020) examined the application of IoT and regulation around such devices in the energy industry. Economists, energy regulators, executives, and specialists explored the problems and prospects of Interconnection in several energy sectors as a result of this study. Encourage the convergence of community and legal data, provide data access, and improve overall system performance as part of IoT policy proposals to decrease ecological impact and regulate energy usage. Digital transformation will be propelled by IoT-enabled fast-paced technological regulations in real-time sensing and reaction, resulting in a data-driven digital government capable of providing regulations, interests of society, and benefits (Chatfield and Reddick 2019).

4.2

Application in Energy Storage

To increase the resilience of renewable energy sources, energy providers are turning to IoT-enabled energy storage technologies. Using interconnected solutions, IoT infrastructures improve the energy environment, allowing green development and sustainable energy technologies (Singh et al. 2020). Power businesses may measure consumption, detect anomalies, and give quantitative data to minimize energy storage needs and expenses using a simple, adaptable power control framework based on IoT. A great variety of research on IoT energy storage systems have been undertaken, including effective energy system design (Jayakumar et al. 2016), energy harvesting, and so on (Adila et al. 2018). The fundamental challenge with renewable energy integration is the energy gap created by renewable energy’s inconsistent existence that makes the system unsustainable. In an effort to expand grid stability, IoT energy storage techniques are being deployed to resolve this concern. Building effective energy retention with high energy density and power, completely integrated with solar, wind, and rectenna energy storage systems, continues to be a problem. In establishing effective energy storage devices, ultrathin super-capacitors and nanomaterials are required to tackle these difficulties. Furthermore, power generation is changing away from outdated fossil fuels and power plants and toward more localized, sustainable sources like wind, solar, and storage. To maintain compliance with regulations, handling the infrastructure of power plants, avoiding expensive breakdowns, and offering dependable renewable energy, enhanced insights, and improved automation are required. Modern

Recent Developments in the Smart Energy Systems

335

computers, industrial IoT, and one of the world’s most well-known IT behemoths are leading the charge. Energy storage is beginning to transform transportation, energy supply, and life’s possibilities, especially when paired with wind and solar energy. IoT energy storage devices too are assisting in the improvement of electric car battery quality.

4.3

Application in Grid Control and Power Generation

The Internet of Things (IoT) has enabled a variety of technologies in the power industry, including distribution and transmission, as well as power supply and demand. IoT devices may be utilized to boost renewable energy production, increase energy efficiency, and minimize the environmental effect of energy usage (Hui et al. 2020). Sensors for distribution, generating, and transmission equipment are part of IoT’s renewable energy generation technology. Businesses may use such solutions to track and manage equipment operation in real time from afar. This minimizes our reliance on fossil fuels while also lowering operating expenses. Renewable energy sources have several advantages over conventional energy sources. The Internet of Things (IoT) infrastructure is a significant method for triggering control grid information fast. Ongoing technological advancements in the Internet of Things era provide improved capability to manage these concerns and undertake a variety of smart grid programs. Smart metering and sophisticated metering facilities empower techniques, which can optimize conventional electricity grids while also disclosing confidential electrical power details via two-way communication schemes implemented across power exchange processes among distributors and customers (Al-Turjman and Abujubbeh 2019). The system is continually monitored and defect circumstances that might compromise the system’s efficiency are identified using an IoT-based, real-time remote operation and computerized grid management system. It conducts a comprehensive investigation and delivers automated alerts with suitable operational logs to specialists who evaluate data remotely and determine if the device is acting abnormally (Zervakis 2019). To defend vulnerable grids and interlinked distributed cyber-attack sources, distributed energy supplies, cybersecurity for renewables, and smart inverters provide a holistic, attack-resistant architecture and adaptable spectrum of cyber-physical remedies. IoT for grid infrastructure aims to create benchmark grid models for partners and the power business, as well as a common framework for planning, roadmap innovations, and market frameworks.

4.4

Application in Actuation and Sensing

Sensors are an important part of the Internet of Things database. A sensor is referred to as a transducer. Every physical mechanism that transfers one source of energy to another is referred to as a transducer. The transducer in a sensor converts any physical event into an electrical signal that may later be interpreted as readable.

336

A. Wazeer and A. Das

In current IoT networks, the sensor might gather data and share it to the command center, where a decision has to be made and a following instruction is given to the actuator, in order to adapt to perceived data. Within energy industry, many kinds of IoT-based sensors are employed, including humidity sensors, thermostats, flow sensor systems, and voltage actuators. Three key elements of the IoT network infrastructure have indeed been identified. Physical layers, network layers, and application layers are the three types of layers. Within physical layer, the sensors would seek data from the surroundings or the thing under investigation and increase safety. Water level sensors, robotic surveillance systems, ambient air detectors, home voice controls, and smart monitoring systems are all part of the IoT phase. The data from sensors and actuators is highly sloppy. Data collected from actuators or sensors must be pooled and digitalized sources for network layer processing. The application layer is in charge of delivering user-specific programs to users. Data can be transmitted to the database to examine and apply new goods and services once it has been collected, processed, and removed from the network layer. As a starting point, the data type of computing in actuating and sensing as a service was chosen (Satpathy et al. 2018).

4.5

Application in Data Visualization

IoT devices may be used to gather tolerable and verifiable data for smart energy monitoring. To administer and regulate smart meters, smart IoT metering devices are employed. The IoT approach gives pre-built network infrastructure for gathering data, connecting smart meter apps, analyzing smart data tracking, data storage, and distributing analysis findings with end users and consumers. Deep learning methods were used to control large amounts of data and guarantee the safety of IoT devices (Amanullah et al. 2020). The next era of visual analytics for subsurface modeling and energy trading is enabled by new automated data visualization apps designed with better productivity development tools. Data on pollution, power generation, and renewable energy potential are combined with national legislation using IoTbased solutions. While analytics and big data are playing a greater role in numerous industries and services than ever, space capability is beginning to be acknowledged. When it comes to developing a contemporary power grid, organizations have gone away from just providing latest software innovations and have instead revolutionized operating devices.

4.6

Application in Forecasting

Among the most important advantages of the Internet of Things is its ability to estimate energy use. Electricity companies can evaluate and link energy use with temperature, daylight, and other data to detect patterns in use throughout the area using a cloud-based analytic platform. Organizations may create a clear

Recent Developments in the Smart Energy Systems

337

vision of energy use based on predicted weather data and take necessary actions to meet demands until it becomes costly. IoT data will inspire renewable energy providers to establish goals for their long-term storage systems, resulting in more effective utilization of renewable energy sources and lower peak demand. Hong et al. (Xuesheng et al. 2020) used IoT to predict the thermal energy of smart houses. It is concluded that the building functions efficiently due to the usage of modern technologies for IoT gadgets. Because the construction sector consumes so much energy, forecasting is critical for long-term planning. Emerging innovations like cloud computing, IoT, and fog/edge computing are being used to control and anticipate the electricity system (Saputro and Akkaya 2017). Adaptive demand response, building energy management, and incorporating new equipment into the power grid are all critical topics of interest for possible IoT-enabled smart grid solutions (Ahmad et al. 2020). IoT devices are presently generating greater data than social networking websites. The system may send data several times per second, and handling billions of these events each day may need a conventional data processing infrastructure containing millions of smart devices. Though maintaining this volume of data is a significant technological problem, it is evident that data saved on the computer, even if it has been filtered, cannot be used. In order to get understanding, the data obtained should be examined. Spotting abnormalities from this data is one sort of issue that may be efficiently handled utilizing IoT.

4.7

Application in Real-Time Environment Monitoring

Changes in weather patterns are constantly monitored by climate monitoring devices. The data generated by these sensors is used to forecast the climate, follow the weather, and detect important alterations in the environment (Kim et al. 2017). While powerful automated systems are in use now, they enable for in-depth observation. Fine-grained data, adaptability, and precision are all advantages of new IoT advances. Precise forecasting necessitates a high degree of information as well as diversity in scale, shape, and deployment. This enables for early detection and intervention, reducing the risk of fatalities and injuries. Furthermore, IoT devices may be used to monitor real-time weather data in order to ensure grid reliability. It will contribute to the establishment of a robust and dependable end user electricity system.

4.8

Recent Advancement in IoT

The application of IoT in the energy domain is evolving on regular basis. It comprises smart grid as well as grid management, incorporated monitoring of electric vehicles, network administration, monitoring of micro-grids, managing district cooling and heating necessities, demand response (Huang et al. 2021), progression metering infrastructure, smart buildings, energy storage, wind farm functioning, photovoltaics, enhancing energy competence, failure repairs, digitized

338

A. Wazeer and A. Das

power production, analysis of load demand and forecasting, grid balancing, novel power solutions, zero net energy buildings, process optimization, to attain affordable energy prices, decrease cost of maintenance, power plants, and smart factories. The IoT’s distinct objective is to create assets or places to realize the quality of the data acquired from the cloud by using analysis of the data to acquire resources and insight from connected devices, based on current interfaced services and infrastructure. Device engineers will employ simulation designs, efficient analytics, and what-ifany-hypotheses to deliver accurate grid status projections and associated conditions after data has been received from every section of the distribution network. The prospects connected with data analytics and its transition from reactive to productive are the major elements of the IoT power grid. Organizations may optimize their processes by distributing changes to the full city network/database without harming performance of the system using a variety of IoT network settings. Maybe there are automobiles, enterprises, cities, residences, industries, or any other collection of industry-specific procedures. IoT allows wearable gadgets, smart power panels, computers, and tablets to operate such devices remotely. Improved technology nodes, optimized flash access, energy generation, flexible, low-power, and analogue front ends, incorporating major characteristics into different digital chips, supplying extremely versatile energy mechanisms, power solutions, as well as using power-optimized devices are just a few of the steps that energy providers must take to enhance the network reliability of electrical systems. Till date, just a few electricity firms have pioneered these new IoT advancements. The majority of companies, on average, follow classic industrial trends. Nevertheless, given the industry’s growth and the interconnected nature of local power grids, power firms will have to expand their IoT activities in the future. IoT sensors provide a real-time picture of power generation and can aid in the transition to centralized production. It will allow businesses to track the health of power system components, power grids, grid infrastructure, solar panels, and wind turbines from afar.

5

Case Study: Smart Energy Towns in China

Due to the enormous issues of urbanization and energy amid fast economic and social growth, China was a major instance for regional research on this topic. Since 1979, China has seen significant urbanization, with the rate of urbanization growing from 18.96% in 1979 to 58.52% in 2017 (National Bureau 2018). China’s towns and cities were suffering major issues in many areas as a result of rapid urbanization, including increased pollutants and CO2 mostly due to energy consumption (Ma et al. 2011). The idea of smart energy city will be discussed in light of China’s reality, providing a beneficial model for spatial planning and smart city and smart energy practice in emerging nations. Because China’s SET involves the cross-disciplinary interaction of power systems, urban planning, ICT, as well as other sectors, adopting a single disciplinary method to research is problematic. As a cross-sectional field specialized in multidisciplinary challenges, systematics is supposed to give a better

Recent Developments in the Smart Energy Systems

339

methodological foundation. Because the idea and practice of China’s SET entail a wide variety of linked issues, which fall under the category of a conventional soft challenge of weak structure or referred to as a contentious issue, the soft method of systematics is more appropriate. Checkland’s technique is one of the most wellknown soft methodologies in systems research (Checkland 1990).

5.1

Constraints to Build China’s SET

In certain locations, the discrepancy in economic growth between provinces causes an underinvestment and makes it difficult to launch SET initiatives. The infrastructure building for SET projects, which includes energy systems and ICT, necessitates a significant initial investment. The original funding sources for the successful projects revealed comprise governmental economic help, state-owned firm funding, and some local private company investment. In certain impoverished places, depending entirely on government funding was not enough to ensure the project’s smooth growth, especially given the impoverished local market. Facilities also failed to foster an appropriate business model, resulting in participation from state-owned companies, private firms, and other players. Significant disparities in energy resource allocation, energy requirements, and industrial change have resulted in unequal SET project success. Rich energy assets and a little outdated industrial structure in some locations in desperate need of energy and manufacturing transition (like as Inner Mongolia) encourage local governments to aggressively deploy SET initiatives to continue developing local high-quality sources and pursue industrial upgrading. Because economic building is still the major need in some locations where energy supplies are few and economic growth is poor (e.g., the Northeast), the advancement of SET initiatives is more slow and encounters significant opposition. Simultaneously, there are a few glimpses of optimism. The extra focus of coastal inhabitants and city officials to urban living standards, for instance, has aided the development of SET projects.

5.2

Driving Force for China’s SET

• The fundamental driving factor behind promoting the implementation of SET is the power saving and reduced emissions strategy, which has been updated as the idea of sustainable civilization development and greener industrialization, as recommended by President Xi Jinping (Report 2017). • China has become the first developing nation to declare a national obligation to address global climate change as a result of international pressure and collaboration. • Aside from national policy and international collaboration as driving forces, several communities and towns spontaneously and proactively encourage the practice of SET, taking into consideration their unique development needs. For example, massive waste of energy output, including wind and hydro curtailment,

340

A. Wazeer and A. Das

is nearly ubiquitous in various regions of Northwest and Southwest China. Local governments are highly motivated to promote SET by implementing different energy storage equipment and systems, such as the building of an electric car public transportation network, the retirement of a battery management system, and so on. Simultaneously, minimizing energy waste has resulted in significant economic advantages, inviting a diverse range of stakeholders to join in project development. • Private corporations investment makes for a significant share of overall SET project financing.

5.3

Recommendations to Build China’s SET

• To enhance the planning direction of SET, the central government should analyze and issue the smart city program, smart energy systems strategy, and SET plan as well as related supporting policies, as early as feasible. Simultaneously, it will improve cooperation and interaction with regional governments, as well as provide regional governments more project management power and monetary incentives to encourage local innovation. • State-owned businesses, private firms, and foreign enterprises must work together to improve collaboration models and support the creation of SET in China through innovative business models and cooperation mechanisms. State-owned firms, in particular, might be more engaged in the implementation of China’s SET initiatives. • International bodies should pay greater attention to China’s SET programs and help to direct more global recognition to the country. Simultaneously, they will continue to convey innovative designs and real-world experience from industrialized countries to China, as well as work to give meaningful support for China’s SET strategy, investigation, and project financing.

6

Challenges in Smart Energy Systems

Owing to the rapid expansion of the national economy and continuous advancements in science and technology, power generation, transmission, and demand have progressively acquired liberalization, multiplicity, and intellect. The ever-increasing demands that society places on the energy service industry necessitate continuous innovation and advancements in a range of SES (Smart Energy Systems) techniques. As a result, the following are the key issues that important SES technologies face: • SESs are primarily formed in locations with more advanced information technology since they rely heavily on effective intelligent technology. It would be difficult to construct SES in locations where situations are somewhat backward in all respects, particularly in isolated mountainous where information technology has still not become widely used. As a result, meeting the energy demands

Recent Developments in the Smart Energy Systems

•

•

•

•

7

341

of people in remote locations, constructing a smart energy supply system, and achieving intelligent energy usage technology that is not limited by geography will be a big issue for future SES construction. Prediction technology nowadays is mostly focused on anticipating a single entity. In such a system with many dispersed energy sources, for example, photovoltaic power generation prediction technology is used. This type of prediction technology, on the other hand, just looks at the law of solar power generation. Other unpredictable components in the system, including wind energy volatility and user energy randomization, are difficult to anticipate at the same time. Furthermore, present prediction technology is in its infancy. For instance, it is difficult to forecast certain extreme accidents, and it is also difficult to estimate and analyze the impact that an event does to the system. As a result, more research into an effective prediction tool capable of performing comprehensive real-time online evaluation of numerous uncertainties of different entities in the SES is required. There are many different types of energy, and the new energy that is presently being generated and put to use is simply the beginning; the production and application of new energy is far from sufficient. As a result, how to collect society’s distributed and diversified energy while also protecting the ecological landscape while mining and using RE on the planet is critical to the expansion of SESs. The amount of societal energy waste that currently exists is enormous, with the majority of it centered on commercial building power use, particularly the central air-conditioning unit. As a result, a fundamental difficulty in the construction of SESs is how to improve citizens’ energy-saving ideas, establish an effective punishment mechanism, involve residents in energy-saving actions, raise energy use efficiency, and decrease energy waste. Various energy markets chose various clearance standards and function on different time frames. To some degree, this stifled the eagerness of many SES companies to participate in the market. The deployment of DR was further hampered by the complicated market operating system. As a result, how to structure multiple energy markets jointly to enable concurrent functioning, rationalize strong competiveness, and clarify the sophisticated mechanics of multi-energy markets will be essential concerns to address as SESs grow.

Future Prospects in Smart Energy Systems

The design and construction of SESs face new opportunities with the active progress of global carbon neutrality attempts and the development of smart cities, or the ongoing growth of smart data technologies like AI, cloud computing, big data, and IoT: • Future SES work must start with a mix of technical innovation and application in practice, maximizing the use of digitized and smart technology while permitting

342

•

•

•

•

•

8

A. Wazeer and A. Das

the extensive adoption of advanced technology in real-world settings. Another of the main hurdles to SES advancement right now is the divergence from breakthrough studies on many important technologies and their actual imbalanced production. As a consequence, establishing a solid infrastructure is critical for the advancement of different technologies within SES. Encourage the growth of ES device materials that are high-density, low-cost, and long-lasting. The continued development of high-efficiency and sophisticated ES techniques has emerged from the deployment of sizable and high-proportion RE. Encourage the creation of individualized and creative SES services, as well as the active development of innovative solutions that address societal demands and people’s lives. In the framework of energy saving and emission reduction, government bodies heavily encourage the development of SESs. The SESs’ goal is to attain energy transformation, improve energy efficiency, and make it easier to reach carbon peak and carbon neutral objectives. Researchers’ in-depth investigation of diverse studies results in the discovery of new topic areas on a regular basis. This breathes fresh life into the effective ES industry and opens up new opportunities for the future implementation of more large-scale and consistent ES forms. Furthermore, advances in smart information technology would help to expand the future energy integrated service business model and system, as well as the digitized energy integrative market and service operations mechanism.

Conclusions

In latest years, SESs have indeed been investigated and improved on a regular basis to support sustainable growth. The SES provides adaptable multi-energy control and management in a variety of uncertain or hostile environments, as well as more diversified energy sources and applications. Smart energy systems have gotten a lot of attention from experts in recent years as a development trend for future energy systems. This study provided a thorough review of the most recent studies on smart energy systems. The notion of smart energy systems was discussed, backed by the major technological advancements in smart energy systems. Also there is a review on the use of IoT in smart energy systems, as well as a case study of China. Ultimately, the obstacles and future opportunities in the field of smart energy systems are discussed.

References A.S. Adila, A. Husam, G. Husi, Towards the self-powered internet of things (IoT) by energy harvesting: Trends and technologies for green IoT, in 2018 2nd International Symposium on Small-Scale Intelligent Manufacturing Systems, (SIMS, 2018), pp. 1–5 J. Aghaei, H. Shayanfar, N.M. Amjady, Multi-objective market clearing of joint energy and reserves auctions ensuring power system security. Energy Convers. Manag. 50(4), 899–906 (2009)

Recent Developments in the Smart Energy Systems

343

R. Aghamolaei, M.H. Shamsi, M. Tahsildoost, J. O’Donnell, Review of district- scale energy performance analysis: Outlooks towards holistic urban frameworks. Sustain. Cities Soc. 41, 252–264 (2018) T. Ahmad, H. Zhang, B. Yan, A review on renewable energy and electricity requirement forecasting models for smart grid and buildings. Sustain. Cities Soc. 55, Article 102052 (2020) S.H.R. Ahmadi, Y. Noorollahi, S. Ghanbari, M. Ebrahimi, H. Hosseini, A. Foroozani, A. Hajinezhad, Hybrid fuzzy decision making approach for wind-powered pumped storage power plant site selection: A case study. Sustain. Energy Tech. Assess. 42, 100838 (2020) C. Ahn, C.T. Li, H. Peng, Optimal decentralized charging control algorithm for electrified vehicles connected to smart grid. J. Power Sources 196(23), 10369–10379 (2011) F. Al-Turjman, M. Abujubbeh, IoT-enabled smart grid via SM: An overview. Futur. Gener. Comput. Syst. 96, 579–590 (2019) M.A. Amanullah, R.A.A. Habeeb, F.H. Nasaruddin, A. Gani, E. Ahmed, A.S.M. Nainar, et al., Deep learning and big data technologies for IoT security. Comput. Commun. 151, 495–517 (2020) S. Bahramara, M.P. Moghaddam, M.R. Haghifam, Optimal planning of hybrid renewable energy systems using HOMER: A review. Renew. Sust. Energ. Rev. 62, 609–620 (2016) A.T. Chatfield, C.G. Reddick, A framework for internet of things-enabled smart government: A case of IoT cybersecurity policies and use cases in U.S. federal government. Gov. Inf. Q. 36(2), 346–357 (2019) P. Checkland, Systems Thinking, Systems Practice (Huaxia Press, Beijing, 1990) N. Duch-Brown, F. Rossetti, Digital platforms across the European regional energy markets. Energy Policy 144, 111612 (2020) Z. Gao, D. Mao, J. Wang, Distribution grid response monitor. IET Generation. Transm. Distrib. 13(19), 4374–4381 (2019) R. Hiteva, T. Foxon, Beware the value gap: Creating value for users and for the system through innovation in digital energy services business models. Technol. Forecast. Soc. Change 166, 120525 (2021) M.A. Hossain, R.K. Chakrabortty, S. Elsawah, M.J. Ryan, Very short-term forecasting of wind power generation using hybrid deep learning model. J. Clean. Prod. 296, 126564 (2021) J. Hu, J. Heng, J. Tang, M.M. Guo, Research and application of a hybrid model based on meta learning strategy for wind power deterministic and probabilistic forecasting. Energy Convers. Manag. 173, 197–209 (2018) C. Huang, H. Zhang, Y. Song, L. Wang, T. Ahmad, X. Luo, Demand response for industrial micro-grid considering photovoltaic power uncertainty and battery operational cost (IEEE Transactions on Smart Grid, 2021) H. Hui, Y. Ding, Q. Shi, F. Li, Y. Song, J. Yan, 5G network-based internet of things for demand response in smart grid: A survey on application potential. Appl. Energy 257, Article 113972 (2020) H. Jayakumar, A. Raha, Y. Kim, S. Sutar, W.S. Lee, V. Raghunathan, Energy-efficient system design for IoT devices, in Proceedings of the Asia and South Pacific Design Automation Conference, (ASP-DAC, 2016), pp. 298–301 A. Joseph, T.R. Chelliah, A review of power electronic converters for variable speed pumped storage plants: Configurations, operational challenges, and future scopes. IEEE J. Emerg. Selected Topics Power Electron. 6(1), 103–119 (2017) S.H. Kim, J.M. Jeong, M.T. Hwang, C.S. Kang, Development of an IoT-based atmospheric environment monitoring system, in International Conference on Information and Communication Technology Convergence: ICT Convergence Technologies Leading the Fourth Industrial Revolution, (ICTC 2017, 2017), pp. 861–863 I. Kouache, M. Sebaa, M. Bey, T. Allaoui, M. Denai, A new approach to demand response in a microgrid based on coordination control between smart meter and distributed superconducting magnetic energy storage unit. J. Energy Stor. 32, 101748 (2020) S. Koutra, V. Becue, M.-A. Gallas, C.S. Ioakimidis, Towards the development of a net-zero energy district evaluation approach: A review of sustainable approaches and assessment tools. Sustain. Cities Soc. 39, 784–800 (2018)

344

A. Wazeer and A. Das

B. Krawczyk, L.L. Minku, J. Gama, J. Stefanowski, M. Wo’zniak, Ensemble learning for data stream analysis: A survey. Inf. Fusion 37, 132–156 (2017) J. Lee, S. Jeong, Y.H. Han, B. Park, Concept of cold energy storage for superconducting flywheel energy storage system. IEEE Trans. Appli. Supercond. 21(3), 2221–2224 (2010) R. Li, Y. Zhang, H. Chen, H. Zhang, Z. Yang, E. Yao, H. Wang, Exploring thermodynamic potential of multiple phase change thermal energy storage for adiabatic compressed air energy storage system. J. Energy Stor. 33, 102054 (2021) Y. Liu, J.-L. Du, A multi criteria decision support framework for renewable energy storage technology selection. J. Clean. Prod. 277, 122183 (2020) H. Lund, A.N. Andersen, P.A. Østergaard, B.V. Mathiesen, D. Connolly, From electricity smart grids to smart energy systems – A market operation based approach and understanding. Energy 42(1), 96–102 (2012) H. Lund, P.A. Østergaard, D. Connolly, B.V. Mathiesen, Smart energy and smart energy systems. Energy 137, 556–565 (2017) L.W. Ma, P. Liu, F. Fu, et al., Integrated energy strategy for the sustainable development of China. Energy 36, 1143–1154 (2011) D. Mao, H.J. Khasawneh, M.S. Illindala, B.L. Schenkman, D.R. Borneo, Economic evaluation of energy storage options in a microgrid with flexible distribution of energy and storage resources, in 2015 IEEE/IAS 51st Industrial & Commercial Power Systems Technical Conference (I&CPS), (IEEE, 2015), pp. 1–7 D. Mao, J. Wang, J. Tan, G. Liu, Y. Xu, J. Li, Location planning of fast charging station considering its impact on the power grid assets, in 2019 IEEE Transportation Electrification Conference and Expo (ITEC), (ITEC, 2019), pp. 1–5 B.V. Mathiesen, H. Lund, D. Connolly, et al., Smart energy systems for coherent 100% renewable energy and transport solutions. Appl. Energy 145, 139–154 (2015) M.A. Miller, J. Petrasch, K. Randhir, N. Rahmatian, J. Klausner, Chemical energy storage, thermal, mechanical, and hybrid chemical energy storage systems (Academic Press, 2021), pp. 249–292 M. Mohammadi, Y. Noorollahi, B. Mohammadi-ivatloo, M. Hosseinzadeh, H. Yousefi, S.T. Khorasani, Optimal management of energy hubs and smart energy hubs – A review. Renew. Sust. Energ. Rev. 89, 33–50 (2018) N.H. Motlagh, M. Mohammadrezaei, J. Hunt, B. Zakeri, Internet of things (IoT) and the energy sector. Energies 13(2) (2020) P. Mukherjee, V. Rao, Design and development of high temperature superconducting magnetic energy storage for power applications-a review. Phys. C. Superconductivity Appl. 563, 67–73 (2019) B. Nastasi, B.G. Lo, Hydrogen to link heat and electricity in the transition towards future smart energy systems. Energy 110, 5–22 (2016) National Bureau of Statistics of China, China Statistical Yearbook (China Statistic Press, Beijing, 2018) J. Naughton, H. Wang, S. Riaz, M. Cantoni, P. Mancarella, Optimization of multi-energy virtual power plants for providing multiple market and local network services. Elec. Power Syst. Res. 189, 106775 (2020) A. Olabi, T. Wilberforce, M. Ramadan, M.A. Abdelkareem, A.H. Alami, Compressed air energy storage systems: Components and operating parameters–a review. J. Energy Stor. 102000 (2020) F. Orecchini, A. Santiangeli, Beyond smart grids - the need of intelligent energy networks for a higher global efficiency through energy vectors integration. Int. J. Hydrog. Energy 36(13), 8126–8133 (2011) S.C. Pryor, R. Barthelmie, A global assessment of extreme wind speeds for wind energy applications. Nat. Energy 6(3), 268–276 (2021) Report (2017) at the 19th national congress of Chinese Communist Party J. Reynolds, Y. Rezgui, J.-L. Hippolyte, Upscaling energy control from building to districts: Current limitations and future perspectives. Sustain. Cities Soc. 35, 816–829 (2017) N. Saputro, K. Akkaya, Investigation of smart meter data reporting strategies for optimized performance in smart grid AMI networks. IEEE Internet Things J. 4(4), 894–904 (2017)

Recent Developments in the Smart Energy Systems

345

S. Satpathy, B. Sahoo, A.K. Turuk, Sensing and actuation as a service delivery model in cloud edge centric internet of things. Futur. Gener. Comput. Syst. 86, 281–296 (2018) C.C. Shao, X.L. Wang, W. By, Probe into analysis and planning of multienergy systems. Proc. CSEE 36(14), 3817–3828 (2016) H. Shi, N. Blaauwbroek, P.H. Nguyen, et al., Energy management in multi-commodity smart energy systems with a greedy approach. Appl. Energy 167, 385–396 (2016) S. Singh, P.K. Sharma, B. Yoon, M. Shojafar, G.H. Cho, I.H. Ra, Convergence of blockchain and artificial intelligence in IoT network for the sustainable smart city. Sustain. Cities Soc. 63, Article 102364 (2020) Y. Su, Smart energy for smart built environment: A review for combined objectives of affordable sustainable green. Sustain. Cities Soc. 53 (2020) L. Suganthi, S. Iniyan, A.A. Samuel, Applications of fuzzy logic in renewable energy systems – A review. Renew. Sust. Energ. Rev. 48, 585–607 (2015) B. Thormann, P. Puchbauer, T. Kienberger, Analyzing the suitability of flywheel energy storage systems for supplying high-power charging e-mobility use cases. J. Energy Stor. 39, 102615 (2021) Z. Tong, Z. Cheng, S. Tong, A review on the development of compressed air energy storage in China: Technical and economic challenges to commercialization. Renew. Sust. Energ. Rev. 135, 110178 (2021) G. Venkataramani, P. Parankusam, V. Ramalingam, J. Wang, A review on compressed air energy storage – A pathway for smart grid and polygeneration. Renew. Sust. Energ. Rev. 62, 895–907 (2016) H. Wang, Z. Lei, X. Zhang, B. Zhou, J.M. Peng, A review of deep learning for renewable energy forecasting. Energy Convers. Manag. 198, 111799 (2019) A. Wazeer, A. Singh, Smart grid. Int. J. Adv. Res. Sci. Eng. 7(5), 201–205 (2018) L. Wu, X. Gao, Y. Xiao, Y. Yang, X. Chen, Using a novel multi-variable grey model to forecast the electricity consumption of Shandong Province in China. Energy 157, 327–335 (2018) L. Xuesheng, L. Ma, C. Chong, Z. Li, W. Ni, Development of smart energy towns in China: Concept and practices. Renew. Sust. Energ. Rev. 119, 109507 (2020) H. Yi, W. Lin, X. Huang, X. Cai, R. Chi, Z. Nie, Energy trading IoT system based on blockchain. Swarm Evol. Comp. 64, 100891 (2021) G. Zervakis, IoT for smart grids (2019, pp. 163–180)

Constructing Renewable Energy Systems Using Big Data Applications Nassim Sohaee

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Renewable Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Big Data Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 An Integrated Big Data Architecture in Renewable Energy . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Blockchain for Smart Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Forecasting models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

348 348 350 350 352 353 357 357

Abstract

In this chapter we will discuss how big data helps in advancing the field of smart energy systems. Massive data is collected over time in the energy sector. Data are gathered from sources like wireless transmission, network communication, and cloud computing technologies. We will discuss analytical techniques to effectively and efficiently integrate renewable energy sources into the system. Constructing a data-driven smart energy system can provide essential support for the efficient expansion of the renewable energy industry. We will discuss some applications that can be developed or implemented only in the presence of big unstructured data. Keywords

Big data · Forecast · Neural network · Blockchain

N. Sohaee () Information Technology and Decision Science, University of North Texas, Denton, TX, USA e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_176

347

348

1

N. Sohaee

Introduction

Energy is essential to modern human society’s quality of life. The source of energy indicates whether it is self-renewing or not. Energy resources that can naturally replenish are called renewable energy, implying that nonrenewable resources are limited in supply and cannot be used sustainably. The primary sources of renewable energy can be listed as: • • • •

renewable energy sources electricity and smart grid renewable energy sources fossil fuels renewable energy sources wind power and solar energy sources geothermal

Renewable sources of clean energy offer many advantages, including lower costs and almost zero emissions from generation. The major disadvantage of renewable energy is variability and unpredictability. The energy production by sources like solar energy and wind power are primarily affected by weather conditions. Renewable energy distribution and management need efficient, effective, and innovative system, known as smart grid. A conventional power grid is a network of transmission lines, substations, and transformers that delivers electricity from the power plant to consumption destinations. Smart grids are adding digital technology to the grid system to make the system reliable, available, and efficient. The smart grid system has many known benefits, such as the efficient transmission of electricity, reducing operation and management costs, reducing peak demand, increasing integration of large-scale renewable energy systems, and improving security. The focus of the smart grid is to calculate the optimum generation, transmission, and distribution pattern (Abdollahy et al. 2013).

1.1

Renewable Energy

Based on the information released by the US Department of Energy Information Administration (EIA), more than 90 percent of energy consumption in the United States will be nonrenewable by the second decade of the twenty-first century. Nonrenewable energy sources are petroleum, hydrocarbon gas liquids, natural gas, coal, and nuclear energy. There are five primary renewable energy sources, solar, geothermal, wind, biomass, and hydropower. Figure 1 illustrates the current position of the United States in energy consumption. Primary energy sources have changed throughout history. Figure 2 shows these changes since the independence of the United States . Today, renewable energy provides nearly 12 percent of total US energy consumption. The significant benefit of increasing renewable energy consumption is the reduction of greenhouse gas emissions. Many countries – including the United States – have had policies to mitigate greenhouse gas production in the past few decades. These policies are

Constructing Renewable Energy Systems Using Big Data Applications

349

Fig. 1 United States energy consumption by source, 2020 (What is energy 2021)

Fig. 2 Shares of total US energy consumption by major sources in selected years (1776–2020) (Share of total us energy consupmtion by major sources in selected years 2021)

enforced in many forms, from gasoline taxes to requiring that a certain amount of electricity in a state come from renewable sources. Many factors such as cost, change in state policies, and, finally, technological advancements contribute to moving renewable energy forward. According to Forbes, renewable energy production is becoming more affordable and can become cheaper than traditional fossil fuels within a few years (Dudley 2018). Meanwhile,

350

N. Sohaee

many countries are creating energy policies to diversify energy generation resources. Within the United States, almost two-thirds of states set a goal to replace fossil fuels with forms of renewable energy. Above all, renewable energy technology has improved significantly in past decades. We have a massive amount of data collected that can be used to optimize the energy generation and distribution process.

1.2

Big Data Technology

In the past 50 years, with the help of computer devices, we have been experiencing an information explosion. Nowadays, almost everything can be formatted to be a computer-friendly object. Also, computerized tools are vastly available to collect information and make them more accessible and extend their scope. As a result, we deal with massive datasets in size and complexity collected from different sources. This new concept is referred to as big data. Kim et al. (2014) summarized the idea of big data as 3 Vs, volume, variety, and veracity. In big data, massive volumes of data can be used to address business problems that previously were not possible to tackle. Velocity is addressing the rate of data collection and processing. Many big data applications need real-time or near realtime evaluation and actions. And finally, variety refers to diverse types of available data. Contrary to the traditional dataset, big data may come in new unstructured data types. The unstructured feature in big data requires additional preprocessing steps to prepare data. Applications of big data and artificial intelligence are emerging technologies in renewable energy. Over time the architecture of the energy distribution system has changed from unidirectional energy flow to a smart grid that incorporates multiple energy sources into the infrastructure. Smart grids integrate information and technology into every aspect of energy generation, delivery, and consumption. The ultimate goal is to minimize the environmental impact, enhance the market, improve reliability and service, and reduce costs while improving efficiency (The role of smart grid 2022).

2

An Integrated Big Data Architecture in Renewable Energy

The world is moving fast toward replacing traditional power grids with more efficient systems. A power grid is a network of transmissions, substations, and transformers to deliver electricity from power plants to consumers. Power stations are often located near energy sources and away from the consumers in this setting. The substations are there to adjust the electricity voltage for different usage. Then a massive load of transmission lines carries the power long distances to the power distribution centers, and, finally, these centers will distribute power to individual consumers. As early as 1882, when Thomas Edison invented electricity as a source of energy, there was an interest in measuring the amount of energy produced and consumed.

Constructing Renewable Energy Systems Using Big Data Applications

351

Samuel Gardiner created the first electric meter in 1872, and in 1883, Hermann Aron invented a recording meter to show the energy consumed on clock dials (The history of making the grid smart 2022). Collecting data on energy consumption over time helped power grid companies to learn about electricity demands and manage peak loads. One of the methods of handling peak loads was to build generating capacity to meet the markets’ demand during the peak time and stay idle during nonpeak loads. Since the 1970s, when automatic meter devices were introduced, power grids recorded data on many dimensions. Many of these collected data eventually was a trigger to move toward a safer, more efficient, and more reliable energy distribution which is a building block of a smart grid. The national coordinator’s office for smart grid interoperability at the US department of commerce released the first framework and roadmap for smart grid standards in January of 2010. The latest update of this framework is dated October 2014. The conceptual structure and overall organization presented in this report are displayed in Fig. 3. In this conceptual design, a smart grid is a composition of many independent systems working in harmony with one another. The goal of this approach was to maximize flexibility in design, and implementation (Greer et al. 2014). The first step toward a smart grid system is to adopt new energy sources. The main disadvantage of fossil fuels as the primary source of power energy is the everdecreasing supplies and emission of greenhouse gases. As of now, most countries are moving toward clean energy sources like solar, wind power, and geothermal. Therefore, a smart grid system should support a multitude of clean and renewable energy sources. Smart grids are designed to support both central and decentralized sources of energy. Traditional grids are designed to rely on only one energy source and cannot overload with different sources of energy production. Also, the conventional grids cannot accommodate and control systems in uncertain situations. Therefore, overloading renewable energies on the traditional grids results in an unreliable system. On the other hand, there are many methods to monitor and control the transmission systems in real time in the smart grids.

Fig. 3 Conceptual diagram of smart grids (Ahmad et al. 2019)

352

2.1

N. Sohaee

Blockchain for Smart Grid

Many national and international organizations contributed to developing roadmaps to define and understand technical, financial, and legal obstacles around the concept of smart grids. Considering the advancement in the Internet of Things (IoT), a broad range of data formats, and the impact of security as the main challenges in a smart grid, IEEE published IEEE 2030.5 as a standard for smart grids. The latest update is dated back to December 21, 2018. The standard complies with the IoT concept, which gives users the ability to manage energy consumption. To date, there is not any single comprehensive definition of smart grids. Still, the core of most definitions is around the concept of a secure, reliable heterogeneous system that emphasizes communications for measurement, monitoring, management, and control (Abrahamsen et al. 2021; Ghalib et al. 2030). Figure 4 shows that smart grids have the potential to integrate electrical energy from a wide range of sources. Moreover, advancements in technologies like automated systems, smart consumer devices, and electrical energy storage generate a massive amount of data. The data collected in smart grids should serve many purposes as reducing the energy dependency on nonrenewable sources, reducing the impacts of environmental pollution, and, at the same time, increasing the energy security of the operation and management efficiency (Colmenares-Quintero et al. 2021). In the remaining of this chapter, we will discuss how data analytics and intelligence have the potential to support the technical and managerial decisionmaking process. The smart grid revolution made it possible to expand the sources of electrical energy. Even though, to this date, fossil fuel is the primary source of electrical energy. The power grid systems integrated many other clean energy sources like solar, wind power, etc. The automated and smart power grids made it possible to change how energy is generated and distributed. Also, technologies like solar panels

Fig. 4 Schematic view of smart grids (Smart grid evolution 2018)

Constructing Renewable Energy Systems Using Big Data Applications

353

are now widely available to let the consumers generate electricity for both individual and commercial usage. In the past decade, the concept of blockchain received significant attention in security. Initially, Satoshi Nakamoto proposed the concept of blockchain for Bitcoin cryptocurrency as a solution for the double-spending problem in cryptocurrency without the power of a third party (Nakamoto 2008). At the same time, the demand for having a secure electricity supply-demand is on the rise, which will add complexity to design and implementation. A blockchain is a form of a distributed database system that is shared among connected computers to maintain a secure and decentralized record of transactions. This new technology can guarantee the fidelity and security of transactions without the need for a trusted third-party intermediator. One of the challenges in smart grids is determining the price of electricity in a distributed and decentralized smart grid system. Blockchain, as a peer-to-peer distributed ledger of transactions, can safely store the information of all parties involved. Some initial and preliminary research in this field has already proposed a blockchain structure for smart grids transactions (Hahn et al. 2017; Münsing et al. 2017). Inspired by Bitcoin, Aitzhan and Svetinovic (2016) proposed a design for smart grids with a bidirectional communication flow that solves the transaction security in a decentralized smart grid. Zheng et al. (2018) combined blockchain technology, proof of stake consensus mechanisms, and cryptography tools to build a new smart grid power trading system. However, this proposed method lacks a practical implementation and evaluation (Mollah et al. 2020). Agung and Handayani proposed a novel blockchain architecture for smart grids. In their architecture, a smart electric grid consists of a few clusters, and each cluster represents one of the contributing factors in a transaction. Figure 5 shows the schematic view of this blockchain architecture. Cluster A represents traditional government-owned power plants, cluster D consists of privately owned power plants, and clusters B, C, and E represent buildings with and without of capability of generating electricity. Finally, cluster E represents public outlets. Grid storage is supposed to store excess electricity generated in the system. We can also assume that there is a computerized system like an electricity meter that keeps track of electricity generation, usage, and distribution. In this design, electricity is a token in the system, and consumers pay for the electricity they need in advance. Power plants create tokens for the electricity generated and make the tokens available on the system. Consumers exchange tokens with energy. This smart design will give the consumers options to buy energy from multiple producers (Agung and Handayani 2020).

2.2

Forecasting models

As we discussed before, one of the goals in the energy sector is to replace fossil fuels with clean and renewable energy. Also, the ever-growing energy demand requires energy-intensive management to enhance energy system efficiency and performance. One of the crucial aspects of energy management in smart grids is

354

N. Sohaee

Fig. 5 Blockchain architecture for smart grids (Agung and Handayani 2020)

to guarantee a reliable balance between energy production and consumption. This requirement adds different layers of complexity to the design of smart grids. For example, the energy system reliability and security should depend on the operation’s capability to support planned or unexpected disruption on both ends of production and consumption. Therefore, in a smart grid, we should have a reliable expectation of the network load and capacity for short- and long-term planning (Ahmad et al. 2020). To adequately address this vital challenge, many forecasting techniques can be applied to big data collected in a smart grid and other contributing organizations (Ahmad et al. 2020). However, the major obstacle in load and capacity forecasting remains a large number of contributing factors. Sometimes, some of these factors may not trace the electricity load and consumption. Also, there was no meaningful medium to collect and analyze data with the traditional power grid systems. With the emerging smart grid power systems and computerized meters, we can collect data at each step from the power plan to the consumers. Meanwhile, many other practical and affecting information like weather information can be integrated into the system (Li et al. 2017). Big data and sophisticated machine learning forecasting algorithms can be used to address the critical problem in smart grids. Electricity consumption forecasting is essential in detecting unexpected power usage patterns. However, this task depends on many complex factors like historical consumption, consumption habit, environmental, and social factors (Lei et al. 2019). Electricity demand prediction is not a new field of research, and it has been well studied for over a century. However, these model predictions are heavily used only

Constructing Renewable Energy Systems Using Big Data Applications

355

on the traditional power plants without much flexibility to merge into smart grids (Singh et al. 2011). In renewable energy, forecasting can be an essential tool in many areas such as electricity production and consumption, contract evaluation, load switching, power extension, and many more. Also, the objective can change from short-term to long-term forecasts for all these models. Generally, a short-term forecast is ideal for production control and management prediction, and long-term projections are mainly needed for production planning. Most research papers published in this regard focus on short-term energy prediction for various load situations (Ahmad et al. 2020).

2.2.1 Short-Term Forecast Daily operations of smart grids require constant and immediate load forecasting for control, management, and daily planning of energy generation, load switching, and infrastructure maintenance. To accurately and efficiently forecast any of the objectives listed above, it is necessary to consider that the massive collective data on the field of interest is fundamental. Various forecasting models have been proposed and tested in different situations in the past two decades. Many learning models have the flexibility to learn from a small load of data. However, most of them require big data for complex datasets. Considering the magnitude of complexity in smart grids, it is necessary to have massive data to train and evaluate learning models before deploying them for actual and field prediction. For this part, we will have a quick review of models proposed to perform short-term forecasts. In smart grids, the term short-term refers to a few hours to a day ahead of the prediction horizon (Ahmad et al. 2019). The simplest predictive models are linear models, predicting the load based on a linear combination of underlying features considered in the model. Linear models are generally applauded for their simplicity, but as we discussed before, they may not be a good fit for smart grid load prediction. The limitations of linear models are discussed in Hagan and Behr (1987). On the other hand, many nonlinear learners can adapt to the complexity of the smart grids. One of the well-regarded models that have the flexibility of adapting to complex datasets is the support vector machine (SVM) model. Ayub et al. (2019) applied SVM along with the XGBoost model for feature selection. They claimed that their model performance is acceptable for short-term and midterm forecasts. Zang et al. (2015) used an ensemble of decision tree models and SVM with different parameters and kernels to maintain nonlinear learning. In their proposed model, first, they predicted the load pattern using a decision tree model, and then the corresponding SVM models were used to forecast the system load. Even though traditional and nonlinear models like SVM had some success in load forecasting, many published researchers suggest that the major problem with SVM is its scalability to large and complex datasets. To solve this problem, alternative artificial intelligence learning models are proposed. The adaptive learning flexibility of neural networks and deep learning models makes them excellent for large and complex datasets.

356

N. Sohaee

The amount of research conducted in energy load forecasting with the help of neural networks and deep learning models has increased significantly in recent years (Vanting et al. 2021). Considering the time-series nature of load forecasting, a recurrent neural network (RNN) can take the time dependency between the instances into account by considering self-connection between hidden layers (Napoli et al. 2015). The RNN model architecture aims to map input sequence to an output sequence. Still, in practice, the long-term dependency between instances is forgotten and not correctly and adequately learned and implemented. This problem is shared among all the gradient-based learning models (Bengio et al. 1994). To overcome this major drawback of RNN models, a modified RNN model by adding a special unit structure to RNN is proposed (Greff et al. 2016). The resulted RNN model is called long short-term memory (LSTM). Some of the related works to use LSTM to forecast short-term mard grid loads are presented in Li et al. (2020), Zheng et al. (2017), Choi et al. (2018), Jahangir et al. (2020), and Muneer et al. (2022).

2.2.2 Long-Term Forecast Worldwide energy consumption has been on the rise in the past few decades due to an increase in the human population and fast-moving toward industrialization. The energy industry and production are complex and not efficiently scalable. Many governmental and nongovernmental agencies need to estimate future energy production accurately for planning purposes. Future long-term energy production is dependent on many known and not-so-known factors. In that sense, such long-term forecasting is difficult compared to stock market forecasting (Bianco et al. 2009). Long-term consumption prediction is essential in developing smart grid systems and capacity expansions. Also, it is required to consider capital investment, revenue analysis, and market management (Ekonomou et al. 2010). On the other hand, numerous long-term attributes increase the uncertainty in future forecasts. Generally, econometric, statistical, and mathematical operations are widely used to predict future consumption. These models are easy to understand and implement, especially on traditional plant networks, but are difficult to be scaled on decentralized and heterogenous systems like smart grids (Khan et al. 2020). The typical difficulty in long-term consumption prediction in smart grids is the lack of knowledge in considering all the necessary information. A significant number of published research for long-term consumption prediction use conventional machine learning models. Da Silva et al. (2019) proposed a model that combines the bottom-up approach and hierarchical linear models. They have used Bayesian inference to include the model uncertainty. To manage the model uncertainty, we can also combine machine learning models with the econometric method. In these approaches, the demand will be predicted based on the conventional econometric methods, and then the results will be passed to other learning models like artificial neural networks (He et al. 2017). Different neural networks and deep learning models are widely sued to predict energy consumption. Similar to short-term forecasting, artificial intelligent algorithms have the flexibility to solve mathematical and optimization problems with the help of historical data available and collected over time in the system. Most of

Constructing Renewable Energy Systems Using Big Data Applications

357

the published research in long-term consumption prediction ignores the distributed energy resources. Many researchers are considering this critical feature to improve the quality of the forecast (Khuntia et al. 2016; Sauter et al. 2018). Yet, training and forecasting smart grid consumption face dire challenges (Chen et al. 2020).

3

Conclusion

The world is moving toward replacing fossil-fueled energy with renewable ones shortly. There is an immediate need to replace the current system with decentralized and heterogeneous systems to meet future needs. The traditional energy systems cannot be scaled to support multiple energy sources. Data-driven modeling plays an essential role in constructing such a system. A massive amount of data is collected in smart grid systems that can help to improve system integration, efficiency, and management planning.

References F.C.A.E.S. Abdollahy, A. Mammoli, J. Johnson, Distributed compensation of a large intermittent energy resource in a distribution feeder, in IEEE PES Innovation Smart Grid Technology, 2013, pp. 1–6 F.E. Abrahamsen, Y. Ai, M. Cheffena, Communication technologies for smart grid: a comprehensive survey. Sensors 21(23), 8087 (2021) A.A.G. Agung, R. Handayani, Blockchain for smart grid. J. King Saud Univ.-Comput. Inf. Sci. 8(1), 18–43 (2020) A. Ahmad, N. Javaid, A. Mateen, M. Awais, Z.A. Khan, Short-term load forecasting in smart grids: an intelligent modular approach. Energies 12(1), 164 (2019) T. Ahmad, H. Zhang, B. Yan, A review on renewable energy and electricity requirement forecasting models for smart grid and buildings. Sustain. Cities Soc. 55, 102052 (2020) N.Z. Aitzhan, D. Svetinovic, Security and privacy in decentralized energy trading through multi-signatures, blockchain and anonymous messaging streams. IEEE Trans. Depend. Secure Comput. 15(5), 840–852 (2016) N. Ayub, N. Javaid, S. Mujeeb, M. Zahid, W.Z. Khan, M.U. Khattak, Electricity load forecasting in smart grids using support vector machine, in International Conference on Advanced Information Networking and Applications (Springer, 2019), pp. 1–13 Y. Bengio, P. Simard, P. Frasconi, Learning long-term dependencies with gradient descent is difficult. IEEE Trans. Neural Netw. 5(2), 157–166 (1994) V. Bianco, O. Manca, S. Nardini, Electricity consumption forecasting in italy using linear regression models. Energy 34(9), 1413–1421 (2009) L. Chen, H. Yu, L. Tong, X. Huai, P. Jin, Y. Huang, C. Dou, Research on load forecasting method of distribution transformer based on deep learning, in 2020 7th IEEE International Conference on Cyber Security and Cloud Computing (CSCloud)/2020 6th IEEE International Conference on Edge Computing and Scalable Cloud (EdgeCom) (IEEE, 2020), pp. 228–233 H. Choi, S. Ryu, H. Kim, Short-term load forecasting based on ResNet and LSTM, in 2018 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm) (IEEE, 2018), pp. 1–6 R.F. Colmenares-Quintero, D.J. Quiroga-Parra, N. Rojas, K.E. Stansfield, J.C. ColmenaresQuintero, Big data analytics in smart grids for renewable energy networks: systematic review of information and communication technology tools. Cogent Eng. 8(1), 1935410 (2021)

358

N. Sohaee

F.L. da Silva, F.L.C. Oliveira, R.C. Souza, A bottom-up bayesian extension for long term electricity consumption forecasting. Energy 167, 198–210 (2019) D. Dudley, Renewable energy will be consistently cheaper than fossil fuels by 2020, report claims. Forbes, 2018. Accessed on 13 May 2022 L. Ekonomou, Greek long-term energy consumption prediction using artificial neural networks. Energy 35(2), 512–517 (2010) M. Ghalib, A. Ahmed, I. Al-Shiab, Z. Bouida, M. Ibnkahla, Implementation of a smart grid communication system compliant with ieee 2030.5, in 2018 IEEE International Conference on Communications Workshops (ICC Workshops), 2018, pp. 1–6 C. Greer, D.A. Wollman, D. Prochaska, P.A. Boynton, J.A. Mazer, C. Nguyen, G. FitzPatrick, T.L. Nelson, G.H. Koepke, A.R. Hefner Jr. et al., Nist framework and roadmap for smart grid interoperability standards, release 3.0 (2014) K. Greff, R.K. Srivastava, J. Koutník, B.R. Steunebrink, J. Schmidhuber, LSTM: a search space odyssey. IEEE Trans. Neural Netw. Learn. Syst. 28(10), 2222–2232 (2016) M.T. Hagan, S.M. Behr, The time series approach to short term load forecasting. IEEE Trans. Power Syst. 2(3), 785–791 (1987) A. Hahn, R. Singh, C.-C. Liu, S. Chen, Smart contract-based campus demonstration of decentralized transactive energy auctions, in 2017 IEEE Power & Energy Society Innovative Smart Grid Technologies Conference (ISGT) (IEEE, 2017), pp. 1–5 Y. He, J. Jiao, Q. Chen, S. Ge, Y. Chang, Y. Xu, Urban long term electricity demand forecast method based on system dynamics of the new economic normal: the case of tianjin. Energy 133, 9–22 (2017) H. Jahangir, H. Tayarani, S.S. Gougheri, M.A. Golkar, A. Ahmadian, A. Elkamel, Deep learningbased forecasting approach in smart grids with microclustering and bidirectional lstm network. IEEE Trans. Ind. Electron. 68(9), 8298–8309 (2020) A. Khan, H. Chiroma, M. Imran, J.I. Bangash, M. Asim, M.F. Hamza, H. Aljuaid et al., Forecasting electricity consumption based on machine learning to improve performance: a case study for the organization of petroleum exporting countries (opec). Comput. Electr. Eng. 86, 106737 (2020) S.R. Khuntia, J.L. Rueda, M.A. van Der Meijden, Forecasting the load of electrical power systems in mid-and long-term horizons: a review. IET Gener. Transm. Distrib. 10(16), 3971–3977 (2016) G.-H. Kim, S. Trimi, J.-H. Chung, Big-data applications in the government sector. Commun. ACM 57(3), 78–85 (2014) M. Lei, L. Tang, M. Li, Z. Ye, L. Pan, Forecasting short-term residential electricity consumption using a deep fusion model, in Proceedings of 2018 Chinese Intelligent Systems Conference (Springer, 2019), pp. 359–371 L. Li, K. Ota, M. Dong, When weather matters: IoT-based electrical load forecasting for smart grid. IEEE Commun. Mag. 55(10), 46–51 (2017) J. Li, D. Deng, J. Zhao, D. Cai, W. Hu, M. Zhang, Q. Huang, A novel hybrid short-term load forecasting method of smart grid using MLR and LSTM neural network. IEEE Trans. Ind. Inf. 17(4), 2443–2452 (2020) M.B. Mollah, J. Zhao, D. Niyato, K.-Y. Lam, X. Zhang, A.M. Ghias, L.H. Koh, L. Yang, Blockchain for future smart grid: a comprehensive survey. IEEE Internet Things J. 8(1), 18– 43 (2020) A. Muneer, R.F. Ali, A. Almaghthawi, S.M. Taib, A. Alghamdi, E.A.A. Ghaleb, Short term residential load forecasting using lstm recurrent neural network. Int. J. Electr. Comput. Eng. (IJECE) 9(4), (2022) E. Münsing, J. Mather, S. Moura, Blockchains for decentralized optimization of energy resources in microgrid networks, in 2017 IEEE conference on control technology and applications (CCTA) (IEEE, 2017), pp. 2164–2171 S. Nakamoto, Bitcoin: a peer-to-peer electronic cash system, in Decentralized Business Review, 2008, p. 21260

Constructing Renewable Energy Systems Using Big Data Applications

359

C. Napoli, G. Pappalardo, G.M. Tina, E. Tramontana, Cooperative strategy for optimal management of smart grids by wavelet rnns and cloud computing. IEEE Trans. Neural Netw. Learn. Syst. 27(8), 1672–1685 (2015) P.S. Sauter, P. Karg, M. Kluwe, S. Hohmann, Load forecasting in distribution grids with high renewable energy penetration for predictive energy management systems, in 2018 IEEE PES Innovative Smart Grid Technologies Conference Europe (ISGT-Europe) (IEEE, 2018), pp. 1–6 U.S. Energy Information Administration, Share of total us energy consupmtion by major sources in selected years (1776–2020). Online, April 2021, https://www.eia.gov/energyexplained/whatis-energy/sources-of-energy.php. Accessed on 14 May 2022 V. Singh, D. Joung, L. Zhai, S. Das, S.I. Khondaker, S. Seal, Graphene based materials: past, present and future. Prog. Mater. Sci. 56(8), 1178–1271 (2011) Eolas, Smart grid evolution. Online, March 2018, https://www.eolasmagazine.ie/smart-gridevolution/. Accessed on 13 May 2022 R. C. Lanphier, The history of making the grid smart. Online, March 2022, https://ethw.org/ The_History_of_Making_the_Grid_Smart#:∼:text=Automatic%20meter%20reading%20devices %20introduced,technology%20patented%20by%20Theodore%20Paraskevakos. Accessed on 13 May 2022 Infoplus, The role of smart grid, IoT, and big data in renewable energy. Online, May 2022, https:// www.infopulse.com/blog/role-smart-grid-iot-big-data-renewables. Accessed on 13 May 2022 N.B. Vanting, Z. Ma, B.N. Jørgensen, A scoping review of deep neural networks for electric load forecasting. Energy Inform. 4(2), 1–13 (2021) U.S. Energy Information Administration, What is energy? Online, May 2021, https://www.eia.gov/ energyexplained/what-is-energy/sources-of-energy.php. Accessed on 14 May 14, 2022 P. Zhang, X. Wu, X. Wang, S. Bi, Short-term load forecasting based on big data technologies. CSEE J. Power Energ. Syst. 1(3), 59–67 (2015) J. Zheng, C. Xu, Z. Zhang, X. Li, Electric load forecasting in smart grids using long-short-termmemory based recurrent neural network, in 2017 51st Annual Conference on Information Sciences and Systems (CISS) (IEEE, 2017), pp. 1–6 D. Zheng, K. Deng, Y. Zhang, J. Zhao, X. Zheng, X. Ma, Smart grid power trading based on consortium blockchain in internet of things, in International Conference on Algorithms and Architectures for Parallel Processing (Springer, 2018), pp. 453–459

Control and Monitoring of Wind Farms Based on IoT Application for Energy Conversion Mohammad Niyazi and Adel Nazemi Babadi

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Current Status (Before Utilization of IoT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Analytics Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

362 363 364 364 365

Abstract

Wind farms are comprised of large, expensive wind turbines. The bigger the turbine, the more power it is capable of producing. These wind farms are frequently located in remote areas and require high reliability, and the monitoring and controlling of such farms are very tedious. Wind farms may be located far inside the sea, between the mountains, or in forests. Thus, when minor faults occur, the people have to go to that particular location for fault clearing. It requires a lot of manpower and time, and faces severe economic difficulties. These limitations can be overcome by the use of the Internet of Things concept. Most of the existing products use memory cards or PCs for data logging. This stored data is accessible only on that particular PC. This limitation is also addressed by IoT technology. Keywords

Wind farms · Condition-based monitoring · IoT · SCADA

M. Niyazi () Smart MPower Company, Limerick, Ireland A. N. Babadi KNT University, Tehran, Iran © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_177

361

362

1

M. Niyazi and A. N. Babadi

Introduction

In the coming years, we will see IoT-enabled “talking” turbines, drones, and robots designed to replace the manual inspection of wind turbines and blades, as well as lightweight, high-strength composite materials to drive wind turbine size. However, the industry still faces a plethora of challenges. These include “integration with the electricity grid, withdrawal of government subsidies, and political instability.” Also, most of the sensor data collected today, such as for weather, traffic monitoring, or even healthcare, is outdated seconds after it is collected. In addition, shuttling data from the sensor at the offshore turbine to the cloud not only takes time but is also limited given the bandwidth to the remote locations. Therefore, if we can analyze some of the data in real time where it is collected, we can make decisions faster and, in some cases, automate them, such as shutting down a turbine to avoid cascading damage. Despite these challenges, the sector is growing both in its existing strongholds in Europe and North America, and via new investments in Asia. North America and Europe will continue to dominate the sector, accounting for 45.9% of global revenues. However, China will become an increasingly powerful player, generating 39.3% of global revenues. India, Brazil, France, Spain, Germany, and the UK are also leading expansion (Kordestani et al. 2018). Governments are among the biggest investors in the wind market, launching projects that promote energy security and reduce coal consumption in response to global warming. With the growing number of wind power plants being erected offshore, there is a need for commercial, predictive, and proactive maintenance. A major portion of wind turbine downtime is due to bearing failures, particularly in the generator and gearbox. This can be overcome by installing vibration sensors. The fault rate of offshore wind turbines is so high that it is essential to build a reliable condition monitoring system to improve the maintenance efficiency (Singleton et al. 2017). By using the central station IP address and port number of the SCADA program into the address line of a web browser, we can monitor the system via any web browser on an internet-connected computer. But SCADA monitoring and control systems are very expensive; the installation of a SCADA system is also very difficult; frequent maintenance is required; a SCADA system is more complex and has a measurement delay; and also it occupies more space for installation. These limitations can be easily overcome by using the IoT-based wind turbine monitoring and control system (WTMCS). IoT-based WTMCS consists of sensing unit, monitoring unit, and control unit. The purpose of sensor unit is to sense the wind turbine parameters; the monitoring unit’s purpose is to monitor the sensed values of turbine parameter by comparing with the reference limits and to give instruction to the control unit. The control unit controls the turbine parameters within the safe limit by controlling the relays (ON/OFF) and also giving alarms to alert. To maximize the output, wind turbines must be adjusted to deal with wind direction and wind speed. Whenever a turbine is down for maintenance, it is not producing energy, which lowers output and increases costs, thus decreasing

Control and Monitoring of Wind Farms Based on IoT Application for Energy. . .

363

profitability. If the angle of a blade on a turbine is off by a few degrees, it can have a substantial impact on the output of that turbine. As wind speeds and conditions change, the blades need to be adjusted in real time to compensate (Yu et al. 2015). Forecasting for fossil fuels like oil, gas, and coal is a much more predictable process. Once these resources are found, the methods of forecasting are fairly static. With renewables like wind and solar, forecasting is extremely dynamic due to changes in the weather. Real-time decision-making based on sensor data could create a huge competitive advantage for companies that could optimize for reliability and current conditions.

2

Current Status (Before Utilization of IoT)

The capacity of the wind power has increased in recent years, but the level of operation and maintenance management of the wind power plant is inadequate. There are the main current problems listed below. 1. Low efficiency of the wind turbine breakdown maintenance: The operators monitor the wind turbine status in the center-control room through the SCADA system, and when a breakdown occurs, operators can send fault messages to the maintenance staff. According to the fault messages from the SCADA system, the maintenance staff can deploy the wind farm operation and maintenance resource. However, the SCADA system can just supply the binary information of the faults, but not a clear description. Maintenance staffs need to grasp the specific situation after arriving at the spot. When they are unable to solve the faults, they have to ask for assistance from the remote maintenance experts or the third-party manufactories. Wind farms are generally located in remote regions with low communication signals; therefore, it is not effective to exchange the information between the operation and maintenance staff and the remote experts. In this situation, the remote experts should have to travel personally to the location to solve the faults, which cause the long breakdown time of wind turbines. In other words, experience from experts cannot be effective to copy and shared, and maintenance resource cannot get effective to utilize to keep the breakdown maintenance efficiency. 2. Long period of the wind turbine inspection: The general way of the wind turbine inspection is spot check and the routing inspection mode. In the inspection process, the maintenance staff need to climb the wind turbine one by one, because the wind power is the typically distributed generate system with little capacity standalone, large quantity, and broad distribution. However, the time and human cost are very high to inspect the wind turbine. In addition, because the maintenance focus is not prominent, the maintenance staffs have to climb the wind turbine one by one, which means that the maintenance resources cannot be planned optimal. The long period of the maintenance affects the general generation level of the wind farm.

364

M. Niyazi and A. N. Babadi

3. Low level of the maintenance management: At present, most wind farms use the traditional artificial mode and need to be approved step by step to manage the worksheets. Nevertheless, because the manager is out of the wind farm, the approbation processing will be delayed, which will extend the response time after the faults occur and decrease the generation time of the wind turbine. Moreover, during the maintenance processing, paper records are required, which are kept in the monitoring center. When it is needed to search the historical records, the maintenance staff should go to the control center room, which would cost the delay of the repair time. Meanwhile, paper records are not convenient to keep or even to search. In wind farm asset management, all the information will be managed in the paper ledger, which includes the equipment data, factory information, debugging, fault information, etc. When it is needed to search the information, the maintenance staff should go to the control center room, which would waste time and restrict the maintenance efficiency seriously.

3

Analytics Platform

For a wind farm, maximizing power output is obviously critically important. A powerful and reliable industrial computing platform is one of the keys to ensuring efficient power generation – together with advanced software to control the wind turbine and respond to changes in wind speed and direction. The computing platform has three important tasks in wind power generation: • Preprocess the data to filter garbage information and only send useful information back to the control center. This makes the best use of control center resources and also ensures timely delivery of data by reducing bandwidth demands. • Constantly monitor critical devices’ status to ensure they are functioning correctly. This allows operators to take appropriate steps at an early stage and avoid turbine downtime. In addition, if the computer detects an abnormal situation, it sends an alert, so that timely repair and maintenance can minimize losses. • Communicate in real time with the remote-control center, allowing the control center to adjust turbine blade pitch angles and other parameters according to changes in wind speed and other conditions (Fig. 1).

4

Conclusion

Today’s wind turbine monitoring platform uses low- and high-frequency SCADA scheme that has technical and economic challenges such as a time-consuming and ineffective cost process, binary information of faults and failures, short-term data storage, and an unintegrated system. In this chapter, a comprehensive IoT-based methodology including sensors, IT networks, local and wide area communication media, database, and data analytics platform is presented that can guarantee

Control and Monitoring of Wind Farms Based on IoT Application for Energy. . .

365

Fig. 1 Proposed platform

long-term data storage, full-range faults and failure root cause analysis, and low noise data transfer media. The proposed IoT scheme is categorized into three different layers: hardware, software, and services, and each layer has some needed component with explained functionality. The presented methodology can improve cost, time, and personnel efficiency with increased remaining useful lifetime and fault-tolerant mechanical and electrical structures.

References M. Kordestani, M.F. Samadi, M. Saif, K. Khorasani, A new fault prognosis of MFS system using integrated extended Kalman filter and Bayesian method. IEEE Trans. Ind. Inf., 1–11 (2018) R.K. Singleton, E.G. Strangas, S. Aviyente, The use of bearing currents and vibrations in lifetime estimation of bearings. IEEE Trans. Ind. Inf. 13(3), 1301–1309 (2017) M. Yu, D. Wang, M. Luo, An integrated approach to prognosis of hybrid systems with unknown mode changes. IEEE Trans. Ind. Electron. 62(1), 503–515 (2015)

Smart Grid and Resilience Zahra Zand, Muhammad Reza Ghahri, Soheil Majidi, Mostafa Eidiani, Morteza Azimi Nasab, and Mohammad Zand

Contents 1 2 3 4

5 6 7 8

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introducing the Challenge and Necessity of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Part I: Investigation of the Effect of Charging Electric Combined Vehicles on Power Losses and Voltage Deviations in the Distribution Network . . . . . . . . . . . . . . 4.1 Uncoordinated Charging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Coordinated Charging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Part II: Investigating the Effect of Increasing the Number of Vehicles on Losses and the Cost of Investing in the Power Network . . . . . . . . . . . . . . . . . . . . . . . Reduce Investment in Peak Hours with Smart Charging Strategy . . . . . . . . . . . . . . . . . . Transfer of Charging Time to Non-peak Hours . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1 Increased Energy Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Part III: Optimal Charging of Electric Vehicles by Observing the Restrictions of the Maximum Distribution and Transmission Network . . . . . . . . . . . 8.1 Variables Required to Operate the System under Study . . . . . . . . . . . . . . . . . . . . . . 8.2 Analysis of the Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

369 371 372 375 375 376 378 380 380 381 381 382 382

Z. Zand Razi University, Kermanshah, Iran M. R. Ghahri Sharif University of Technology, Tehran, Iran S. Majidi Research and Development Department, BLUE&P group, Tehran, Iran e-mail: [email protected] M. Eidiani Khorasan Institute of Higher Education, Mashhad, Iran e-mail: [email protected] M. A. Nasab · M. Zand () CTIF Global Capsule, Department of Business Development and Technology, Denmark and Renewable Energy Lab (REL), Aarhus University, Herning, Denmark © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_178

367

368

Z. Zand et al.

9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Verify of Results by Conventional Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Suggestions for Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

388 388 388 389

Abstract

Given that understanding the effects of electric vehicle charging on smart grids and taking action to eliminate or reduce these effects requires planning the power grid and implementing long-term preventive strategies, it is necessary to gain sufficient knowledge in this field before using them extensively. For this reason, many researchers have studied the various effects of connecting these vehicles in the fields of changing the load profile of consumers, increasing losses and decreasing the voltage of the power grid. However, no special attention has been paid to the effects of charging electric chargers on the life of smart grid electrical equipment. However, by connecting electric vehicles to the grid, the amount of power passing through the grid equipment increases and this can reduce their lifespan. Due to the connection of electric vehicles to household sockets, it is expected that smart grids will be more affected by the charge of these vehicles than the production and transmission sectors. Distribution transformers are also one of the most important smart grid equipment’s that have high costs and are found in abundance in distribution networks. For this reason, this chapter focuses on studying the optimal scenario of energy management scenario in the electric vehicle solar charging station in smart grids along with solid-state transformers in the presence of electric vehicles. The following book is followed: (1) In this chapter of the book, we design a solar charging station and present an automated energy management (EMS) strategy for solid-state transformers by adding optimization techniques to the automated energy management, to design an efficient and optimal real EMS strategy for PVCS stations in Smart network ancillary services are paid. In this way, first, the solid-state transformer is selected for PV integration according to the different initial charge conditions of the EVs batteries, and then, considering that the solar charging stations (PVCS) are active as a controllable unit in smart grids. It was active once that using a solid-state transformer, the volume of PVCS would be much smaller. (2) In this chapter of the book, the solar charging stations of the electric vehicle in the smart grid, which is managed by SST transformers, are examined and its optimization is examined. Also, the energy management department in charging stations will be improved by using the MNHPSO-JTVAC optimization method in order to increase the efficiency and accuracy of the optimal power allocation performance. The optimal scenario-based strategy presented in the book also includes the connection of electric vehicles to the network and the determination and evaluation of different scenarios due to different penetration coefficients of the solar charging station and different charging intervals of electric vehicles in managed smart grids. Will be reviewed by SST.

Smart Grid and Resilience

369

Keywords

MNHPSO-JTVAC optimization · Smart grid · Photovoltaic Assisted Charging Station (PVCS) · Energy management strategy · SST transformers Abbreviations

BMS C(t) Ci (n) CPAA DAB DES EBC Ei chg (n) EMS ERCOT EV I(t) MIDC MPPT NREL PHEV Pi chg (n) PV PVCS PWM Q Qi S(t0 ) SOC SST STC Ui (n) t

1

Battery management system Electric vehicle charge rate The charge rate of an electric vehicle in the time interval is n Charging power allocation algorithm Dual Active Bridge Distributed energy resources Energy Bound Calculation Electric vehicle energy demand in time interval n Energy management strategy Electric Reliability Council of Texas Electric vehicles Electric vehicle charging current Measurement and Instrumentation Data Center Maximum power point tracking National Renewable Energy Laboratory Plug-In Hybrid Electric Vehicle Electric vehicle charging power in time interval n Photovoltaic Photovoltaic-assisted charging station Pulse Width Modulation The allowable capacity of electric vehicle battery Unspecified ampere-hour capacity of electric self-contained battery The value of the charge state at time t0 State of charge Solid state transformer Standard test conditions The battery voltage of the vehicle is electric Interval

Introduction

Intelligent network components include distributed energy sources, control and management microsystems, secure communications, and secure information management (Zand et al. 2020a). Solid state transformers with the use of electronic power equipment have additional capabilities such as the supply of reactive power required for the network, high power quality, fault current limitation, energy storage and production management and DC bus for the final consumer. Will also has.

370

Z. Zand et al.

Topology selection is very important for solid state transformers. There are six popular topologies in SST such as Pulse Width Modulation and Dual Active Bridge. For this reason, this chapter focuses on studying the optimal scenario of energy management scenario in the electric vehicle solar charging station in smart grids along with solid-state transformers in the presence of electric vehicles. In general, it can be said that the solid-state transformer has the following functions and benefits: 1. 2. 3. 4. 5. 6. 7. 8. 9.

Voltage level control Volume and weight control Instantaneous and instantaneous voltage adjustment Error separation Power factor correction Controlling the distribution of active and reactive powers Controlling the fault current on the strong and weak pressure side High voltage adjustment capability Possibility of difference between output frequency and input and number of output phases 10. Possibility of having DC input or output 11. Fix voltage drop problems (if you have a storage) 12. Fixing the problem of load imbalance and solving the problem of zero wire current and losses due to this problem, as well as eliminating or reducing the phenomenon of imbalance in the distribution network The following book is followed: (1) In this chapter of the book, we design a solar charging station and present an automated energy management (EMS) strategy for solid-state transformers by adding optimization techniques to the automated energy management, to design an efficient and optimal real EMS strategy for PVCS stations in Smart network ancillary services are paid. In this way, first, the solid-state transformer is selected for PV integration according to the different initial charge conditions of the EVs batteries, and then, considering that the solar charging stations (PVCS) are active as a controllable unit in smart grids. It was active once that using a solid-state transformer, the volume of PVCS would be much smaller. And (2) In this chapter of the book, the solar charging stations of the electric vehicle in the smart grid, which is managed by SST transformers, are examined and its optimization is examined. Also, the energy management department in charging stations will be improved by using the MNHPSO-JTVAC optimization method in order to increase the efficiency and accuracy of the optimal power allocation performance. The optimal scenario-based strategy presented in the book also includes the connection of electric vehicles to the network and the determination and evaluation of different scenarios due to different penetration coefficients of the solar charging station and different charging intervals of electric vehicles in managed smart grids. Will be reviewed by SST.

Smart Grid and Resilience

2

371

Introducing the Challenge and Necessity of the Study

The rapid development of technology in the transportation sector, along with environmental concerns about rising oil prices, has led to efforts to reduce dependence on fossil fuels in many countries around the world. One of the efforts made in this field is investing in the design and construction of electric vehicles. Based on this, many reputable Vehicle companies have started commercial production of electric vehicles. Commercial construction of electric vehicles is done in the form of fully battery-based vehicles and gasoline-electric combined vehicles. Fully batterypowered vehicles are powered by rechargeable batteries mounted on the vehicle. For the daily operation of these vehicles, the batteries in the vehicles must be charged at stations in the city. The use of these vehicles was not welcomed due to the long waiting time to charge the battery in the charging stations as well as the lower performance in terms of speed and acceleration compared to gasoline vehicles. However, with many reputable Vehicle companies focusing on improving the quality of fully battery-based vehicles and using fast charging to charge them, these vehicles are expected to find their market in specific applications in the near future. On the other hand, over the past few years, the production of electric vehicles has been on the agenda of Vehicle manufacturers and the supply of these products to the market is increasing. Due to the reduction in the price of the technology used in these Vehicles, it is expected that the customer acceptance of this Vehicle will increase significantly in the near future. In gasoline-electric combined vehicles, the driving force of the vehicle is produced by combining the combined operation of a gasoline combustion engine and an electric motor. In this combination, on the one hand, the amount of fuel consumption in these vehicles is reduced compared to gasoline vehicles, and on the other hand, their performance is improved compared to fully battery-based EVs. A noteworthy point in increasing the use of hybrid vehicles is that the technology used in the batteries of these vehicles, allows them to be charged from any external source. This external power source can be the power grid, and these vehicles can be charged in any possible place, such as residential houses where the power grid is easily accessible. The easy charging of these Vehicles allows its owners to use the Vehicle during their daily trips within the city and when they return home, by connecting the Vehicle to the household electrical socket, prepare their Vehicle for the next day’s trip. During the period of connecting these vehicles to the power socket, the energy required to recharge the batteries of combined vehicles is supplied from the power grid. In this regard, it is necessary to pay attention to the fact that the use of these vehicles, which has been developed to save fuel consumption, can cause adverse effects on the power grid. It is possible to reduce the peak load power by transferring power by charging and discharging EVs. However, the design and construction of many of the power networks used has been done without considering the possibility of connecting this equipment to the network. Because unknowingly charging equipment through power grids can lead to

372

Z. Zand et al.

many problems in the normal operation of networks, many researchers have studied the technical effects of connecting rechargeable electric vehicles to the performance of power grids. Given that understanding the effects of charging electric vehicles on the power grid and taking action to eliminate or reduce these effects requires planning the power grid and implementing long-term preventive strategies, it is necessary to gain sufficient knowledge in this field before using them extensively. For this reason, in many references (Zand et al. 2020a, b; Chen et al. 2021a; Azimi Nasab et al. 2021; Makolo et al. 2021), researchers have studied the various effects of connecting these vehicles in the areas of changing the load profile of consumers, increasing losses and decreasing the voltage of the power grid. However, no special attention has been paid to the effects of charging electric charges on the life of electrical equipment in the power grid. However, by connecting electric vehicles to the power grid, the amount of power passing through the grid equipment increases, and this can reduce their lifespan. Due to the connection of electric vehicles to household sockets, distribution networks are expected to be more affected by the charge of these vehicles than the production and transmission sectors. Distribution transformers are also one of the most important distribution networks equipment’s that have high costs and are found in abundance in distribution networks. For this reason, this chapter of the book focuses on the study of solid-state transformers in the presence of electric vehicles.

3

Literature Review

Different structures for EMS using different optimization algorithms and different configurations for smart microgrids are presented in the references, some of which aim to optimize the performance of each resource in the microgrid system in order to minimize system operating costs. Much research has been conducted. Much research has been done in this field, including the most recent research, which can be referred to in Zand et al. (2020a), which uses the SST for the energy management strategy for solid state transformers for smart Photovoltaic vehicle charging stations (SPVCS). Also, the energy management department in charging stations has been improved by using the scenario-based optimization method. Solid state transformers have also been selected for PV integration and storage. The advantages of this method include: SST enables the control of the passage of electrical power and the flexible connection of distributed generation sources to the network, as well as power distribution that is safe and stable operation. It is important for the network to be controlled by this equipment, it is also noted. Corresponding charge strategies for Plug-In Hybrid Electric Vehicles (PHEV) by providing an instant energy management algorithm for grid-connected charging parks in industrial or commercial locations are presented in Chen et al. (2021a). In Zand et al. (2020b), an adaptive fuzzy strategy for energy management of hybrid vehicles is presented. Due to their independence from mathematical modeling and high flexibility, which ultimately compares cost, fuel consumption, and pollution among traditional hybrid vehicles, new vehicles have been tested. The simulation

Smart Grid and Resilience

373

results in MATLAB software show that the proposed design has better performance than a tracking controller due to the online updating of its parameters, reducing fuel consumption and tracking the reference speed. In Azimi Nasab et al. (2021), an interest strategy in innovative radar has been proposed for microgrids of commercial smart buildings, including photovoltaic array systems and EVs. Principles and decision-making strategies to improve photovoltaic energy consumption and reduce adverse effects on the power grid are predefined. In Makolo et al. (2021), the performance of EVs alongside renewable energy sources (RESs) in the smart grid is presented. According to this paper, EVs play their role as service providers for the network, such as voltage and frequency stabilization, pixel load curve through power injection into the network, reactive power support to increase operating efficiency, network power security, and cost reduction exploitation. In Chen et al. (2021b), a wavelet transform-based strategy is proposed to manage the power of the PHEV, fuel cell, battery, and super capacitor. The proposed wavelet transform algorithm has the ability to detect the high frequency of transit and real-time power requirements for the PHEV. In Gaber et al. (2021), a model of EMS is used using a combination of systems from several power supplies, and the system is simulated in the PSPICE environment. The combined system studied in this paper includes a FC, a battery, a super capacitor, and a photovoltaic cell. The battery is used as the main power supply for driving the vehicle. Other sources are used as complementary power supplies for use in the medium range of the vehicle. The control algorithm has been developed as a keying mechanism used in setting up energy sources based on the vehicle’s power system drive. In Ahmadi-Nezamabad et al. (2019), one-way communication of vehicle connection technology to the network is more welcomed by the consumer, and by using the aggregator to control the amount of vehicle battery charge and the price of electricity in the network, it has proposed an intelligent charging algorithm that benefits the consumer, the aggregator, and the utility network. The technology in question will also provide the consumer with maximum battery life at the lowest cost, maximum profit for the aggregator, and increased flexibility for the power grid in the presence of renewable energy sources. In Solanke et al. (2021), by analyzing the two real distribution networks, the amounts of power losses and investment costs for different scenarios of vehicle penetration coefficient are examined and analyzed. Also, in terms of size, population, low-pressure and high-pressure subscribers, a distribution network design model has been studied and analyzed in different penetration coefficients, and the required investment and power losses in different penetration coefficients have been calculated and presented. In Hou et al. (2021), he proposes an innovative operational strategy for micro-networks of commercial buildings, including solar charging and EVs. Principles of decision-making and strategies to improve the self-consumption of photovoltaic energy and reduce the impact on the power grid are predefined. In addition, in order to decide on network charging power under time of use (TOU) pricing, online learning is integrated with a regular decision to achieve an instant online algorithm. In Hao et al. (2021), an electric vehicle classification scheme for a solar charging station that reduces the periodic effects of electricity storage and energy transaction costs. In Chakraborty et al. (2021), one-way communication of vehicle connection technology to the network is more welcomed by the consumer, and by using the aggregator to control

374

Z. Zand et al.

the amount of vehicle battery charge and the price of electricity in the network, it has proposed an intelligent charging algorithm that benefits the consumer, the aggregator, and the utility network. The technology in question will also provide the consumer with the maximum battery charge at the lowest cost, maximum profit for the aggregator, and increased flexibility for the power grid in the presence of RESs. In Dagar et al. (2021), consider a mathematical model for the economic evaluation of EVs for the smart grid response. The aim of this research is to develop mathematical modeling for vehicle aggregation. In fact, this model is formulated to evaluate the transfer of energy from vehicle to network and network to vehicle. The loss capacity model is used for V2G battery performance and is based on the total energy transferred between the network and the vehicle, the money exchanged between them, and the vehicle subscribers. In Baumgarte et al. (2021), the vehicleto-network ether examines the load curve in the smart grid environment, then the network operation costs and the wide participation of distributed generation units in the distribution network in the smart grid environment. In this paper, a more efficient management for vehicle charging and discharging in the form of a multi-objective optimization problem is considered to reduce the total cost of network operation. In Bamisile et al. (2021), the concept of vehicle-to-grid is introduced as a charging station, a place where all EVs from a specific area will discharge their charge or energy from a similar station. The V2G controller and charging station controller are designed to control the energy flow between EVs and the grid using fuzzy logic. Ameli and Ameli (2021) has examined the presence of EVs in the Internet of Things and has produced scenarios of their behavior based on the data collected from EVs by fuzzy logic and Monte Carlo methods. In Bayendang et al. (2021), a deep learning model based on recursive neural networks and convolutional neural networks is proposed that predicts urban traffic. In Vogt et al. (2021), time series, recursive neural network, and long short-term memory (LSTM) methods have been used to predict the Turkish load, and the results obtained in the paper show the success of this model of time series in prediction. In addition, in Luo et al. (2021), the uncertainty of scattered wind generation is also considered for more accurate modeling. In order to model the wind speed uncertainty, rail probability distribution functions have been used, which are created after the discretization of probabilistic scenarios. In Rehman et al. (2021), the mode of operation of centralized charging and unified distribution is introduced, in which the model of a centralized smart charging station is examined. Researchers in EV (Solanke et al. 2020; Turksoy et al. 2020) have examined EV charge demand models for urban settings. And in Goli et al. (2021), they developed a time-space model that integrated a transport analysis with a power system analysis, and in Tao et al. (2021), a time-based EV charge demand forecasting model with multiple charging stations in one. They predict the EV entry rate at a charging station with a cell transfer traffic model and (Lv et al. 2020) the time-space aspect of the charging station using modeling and simulation. In R˘aboac˘a et al. (2021), the problem of modeling of smart parking from the perspective of an electric vehicle integrator, in order to maximize its profit, several economic and technical indicators along with security restrictions have been integrated into the model of smart parking data. The system is intended. Sadr et al. (2020) proposes a dual-innovation model. At the first level, a new model

Smart Grid and Resilience

375

for charging parking for EVs is presented, which accurately models the features of EVs. At the second level, a new model is developed to ensure that technical constraints on distribution networks are developed while minimizing overall system costs. In addition, the entry of renewable sources such as PV and WT is included. Studies have pointed out that the goals of the energy management system have been mainly about economic or environmental costs as well as the quality of electric vehicle services. However, as an important flexible source of demand, EVs are able to have more productive effects by participating in smart grid ancillary services.

4

Part I: Investigation of the Effect of Charging Electric Combined Vehicles on Power Losses and Voltage Deviations in the Distribution Network

Connecting hybrid electric vehicles to the grid through household sockets or public Vehicle charging parking lots increases the Vehicle charging current to the distribution network. Excess load due to Vehicle charging causes power losses and voltage deviations in the network. From the perspective of the power system operator, power losses and factors affecting power quality such as voltage profiles and voltage imbalances while charging vehicles is an economic and technical concern that must be limited and controlled (Lv et al. 2020). By applying controlled charging to vehicles, power losses and voltage deviations are controlled within the allowable range. Therefore, by solving the optimization problem with the aim of minimizing power losses, the optimal charge of each charger is determined. In each charging period, the Vehicle must be ready for the next day’s journey, so in optimization, the condition that the battery reaches full charge during the charging period is added to the optimization problem. The time of disconnection from the charge is also determined by the Vehicle owner. The effect of controlled charging compared to uncontrolled charging has been investigated and analyzed. The IEEE34-nodetest feeder with home load has been used for this purpose. Also, two 2 h daily load profiles based on 15 min load, one for winter and the other for summer, are considered in this dissertation (Lv et al. 2020).

4.1

Uncoordinated Charging

Today, owners of electric vehicles have neither the information nor the specific incentive to schedule and control vehicle charging, and to make optimal use of the distribution network. In this way of connecting the combined electric vehicle to the network, there is no intelligent control or measurement system. Therefore, the Vehicles are charged without coordination and schedule. In other words, the battery starts charging as soon as it is connected to the distribution network or after a fixed time delay. In this method of charging, the current taken from the distribution network is also naturally high (R˘aboac˘a et al. 2021). The effect of uncoordinated charging of vehicles on the sample distribution network is calculated and shown in the table of values of voltage deviation and

376

Z. Zand et al.

Table 1 Loss to power ratio for 4 KW charger and uncoordinated charge (Lv et al. 2020) Charging period 21h00–06h00 18h001h00 10h00–16h00

Penetration level Summer Winter Summer Winter Summer Winter

0% 1.1 1.4 1.5 2.4 1.3 1.7

10% 1.4 1.6 2.4 3.4 1.8 2.2

20% 1.9 2.1 3.8 4.8 2.6 3.0

30% 2.2 2.4 5.0 6.0 3.2 3.6

Table 2 Voltage deviation for 4 kW charge and uncoordinated charge (Lv et al. 2020) Charging period 21h00–06h00 18h001h00 10h00–16h00

Penetration level Summer Winter Summer Winter Summer Winter

0% 3.1 4.2 3.0 4.8 3.0 3.7

10% 3.5 4.4 4.4 6.3 4.1 4.9

20% 4.4 4.9 6.5 8.5 5.6 6.4

30% 5.3 5.5 8.1 10.3 6.9 7.7

power dissipation. Network load includes household load and charger load (if any). As the results show, in all cases, power losses in winter are higher than in summer, and this is due to the higher home load in this season. Also, increasing the number of vehicles (or increasing the penetration rate of vehicles) increases power losses (Lv et al. 2020). As can be seen from Tables 1 and 2, power losses and voltage deviations in the evening charging time range (between 18 and 21 h) have the highest values. The reason for this is two things. First, the charge must be done within 4 h, so the charger must work at its maximum power so that the batteries reach their maximum capacity, and on the other hand, the household load consumption has its maximum amount in this time period. The deviation of the mains voltage from the nominal value (230 V in this paper) is presented in Table 2. According to the results, increasing the number of Vehicles causes a significant increase in voltage deviation. At a penetration coefficient of 30%, the voltage deviation reaches 10% at the peak time of the evening, which is the maximum acceptable deviation in the network. Voltage profiles in one node for penetration coefficients of 0% and 30% in the winter night period show the greatest voltage drop occurs during the charging period between 4 and 23 h when most Vehicles are in charge mode due to the higher power of chargers compared to home load. Is.

4.2

Coordinated Charging

The idea of coordinated charging is presented with the aim of achieving optimal charging and optimal use of the network to reduce losses. The implementation

Smart Grid and Resilience

377

of this technique in practice is possible by direct control of the Vehicle charge, intelligent measurement of node voltage and charge currents and sending a signal to the Vehicle. The optimization program in this dissertation has been done using exponential programming technique. This technique optimizes the quadratic function of several variables, which here are the power of the Vehicle chargers, under linear constraints. With power losses minimized, Vehicle owners will not be able to control the charge profile and will only choose when to finish charging the batteries. The end time of the charge is usually chosen as the time required for the Vehicle to be ready. One of the constraints of the optimization problem is the power of the chargers, which varies between zero and four kilowatts and is a constant value, and another important constraint is that the batteries must be fully charged at the end of the charging phase. Optimization is performed in different time intervals, seasons and penetration coefficients and the results are compared with uncontrolled charging. The values for voltage deviation and power losses in synchronized charge mode are shown in Table 3. By comparing the tables for synchronized charging and uncoordinated charging, it is determined that for all charging intervals and seasons, power losses with synchronized charging are reduced. Also, voltage deviation according to EN50160 (Lv et al. 2020) is acceptable for a penetration coefficient of 30% and its value is less than 10%. But if the number of vehicles increases, the amount of increase in losses and voltage deviations for charging day will be more than night. Figure 1 shows that the maximum voltage deviation occurs during the night charge with zero number of vehicles, at the beginning of the charge interval and when the household loads are still high. As can be seen from the figure, the uniformity of the voltage curve related to the penetration coefficient of 0 and 10%, means that the Vehicles are not charged during the peak load of household loads. At a penetration coefficient of 0 and 10% and at the beginning of the night charging interval, we will have the maximum voltage deviation, which is due to the household peak load. As time goes on and the home load decreases, the Vehicles start charging gradually, causing the voltage in the network to decrease. At a penetration rate of 30%, when a number of Vehicles are inadvertently charged at peak load, the mains voltage is lower.

Table 3 Loss ratio to total power for 4 kW charger and coordinated charge (Lv et al. 2020) Charging period 21h00–06h00 18h001h00 10h00–16h00

Penetration level Summer Winter Summer Winter Summer Winter

0% 1.1 1.4 1.5 2.4 1.3 1.7

10% 1.3 1.5 2.3 3.3 1.7 2.1

20% 1.7 1.8 3.7 4.7 2.3 2.7

30% 1.9 2.1 4.7 5.8 2.8 3.2

378

Z. Zand et al.

Fig. 1 Voltage profiles in one node for 0%, 10% and 30% penetration coefficients for coordinated charge

5

Part II: Investigating the Effect of Increasing the Number of Vehicles on Losses and the Cost of Investing in the Power Network

The significant growth of the use of electric vehicles in the future and the imposition of the charging current of these vehicles on the network, will increase power losses and increase the cost of investment in the network. In Sadr et al. (2020), by analyzing the two real distribution networks, the amount of power losses and investment costs for different scenarios of vehicle penetration coefficient are examined and analyzed. In the study, a residential area (A) and an industrial residential area (B) with the following characteristics in terms of size, population, low pressure and high pressure subscribers by a distribution network design model, in different penetration coefficients are analyzed and analyzed. The required investment and power losses in different penetration coefficients are calculated and presented. Initially, the electricity network was designed in two residential areas, A and B, based on the absence of an electric vehicle. Three scenarios of 35%, 51% and 62% are defined for the two regions. These penetration coefficients are predicted for 2020, 2030 and 2050, respectively. Then, using the distribution network planning model, the performance of the distribution network in each scenario, during peak and off-peak hours and different charging modes are examined and analyzed, and the necessary planning to strengthen the necessary structure and investment, as well as incremental network losses. Calculated and presented in each scenario.

Smart Grid and Resilience

379

The distribution planning model used as a sample design to assess the effect of electric vehicle penetration on investment costs and incremental energy losses. According to the initial design, the base distribution network, based on minimizing investment and operation costs, provides the required electricity to the network by observing the voltage drop and high reliability. The distribution planning model used in the dissertation is based on the innovative algorithm and the use of geographical features. By performing the optimization, the objective function of investment cost and maintenance and energy losses is minimized during the planning period and all network equipment and power lines and substations are designed and sized. Also, by determining the technical and economic parameters of the model, investment costs, maintenance and losses are shown along with details of voltage level and how to install in the network. In addition, a graphical representation of the network position is provided, such as Figs. 2 and 3. By applying different scenarios, the penetration coefficient of the electric vehicle, the network requirements and the amount of reinforcement of the network components are determined. In this dissertation, the charging points of Vehicles with the corresponding power are randomly placed in the area. Two loads are played for each scenario. One load is for charging during peak hours and the other is for charging during off-peak hours.

Fig. 2 Residential area A (Sadr et al. 2020)

380

Z. Zand et al.

Fig. 3 Residential industrial zone B (Sadr et al. 2020)

6

Reduce Investment in Peak Hours with Smart Charging Strategy

In the networks studied in this research, the effect of using smart charging strategy to prevent simultaneous charging of electric vehicles during peak hours has been analyzed and the results are shown in Fig. 6. In this figure, two different modes are considered for two regions, one with a synchronization coefficient of one and the other less than one, which, as shown in the figure, reduce the synchronization coefficient by using smart charging, no need for reinforcement and incremental investment. Follows on the network. It should be noted that the synchronization coefficient is calculated as the probability of synchronization of Vehicle charging, assuming a uniform distribution of charging points and relevant time periods.

7

Transfer of Charging Time to Non-peak Hours

According to the evidence presented in the research, both zones A and B of distribution networks are designed to provide load consumption in peak mode, which can provide the required power to vehicles without the need to strengthen the network. Therefore, the idea of shifting the charging period from peak to nonpeak time can be proposed. To reduce the investment cost by transferring the Vehicle charge from peak hours to non-peak hours, the cost of investing during peak hours is reduced. Of course,

Smart Grid and Resilience

381

if the charge of all Vehicles is transferred to non-peak hours, there is a possibility of overload in the distribution system equipment and the need to strengthen the structure. The initial investment to supply electricity to electric vehicles for both areas has been calculated based on the percentage of base investment and without considering the transfer of charge period. By transferring the charging period from peak to nonpeak hours, the investment cost is reduced and their difference as investment savings is calculated as a percentage of the initial incremental investment and is shown in Sadr et al. (2020).

7.1

Increased Energy Losses

The effect of connecting electric vehicles to the grid on power losses during offpeak hours, where most vehicles are connected to the grid, is calculated and shown in the bar chart in Sadr et al. (2020). Energy losses increase significantly during off-peak hours, relative to baseline and Vehicle-free conditions. Measurements and calculations show that the increase in losses increases with the number of vehicles. On the other hand, the comparison of power losses in the two regions shows that the losses in zone A are higher than in zone B, which is due to high load density in zone A. Reach 62% penetration.

8

Part III: Optimal Charging of Electric Vehicles by Observing the Restrictions of the Maximum Distribution and Transmission Network

One of the methods to control the harmful effects of charging electric vehicles in the distribution network is to control the charging of vehicles in order to transfer maximum power by observing the limits of allowable voltage drop and overload in the network (Rafik et al. 2020). Using linear programming and charge control for a part of a distribution network. The planning results show that if the amount of Vehicle charge is controlled individually, the power required to charge the Vehicle is possible without the need to upgrade network equipment or low need. The results of studies and studies conducted on the analysis of the effects of charging electric vehicles show that if controlled and normal charging is applied to vehicles during non-peak times, the existing networks can provide the necessary energy to vehicles with low penetration coefficient. In contrast, uncoordinated charging, especially fast charging, if it coincides with the peak load of the home, causes overload and voltage deviation in the network. In this way, Vehicle owners are encouraged to charge their Vehicles during off-peak hours, and no Vehicle is disconnected from the network before reaching maximum capacity, and no Vehicle is added to the collection when charging begins. There is also intelligent measurement technology and the ability to control the load in the Vehicle or house, and the operator of the distribution system can control and

382

Z. Zand et al.

Table 4 Charging station parameters by PV SOC Objective 0.850

Energy (kWh) 36.0

Maximum charge discharge rate 0.20

Capacity (Ah) 100

Maximum charge rate 0.30

Calculated voltage (V) 360.0

manage the charge of vehicles remotely. The battery is charged through a singlephase household socket and the connection of the battery to the one-way distribution network is considered.

8.1

Variables Required to Operate the System under Study

To have optimal applications for distribution systems in the presence of electric vehicles need to have information from Vehicles will be like: 1. Vehicle charge level when entering the parking lot 2. Location of the Vehicle (in which parking lot is the Vehicle located) 3. How long does the Vehicle connect to the network? How long does the Vehicle have access to the network or network to the Vehicle? This information will be predicted using historical reference data 1 and neural network method for the next day and will be provided to the network optimization program in a fixed quantity. Also, the hypotheses of the case study are taken from Zand et al. (2020a) and are as follows: Permitted capacity of photovoltaic system is 200 kW, which has four charging station parameters by PV. Has been in Table 4. The number of electric vehicles (EVs) is equal to 60 Initial SOC value between 0 (fully discharged) and 1 (fully charged), as shown in Fig. 4 The starting time and charging time of electric vehicles are assumed from the statistical data of a town or a university as shown in Fig. 4. In the simulation, each time slot is set to 10 min, 78 time slots

8.2

Analysis of the Proposed Method

In this section, the results of the proposed improved method based on MNHPSO meta-optimization optimization with non-asymmetric method are compared and analyzed. In this section, MATLAB language is used to test and validate the performance of the proposed improved method. Researchers often use trial and error to determine population size, repetition frequency, and frequency of program

Smart Grid and Resilience

383

Fig. 4 The time of entry and exit of Evs and their SOC values Table 5 Adjustment parameters used in the method

Parameter The amount of

POP 50

Iteration 100

Run 10

execution in meta-optimization because the amount of these adjustment parameters depends on the problem structure, dimensions, type of algorithm, and program size. After trial and error, it was found that based on the structure of the problem and the dimension of the problem, which is considered two-dimensional, the size of the regulatory parameters of the evolutionary method used in Table 5 has been obtained. In this table, pop indicates population size, iteration and number of iterations of the algorithm. Because the number of times the proposed method is performed ten times, the computational time increases but the optimal results are obtained and there is no early convergence. The optimal accuracy obtained from the proposed method based on MNHPSO optimization in each repetition for the maximum repetition of 100 for the first to third cases, respectively, 0.8900, 0.9200, and 0.9325 (Figs. 5 and 6). According to Fig. 7, the values of optimized and non-optimized levels in the SOC schedule for 60 EV are obtained the accuracy of the applied optimization method is normalized between 0 and 1. So that if the accuracy is 1, it means the exact accuracy of the proposed improvement and the fact that it is close to 1 indicates the high accuracy of the improvement. Hence, Fig. 8 had a low accuracy of around 0.5 in the beginning of the repetition, and after 10 repetitions it has improved by 2%, and after that we see an almost direct ratio of repetition to accuracy, so that the repetitions

384

Z. Zand et al.

Fig. 5 Amounts of orders issued by the collector and power management. (a) The proposed method is based on optimization. (b) The method lacks optimization

are far advanced. And increases, the accuracy of improvement is also increased, and after 100 repetitions to 120 repetitions, there is no change in the answers, indicating the convergence of the method in 100 repetitions. Similarly, in Fig. 9. of the third

Smart Grid and Resilience

385

Fig. 6 Optimal scheduling values obtained for solar charging system components. (a) The proposed method is based on optimization. (b) The method lacks optimization

case, we see the upward trend of convergence with increasing repetition, so that with each increase of repetitions, the accuracy has increased, and in this case, as is clear, after 65 repetitions, we see convergence so that Repeat 65 onwards does not

386

Z. Zand et al.

Fig. 7 Values of optimized and non-optimized levels in the SOC schedule for 60 EV. (a) The method lacks optimization. (b) The proposed method is based on optimization

show a change in the accuracy of the proposed improvement result, which means that no change in the improvement responses was obtained from the mentioned iteration, and the MNHPSO optimization method was converged to find the optimal

Smart Grid and Resilience

387

Fig. 8 Accuracy of charge power distribution optimization section – case 2

Fig. 9 Accuracy of charge power distribution optimization section – case 3

scheduling response, and this indicates efficiency. A good and appropriate method of improvement proposed in this dissertation is to present an optimal and efficient scheduling model in solid-state transformers.

388

9

Z. Zand et al.

Conclusion

It seems that the results of the proposed improvement on the third and second case are better than the compared method, in the method without improvement. The reason for this improvement can be found in the advantage of using the MNHPSO technique to optimize the charging schedule in the transformer energy management department. In other words, the advantage of the proposed method compared to the method that is not equipped with the technique of improving the dynamic power distribution using evolutionary methods (method without scheduling improvement with evolutionary optimization) under automatic segmentation and finding global optimal solutions Along with the worst global scheduling solutions to consciously calculate the charging time and changes in the initial charging mode. In other words, to optimize and evaluate different power distribution solutions using the MNHPSO optimization method, Mitt One automatically divided the entire power distribution solutions into groups so that each group around the state with the best moves all over the world. Among all these cases, the best and worst global solution is found. Secondly, this subdivision allows problem solutions to be able to find all the optimal points simultaneously. Therefore, due to the non-random selection of solutions in the stages after the first iteration, the proposed method optimizes the charging power and then energy management.

10

Verify of Results by Conventional Methods

In this section, the proposed energy management method based on SST is discussed with conventional methods in terms of Moore parameters, charging schedule, etc., which are reported in Table 6.

11

Suggestions for Future Work

1. Due to the fact that the photovoltaic charging station consists of a photovoltaic system, the better the control of the mppt of the photovoltaic system, the more power is applied to the circuit, so by combining the mppt control system and the SST transformer, the efficiency of the charging station can be increased. 2. To improve the EMS control system, new super-innovative algorithms such as gray wolf, TLBO, deep neural network, etc. Can be used. 3. Solving the problem in the presence of different percentages of penetration of the photovoltaic system and determining the charging strategy of the electric vehicle. 4. Determining the optimal location of electric Vehicle stations while managing their optimal charge in the presence of solid-state transformers. 5. Using other common models of electric vehicles in comparison with each other in the amount of charge.

Smart Grid and Resilience

389

Table 6 Results of comparison based on key indicators

Indicators Set the highest power command Set the lowest power command Target charge demand Time settings The amount of energy required for electric vehicles Full charge rate The highest setting is finished The lowest setting rate expired Discharge time Revenue from restrictions imposed

Typical charging station 347.23

SST based charging station 311.10

SST-based charging station equipped with evolutionary optimization method 215.00

540.2

−411.59

−330.00

1147.6 27 731.72

1147.6 27 1072.4

1147.6 27 10000.4

63.76% 75.88%

93.44% 100%

94.20% 100%

84.5%

100%

100%

21 7.44

14 15.17

14 18.74

6. Another idea is to apply the neighborhood proximity search procedure to improve people in the population in the MNHPSO optimization algorithm. In other words, in order to further increase the MNHPSO optimization exploration capability, the local search capability of the variable neighborhood search algorithm is used. In creating the neighborhood structure of the algorithm, the main purpose of the algorithm is to reduce the execution cost in addition to managing to increase the total time of the algorithm.

References H. Ahmadi-Nezamabad et al., Multi-objective optimization based robust scheduling of EVs aggregator. Sustain. Cities Soc. 47, 101494 (2019) M.T. Ameli, A. Ameli, EVs as means of energy storage: participation in ancillary services markets, in Energy Storage in Energy Markets, (Academic Press, London, 2021), pp. 235–249 M. Azimi Nasab, M. Zand, M. Eskandari, P. Sanjeevikumar, P. Siano, Optimal planning of electrical appliance of residential units in a smart home network using cloud services. Smart Cities 4, 1173–1195 (2021). https://doi.org/10.3390/smartcities4030063 O. Bamisile et al., Electrification and renewable energy nexus in developing countries; an overarching analysis of hydrogen generation and EVs integrality in renewable energy penetration. Energy Convers. Manag. 236, 114023 (2021) F. Baumgarte, M. Kaiser, R. Keller, Policy support measures for widespread expansion of fast charging infrastructure for EVs. Energy Policy 156, 112372 (2021) N.P. Bayendang, M.T. Kahn, V. Balyan, Power converters and EMS for fuel cells CCHP applications: a structural and extended review. ASTES J. 6(3), 54–83 (2021) N. Chakraborty, A. Mondal, S. Mondal, Intelligent charge scheduling and eco-routing mechanism for EVs: a multi-objective heuristic approach. Sustain. Cities Soc. 69, 102820 (2021)

390

Z. Zand et al.

J. Chen, L. Ramanathan, M. Alazab, Holistic big data integrated artificial intelligent modeling to improve privacy and security in data management of smart cities. Microprocess. Microsyst. 81, 103722 (2021a) J. Chen, S. Huang, L. Shahabi, Economic and environmental operation of power systems including combined cooling, heating, power and energy storage resources using developed multi-objective grey wolf algorithm. Appl. Energy 298, 117257 (2021b) A. Dagar, P. Gupta, V. Niranjan, Microgrid protection: a comprehensive review. Renew. Sust. Energ. Rev. 149, 111401 (2021) M. Gaber et al., Intelligent Energy Management System for an all-electric ship based on adaptive neuro-fuzzy inference system. Energy Rep. 7, 7989–7998 (2021) A. Goli, E.B. Tirkolaee, N.S. Aydin, Fuzzy integrated cell formation and generation scheduling considering automated guided vehicles and human factors. IEEE Trans. Fuzzy Syst. 12, 3686– 3695 (2021) L. Hao et al., Energy management strategy on a parallel mild hybrid electric vehicle based on breadth first search algorithm. Energy Convers. Manag. 243, 114408 (2021) Z. Hou et al., Machine learning and whale optimization algorithm-based design of energy management strategy for plug-in hybrid electric vehicle. IET Intell. Transp. Syst. 8, 20–32 (2021) L. Luo et al., Blockchain-enabled two-way auction mechanism for electricity trading in internet of EVs. IEEE Internet Things J. 9(11), 8105–8118 (2021) Z. Lv, D. Chen, Q. Wang, Diversified technologies in internet of vehicles under intelligent edge computing. IEEE Trans. Intell. Transp. Syst. 22(4), 2048–2059 (2020) P. Makolo, R. Zamora, T.-T. Lie, The role of inertia for grid flexibility under high penetration of variable renewables – a review of challenges and solutions. Renew. Sust. Energ. Rev. 147, 111223 (2021) M.S. R˘aboac˘a, N. Bizon, P. Thounthong, Intelligent charging station in 5G environments: challenges and perspectives. Int. J. Energy Res. 5, 23–26 (2021) M. Rafik et al., in Learning and Predictive Energy Consumption Model Based on LSTM Recursive Neural Networks. 2020 Fourth International Conference on Intelligent Computing in Data Sciences (ICDS) (IEEE, New York, 2020) U. Rehman et al., Network overloading management by exploiting the in-system batteries of EVs. Int. J. Energy Res. 45(4), 5866–5880 (2021) H. Sadr, M.M. Pedram, M. Teshnehlab, Multi-view deep network: a deep model based on learning features from heterogeneous neural networks for sentiment analysis. IEEE Access 8, 86984– 86997 (2020) T.U. Solanke et al., A review of strategic charging–discharging control of grid-connected EVs. J. Energy Storage 28, 101193 (2020) T.U. Solanke et al., Control and management of a multilevel EVs infrastructure integrated with distributed resources: a comprehensive review. Renew. Sust. Energ. Rev. 144, 111020 (2021) H. Tao et al., Shrewd vehicle framework model with a streamlined informed approach for green transportation in smart cities. Environ. Impact Assess. Rev. 87, 106542 (2021) A. Turksoy, A. Teke, A. Alkaya, A comprehensive overview of the dc-dc converter-based battery charge balancing methods in EVs. Renew. Sust. Energ. Rev. 133, 110274 (2020) H.H. Vogt et al., Electric tractor system for family farming: increased autonomy and economic feasibility for an energy transition. J. Energy Storage 40, 102744 (2021) M. Zand, M.A. Nasab, P. Sanjeevikumar, P.K. Maroti, J.B. Holm-Nielsen, Energy management strategy for solid-state transformer-based solar charging station for EVs in smart grids. IET Renew. Power Gener. (2020a). https://doi.org/10.1049/iet-rpg.2020.0399; IET Digital Library, https://digital-library.theiet.org/content/journals/10.1049/iet-rpg.2020.0399 M. Zand, M.A. Nasab, A. Hatami, M. Kargar, H.R. Chamorro, in Using Adaptive Fuzzy Logic for Intelligent Energy Management in Hybrid Vehicles. 2020 28th ICEE (2020b), pp. 1–7. https://doi.org/10.1109/ICEE50131.2020.9260941

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm C. Anuradha, S. Vijayalakshmi, Viswanathan Ganesh, and P. S. Ramapraba

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Permanent Magnet Synchronous Generator – Wind Turbine System Modeling . . . . . . . . 2.1 Simulation Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Modeling of Permanent Magnet Synchronous Generator (PMSG) . . . . . . . . . . . . . . 3 Rotor Side Control Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Fuzzy Logic Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Grid Side Control of WECS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Simulation and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

392 393 393 394 398 398 401 403 407 407

Abstract

Wind Energy Conversion System (WECS) shows nonlinearities in their output power, hence it becomes a challenge for maximizing power and integrating with grid. To ensure maximum power Fuzzy Logic control is proposed at rotor side of Permanent Magnet Synchronous generator with maximum power point algorithm, mechanical speed is controlled. Power quality is maintained well by

C. Anuradha · S. Vijayalakshmi () Department of Electrical and Electronics Engineering, SRM Institute of Science and Technology, Chennai, India e-mail: [email protected]; [email protected] V. Ganesh Department of Electrical Engineering, Chalmers University of Technology, Gothenburg, Sweden e-mail: [email protected] P. S. Ramapraba Panimalar Institute of Technology, Chennai, India © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_180

391

392

C. Anuradha et al.

holding unity power factor grid side using sliding mode control. Hence wind energy is integrated with the grid using these control strategies are verified using MATLAB/Simulink tool. Keywords

Wind energy conversion system · Permanent magnet synchronous generator · Fuzzy logic · Sliding mode control

1

Introduction

In an energy hungry world where countries are running behind fossil fuels for their energy needs, the resources are getting fast depleted. For the very sustenance of our world, it is vital to conserve the very resources that fuel our needs and hence it is important to find alternate resources. Renewable energy resources are those resources, which are classified as inexhaustible. The Wind Energy Conversion Systems (WECS) has always been a top engaging renewable energy technology due to its lower pollution and higher efficiency. However, due to the fact that the energy generated by the WECS depends on the climatic conditions, speed, direction, and humidity of the wind and due to the unreliability of WECS power generation, there may be an increase in the operating costs for the electricity system because the necessities of primary reserves will stand high, which will put a lot of problems to the dependence of electricity supply. Wind energy has huge potential to meet load demand and at the same time is nonpolluting and renewable. Its disadvantage lies in its reliability, which is directly linked to weather condition. To lessen this reserve capacity and boost the wind power generation, a reliable and precise control of wind power is vital before constructing a wind farm. Integration of renewable energy to power grid is very essential nowadays for the conversion of sustainable energy (Ganesh et al. 2021). Also, grid integration of renewables will be constantly enhanced in the future (Arunmozhi et al. 2021). International Renewable Energy Agency portrays that renewable technology is the foremost way to influence zero carbon dioxide (CO2 ) emission. Power electronics converters’ role is significant for energy transition, which would provide efficient electrical energy conversion. Subsequently the development of power electronics technologies use advance semiconductor devices. Wind and solar energy are the most extensively used. Classical proportional integrator is incorporated to limit rotor currents of grid-connected doubly fed induction generator (Kerrouche et al. 2013). Back to back converters, which are widely used along with permanent magnet synchronous generator (Phankong et al. 2013). Maximum power is generated in doubly fed induction generator by means of neuro fuzzy control for different wind fluctuations using hardware in loop (Erramia et al. 2013). Table 1 provides the discussed rule table for the current controller in this chapter.

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm Table 1 Rule table of current controller

2

eix eix NE ZE PE

393 NL NE NE ZE

NS NE ZE ZE

ZE ZE ZE ZE

PS PE ZE ZE

PL PE PE ZE

Permanent Magnet Synchronous Generator – Wind Turbine System Modeling

The mechanical power of wind turbine is given by Pm = 0.5ρπ R2 Vω 3 Cp (λ, β)

(1)

where radius of the turbine is R, wind speed Vm in m/s, air density ρ in Kg/m3 , Cp is the power coefficient, λ is the tip speed ratio, and β is the blade pitch angle. λ=

ωr R Vω

(2)

where ωr is the turbine angular speed. Thereby power from a wind turbine can also written as Popt = Kopt ωr3 opt

(3)

where Kopt =

0.5π Cp max R 5 λ3opt

(4)

Maximum promising power from the turbine must operate at λopt , which is ended by controlling the turbine speed so that it constantly revolves at the optimum speed.

2.1

Simulation Result Analysis

Figure 1 illustrates the simulation diagram of wind turbine. The obtainable simulation’s results of a variable wind speed turbine are presented. The simulation results were obtained for different wind speeds 10, 12, and 15 m/s. Figure 2 illustrates the power coefficient at varying speed. From Figs. 3 and 4 it is seen that the variations of tip speed ratio corresponds to the changes in the speed of rotor for different wind speeds.

394

C. Anuradha et al.

Fig. 1 Simulation circuit – wind turbine

Fig. 2 Power coefficient for various wind speed Fig. 3 Power coefficient at various wind speed

2.2

Modeling of Permanent Magnet Synchronous Generator (PMSG)

The dq0 park transformation is used for the mathematical transformation, which simplifies the analysis of the synchronous machine models. The phase quantities of the PMSG include stator voltages, stator currents, and flux linkages. Applying park’s transformation the three phase quantities abc axes are transformed into two axes reference frame (dq0 transformation). The machine model is synchronously rotating in d-q reference frame with the rotor, d-axis is aligned alongside of the magnetic axis, and q-axis stays

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm

395

Fig. 4 Tip speed ratio – wind turbine

perpendicular to d-axis. Stator variables are converted to rotor reference frame (Rani et al. 2022). It removes the time varying in the voltage equations (Shihabudheen et al. 2019). The voltage equation for the synchronous generator is given by Vds = −Rs ids − ωr λqs + pλds

(5)

Vqs = −Rs iqs + ωr λds + pλqs

(6)

where stator resistance is Rs , ωr is the rotor speed, p number of pole, pairs is p, λds and λqs are the d and q-axis stator flux linkages given by λds = −Ld ids + λm

(7)

λqs = −Lq iqs

(8)

where λm is the rotor magnetic flux and Ld and Lq are the stator dq – axis self-inductances. The electromagnetic torque produced by the permanent magnet synchronous generator is given by (Shihabudheen et al. 2019). 3 Te = − P λds iqs − λqs ids 2

(9)

Substituting λds and λqs in Te 3 P (−Ld ids + λm ) iqs + Lq iqs ids 2 3 = P λm iqs − Ld − Lq iqs ids 2

=

(10) (11)

Figure 5 shows the equivalent circuit obtained from the voltage equation of PMSG. Control system extracts maximum power from wind. By means of power

396

C. Anuradha et al.

Fig. 5 Permanent magnet synchronous generator equivalent circuit

Fig. 6 PMSG- Electromagnetic torque

converter, PMSG is controlled, which enables wind turbine operation variable speed and of decoupling of rotor mechanical speed from the grid electrical system (Figs. 6, 7 and 8). Figure 9 shows the control structure of WECS. Wind turbine is attached straight to the PMSG, which is fed to rectifier. HCS algorithm is used to make generator to operate at maximum power, it generates maximum power from which optimum speed is obtained. Rectifier is fired to maintain DC link voltage constant. Generator side control is done by Vector control, in which speed loop and current loop are

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm

397

Fig. 7 PMSG – dq Voltage

Fig. 8 PMSG – 3ϕ stator voltage

Fig. 9 Control structure of WECS

present. Fuzzy controllers are used in both speed and current loops. Speed is not sensed it is estimated using MRAS estimation method. DC link voltage control loop keep up DC link voltage constant. Cross-coupling effect compensation terms are provided to increase dynamic response of the wind system. Phase locked loop (PLL) is cast to obtain theta for coordinate transformation. Unity power factor is obtained by keeping reference for reactive power be fixed to zero. Reference of d-axis current characterizes system active, and is caused by using SMC. While the inverter works in steady state the DC voltage stands constant at a value set by the reference voltage.

398

3

C. Anuradha et al.

Rotor Side Control Structure

Field-Oriented Control (FOC) (Vijayalakshmi et al. 2021) strategy with fuzzy control is applied toward rotor side converter with the aim to control the speed of generator and likewise to gain the determined electromagnetic torque at lowest current. This objective is reached by keeping the stator currents toward zero; hence the electromagnetic torque is extracted concluded from q component. Current control is accomplished in the rotor reference frame, and coordinate’s transformation is necessary to achieve the reference stator currents. Fuzzy controllers were implemented in both current and speed loops. Speed is estimated with Model Reference Adaptive Estimation method. Optimum speed of rotation is acquired using hill climbing search algorithm (HCS) from maximum power point algorithm using the following relation. ωopt =

Popt Kopt

1 3

(12)

Figure 10 shows the flow chart of obtaining maximum power. From Fig. 11 it is concluded that with HCS algorithm optimum value of power Cp is reached and is maintained at the same value for changes in wind speeds (Fig. 12).

3.1

Fuzzy Logic Controller

A Fuzzy Control (FC) System analyses analog input values with respect to logical variables, which lies between 0 and 1, which operates on discrete values of either 0 or 1, respectively. It is broadly used in a machine control. Fuzzy term relates to the point that the logic involved in the control is referred to “partially true.” Three simple steps are distinctive in every fuzzy logic controllers. It consists of fuzzification of the controller inputs, the accomplishment of the rules of the controller, and defuzzification where output is changed to a crisp value. Two inputs are given to FC and an output Uix * is designed in the controller. Input variables are current error eix (k), and change in current error Δeix (k) is to the controller. eix (k + 1) = ix − ix∗ (k + 1)

(13)

eix (k + 1) = eix (k + 1) − eix (k)

(14)

where ix refers to id and iq currents. Scaling factors G1 , G2 , G3 represents constant gain blocks for obtaining normalized inputs and output of fuzzy logic controller (Figs. 13 and 14). eix N (k + 1) = eix (k + 1) G1

(15)

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm Fig. 10 Flow chart – HCS algorithm

399

400

C. Anuradha et al.

Fig. 11 Power coefficient

Fig. 12 Block diagram – Generator Side

eix N (k + 1) = eix (k + 1) G2

(16)

Uix ∗ (k) = Uix (k − 1) + G3 Uix (k)

(17)

The fuzzy logic control involves three stages: fuzzification, rule execution, and defuzzification. At fuzzification stage, crisp variables eix (k) and Δeix (k) are transformed into fuzzy variables eix * (k) and Δeix * (k) with triangular membership functions presented in the Figs. 11 and 12. Moreover the triangular membership function variation in the command ΔUix and speed control surface is shown in

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm

401

Fig. 13 The membership function of eix

Fig. 14 Membership function – eix

Fig. 13. Universe of discourse is parted to the fuzzy sets, which are denoted as Negative Large(NL), Negative Middle(NM), Negative small(NS), zero(ZE), Positive Small(PS), Positive Middle(PM), and Positive Large(PL). The fuzzification variable-eix are as depicted in Fig. 14. Every fuzzy variable is an associate of the subsets with a degree of membership changing among 0 and 1 (Fig. 15).

4

Grid Side Control of WECS

Wind turbines system are used to distribute the generated power toward electrical grid by means of power converters, which are made commercial nowadays. A classical inverter is attached to the utility grid to a line inductance, which denotes the leakage inductance of the transformer, if any, and the line reactor, which is usually added to the system to decrease the current distortions. The resistance is marginally lesser and has little or no effect on the performance of the system and hence it is neglected. Constant DC link voltage is sustained by means of DC link voltage control loop. For cross coupling effect, compensation terms are added to

402

C. Anuradha et al.

Fig. 15 Membership function – Uix

advance dynamic response of the system. Phase Locked Loop (PLL) is used to acquire the value of theta for coordinate transformation (Ahmed et al. 2020). The reference used for reactive power can be agreed to zero for unity power factor. Reference d-axis current denotes the active power of the system, which is caused by Sliding Mode Controller (SMC) for DC voltage control. SMC generates the d axis current according to the operating conditions. State equation of the grid circuit of the inverter is changed from abc stationary to d-q synchronous and the following equations are gained. At rotating d-q reference frame, when the grid voltage space → vector − u is oriented on d-axis, the voltage across the inductor Lf is specified by Nada Zine Laabidine et al. 2021 as follows: did−f 1 = Vd − Rf id−f + ω Lf i q−f − V dt Lf

(18)

diq−f 1 = Vq − Rf iq−f − ω Lf i d−f dt Lf

(19)

where Rf and Lf are the filter resistance and inductance, respectively; Vd and Vq are d-q-axis voltage components of the inverter, respectively. id-f , iq-f are d- q- axis currents, respectively. Therefore reactive power and active power can be specified as (20) and (21) Q=

3 V iq−f 2

(20)

P =

3 V id−f 2

(21)

DC-side equation are represented as (22)

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm

403

2 1 dUdc C = Pg − P 2 dt

(22)

where Pg is PMSG stator output power. The d-axis reference current is developed with DC-link voltage controller, to regulate converter output power. The fast dynamics are related with the line current control in the inner loop where the SM control is implemented, which tracks the line current control, while in the outer loop slow dynamics is linked with the DC voltage control. Additionally, the PI regulator is engaged so as to cause the reference source current idr-f and to regulate the DC voltage. Likewise, the reference signal of q-axis current iqr-f is formed by the reactive power Qr . Let us present the subsequent sliding surfaces Sd−f = idr−f − id−f

Sq−f = iqr−f − iq−f

where idr-f , iqr-f are the chosen values of d and q axis current, respectively. •

•

•

S d−f = i dr−f − i d−f

•

•

•

S q−f = i qr−f − i q−f

As a result, the controls voltage of d and q axis are defined Vdr−f = Lf

didr−f + Rf id−f − Lf ω iq−f + V + kd−f sgn Sd−f dt Vqr−f = Rf iq−f + Lf ω id−f + kq−f sgn Sq−f

(23) (24)

To conclude, PWM is utilized in order to get the control signal. The structure of the DC-link voltage and current controllers of grid-side converter are presented in Fig. 16.

5

Simulation and Results

MATLAB/Simulink tool is used for attaining the results. Overall Simulink circuit is Illustrated in Fig. 17. Figure 18 shows the variation of wind speed of 10 m/s from 0 to 1 s, 12 m/s from 1 to 2 m/s, and from 2 to 3 s wind speed of 15 m/s. Simulation results for the above varying wind speed is shown below. Comparison of optimum speed (indicated with blue line) that is obtained from MPPT, estimated speed (indicated with red line), which is estimated by MRAS speed estimation, and actual speed as given in the Fig. 19 shows that speed is estimated accurately. It is taking a settling time of 0.25 sec for wind speed of 10 m/s, 0.2 s for wind speed of 12 m/s, and 0.3 s for wind speed of 15 m/s. Fuzzy speed

404

Fig. 16 Block diagram – grid side

Fig. 17 Matlab circuit implementation

Fig. 18 Wind speed

C. Anuradha et al.

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm

405

Fig. 19 MPPT with MRAS speed optimization

Fig. 20 Actual speed of PMSG

Fig. 21 PMSG output voltage

controller is making PMSG to operate at optimum speed so that maximum power is generated. Figure 20 displays the actual speed (rad/sec) at which PMSG rotates when the wind speed is 10 m/s, 12 m/s, and 15 m/s, respectively. Figure 21 displays the PMSG output voltage for various wind speed of 10 m/s, 12 m/s, and 15 m/s. Also the expanded wave form is shown in the figure. Voltage varies according to the wind speed.

406

C. Anuradha et al.

Fig. 22 Electromagnetic torque

Fig. 23 DC link voltage

Fig. 24 Grid voltage

Electromagnetic torque for various wind speeds are shown in Fig. 22. As machine is operated in generation mode, torque is negative. Using SM controller, DC link voltage remains constant throughout change in the wind speed as shown in Fig. 23. Grid voltage and current waveforms generated using SM controller are shown in the Figs. 24 and 25. Power factor of the grid is unity as in Fig. 26.

Grid Integration of Wind Energy Using Fuzzy Logic Algorithm

407

Fig. 25 Grid current

Fig. 26 Power factor

6

Conclusion

The control strategies Fuzzy at generator side and SM at grid side of the variable speed wind system established with PMSG ensures a better performance. It is observed from simulation results that estimated speed is at improved accuracy with optimum speed in MRAS method. Results conclude that Field Oriented Control for variable wind speed in rotor side converter with Fuzzy controller gave improved performance since DC link voltage is maintained constant. Power factor at grid side is maintained unity by implementing Sliding Mode Controller.

References S.D. Ahmed, F.S.M. Al-Ismail, M. Shafiullah, F.A. Al-Sulaiman, I.M. El-Amin, Grid integration challenges of wind energy: A review. IEEE Access 8, 10857–10878 (2020) M. Arunmozhi, S. Senthilmurugan, V. Ganesh, Design and operation strategies for grid connected smart home, in Handbook of Smart Energy Systems, ed. by M. Fathi, E. Zio, P. M. Pardalos, (Springer, Cham, 2021). https://doi.org/10.1007/978-3-030-72322-4-78-1

408

C. Anuradha et al.

Y. Erramia, M. Ouassaid, M. Maaroufia, Control of a PMSG based wind energy generation system for power maximization and grid fault conditions. Energy Procedia 42, 220–229 (2013) V. Ganesh, S. Senthilmurugan, R. Ananthanarayanan, S.S. Srinivasan, N.R.S. Lakshanasri, Integration strategies of renewable energy sources in a conventional community, in Handbook of Smart Energy Systems, ed. by M. Fathi, E. Zio, P. M. Pardalos, (Springer, Cham, 2021). https://doi.org/10.1007/978-3-030-72322-4-120-1 K. Kerrouche, A. Mezouar, K. Belgacem, Decoupled control of doubly fed induction generator by vector control for wind energy conversion system. Energy Procedia 42, 239–248 (2013) N.Z. Laabidine, A. Errarhout, C. El Bakkali, K. Mohammed, B. Bossoufi, Sliding mode control design of wind power generation system based on permanent magnet synchronous generator. Int. J. Power Electron. Drive Syst. 12(1), 393–403 (2021) N. Phankong, S. Manmai, K. Bhumkittipich, P. Nakawiwat, Modeling of grid-connected with permanent magnet synchronous generator (PMSG) using voltage vector control. Energy Procedia 34, 262–272 (2013) P. Rani, V.P. Arora, N.K. Sharma, Improved dynamic performance of permanent magnet synchronous generator based grid connected wind energy system. Energy Sources Part A Recover. Util. Environ. Eff., 1–20 (2022). https://doi.org/10.1080/15567036.2021.2022814 K.V. Shihabudheen, G.N. Pillai, S. Krishnama Raju, Neuro-fuzzy control of DFIG wind energy system with distribution network. Elect. Power Compon. Syst. 46(13), 1–16 (2019) S. Vijayalakshmi, C. Anuradha, V. Ganapathy, V. Padmajothi, Direct driven wind energy conversion system connected to load using variable frequency transformer. Int. J. Electr. Eng. Educ. 58(2), 488–500 (2021)

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy Systems and Their Supply Chains Natasha J. Chrisandina, Shivam Vedant, Mahmoud M. El-Halwagi, Efstratios N. Pistikopoulos, and Eleftherios Iakovou

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Supply Chain Resilience: Definitions and Tenets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Performance and Structural Metrics for Resilience Assessment . . . . . . . . . . . . . . . . . . . . 3.1 Performance Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Structural Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 A Framework for Next-Generation Data-Driven Resilient Energy Supply Chains . . . . . . 5 Conclusions and Future Research Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

410 411 413 414 419 421 424 425 425

N. J. Chrisandina Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA e-mail: [email protected] S. Vedant Energy Institute, Texas A&M University, College Station, TX, USA Department of Multidisciplinary Engineering, Texas A&M University, College Station, TX, USA e-mail: [email protected] M. M. El-Halwagi Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Energy Institute, Texas A&M University, College Station, TX, USA TEES Gas and Fuels Research Center, Texas A&M University, College Station, TX, USA e-mail: [email protected]

© Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_182

409

410

N. J. Chrisandina et al.

Abstract

In order to meet the growing global energy demand in the new era of increased volatility, uncertainty, complexity, and ambiguity (VUCA), next-generation energy systems need to be designed with considerations for increased resilience along with cost efficiency. This need is further demonstrated by an increase in recent publications highlighting the advantages of resilience-aware design and management of energy supply chains. In this chapter, we first present the evolution of resilience, the key motivators, and the antecedents to supply chain resilience. We then provide a taxonomy of performance and structural metrics for quantifying the resilience of energy systems. Building on that, we propose a conceptual framework for next-generation data-driven costcompetitive resilience specifically for energy systems by integrating multiscale modeling approaches. Finally, future research directions for the continued enhancement of the proposed framework towards next-generation energy supply chains are discussed. Keywords

Supply chain resilience · Resilience metrics · Multiscale modeling · Data-driven approach · Energy systems

1

Introduction

Societal reliance on energy for day-to-day functions has led to the development of complex infrastructures to produce, deliver, and use energy as efficiently as possible; these infrastructures are collectively referred to as energy systems (Edenhofer 2015). The three components of energy systems are generation, delivery (which involves transmission between energy generators and distributor, and distribution

E. N. Pistikopoulos Artie McFerrin Department of Chemical Engineering, Texas A&M University, College Station, TX, USA Energy Institute, Texas A&M University, College Station, TX, USA e-mail: [email protected] E. Iakovou () Energy Institute, Texas A&M University, College Station, TX, USA Department of Engineering Technology and Industrial Distribution, Texas A&M University, College Station, TX, USA J. Mike Walker ’66 Department of Mechanical Engineering, Texas A&M University, College Station, TX, USA Mosbacher Institute of Trade, Economics and Public Policy, Bush School of Government and Public Service, Texas A&M University, College Station, TX, USA e-mail: [email protected]

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

411

between energy distributors and direct customers), and utilization (Grigsby 2007). Energy generators transform energy sources such as fossil fuels, wind, or solar power into usable forms of energy that can be transported. These forms of energy, which could include electricity, coal, or oil, can be transmitted to distribution centers or secondary facilities that process the energy further. From distribution centers, energy is then distributed to end customers. Smart energy systems integrate different sectors such as electricity, heating and cooling, and transportation to support endto-end management of energy flows throughout the value chain (Lund et al. 2017). Reliable energy systems are crucial for business continuity and societal welfare. However, energy systems are exposed to many risks that could threaten their ability to operate effectively. These threats can be broadly categorized into environmental risks and system risks (Jasi¯unas et al. 2021). Environmental risks originate from outside the system, including but not limited to natural events such as weatherrelated disaster, pandemics, or geopolitical disruptions such as tariffs, trade wars, economic sanctions, and international conflicts. System risks, on the other hand, originate from parties within the system itself such as institutions regulating the energy industry, market conditions, physical and digital infrastructures, and human operators working in the system. Designing resilient energy systems requires a deep understanding of the key contributors to resilience as well as the specific properties of energy systems that can be exploited to enhance their performance. This chapter will trace major conceptual and quantitative developments in the analysis and design of resilient energy systems. Section 2 focuses on the evolution of resilience as a concept, key motivators, and antecedents to resilience to set the groundwork for a quantitative definition of resilience. Section 3 presents the various resilience indicators and metrics that have been proposed for energy systems. Section 4 discusses a conceptual framework for data-driven next-gen resilience enhancement strategies and multiscale modeling approaches. Finally, Sect. 5 summarizes the chapter and provides future research directions.

2

Supply Chain Resilience: Definitions and Tenets

Resilience of energy systems and energy supply chains as a field of study is still relatively new, with a small but growing number of publications available. Energy systems are supported by their bespoke supply chains (SCs), and therefore, conceptual frameworks that have been developed within the supply chain resilience literature can be utilized as a starting point to study energy system resilience. This section discusses how resilience has been conceptually defined in the supply chain literature and the key motivators that enable a resilient system. This brief review will form the basis of resilience quantification methods, which will be discussed in the Sect. 3. The foundational definition of resilience comes from Holling in 1973, who defined resilience for an ecological system as its ability to “absorb change and disturbance, and still maintain the same relationships between populations or state variables” (Holling 1973). In the following decades, many authors have proposed

412

N. J. Chrisandina et al.

their own definitions to highlight additional or different capabilities and tailor the concept to their specific fields of study. Iakovou and White (2020), further building upon the existing supply chain risk management, provided a comprehensive definition of supply chain resilience as “The ability of a given supply chain to prepare for and adapt to unexpected events; to quickly adjust to sudden disruptive changes that negatively affect supply chain performance; to continue functioning during a disruption; and to quickly recover to its pre-disruption state or a more desirable state.” Definitions for resilience typically point to specific capabilities a system must possess in the face of a disruption or disturbance. When a resilience metric is defined, it quantifies one or more of these capabilities. Problem arises when the resilience definition misses a critical capability that is needed for a system to respond appropriately to a disruption. A comprehensive understanding of the capabilities required to achieve a resilient system is key to building resilient energy systems. Table 1 provides a taxonomy of proposed definitions for SC resilience. The capabilities to absorb and recover from disruptions are frequently discussed in the literature, whereas the ability to anticipate disruptions is less likely to be mentioned explicitly. Furthermore, definitions tend to emphasize either absorptive or adaptive capability, but not both. An emphasis on absorptive capability points to a static view of resilience as the ability to maintain core functionalities in the face of disruptions. An emphasis on adaptive capability, on the other hand, indicates a dynamic view of resilience as the ability to adjust over time in response to disruptions (Iftikhar et al. 2021). A few definitions, such as the one proposed by Ali et al. (2017), added the capability to learn from past experience to their definition of resilience. Incorporating the ability to learn into how resilience is conceptualized is increasingly common as companies attempt to not only anticipate upcoming disruptions but avoid repeating previous mistakes.

Table 1 Summary of the capabilities of resilience as defined in previous works Publication Holling (1973) Closs and McGarrell (2004) Pettit et al. (2010) Carvalho et al. (2012) Yao and Meurier (2012) National Academy of Sciences (2012) Kim et al. (2015) Tukamuhabwa et al. (2015) Kamalahmadi and Parast (2016) Ali et al. (2017) Shekarian and Mellat Parast (2020)

Resilience capabilities Anticipate Absorb

Adapt

Recover

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

413

The key tenets for supply chain resilience are as follows (Iakovou and White 2020): • Rapid detection, response, and recovery. Also known as flexibility or agility, the ability to quickly detect, respond to, and recover from a disturbance is critical to returning to normal operations in a timely manner (Parast and Shekarian 2019). • End-to-end visibility, data-driven, supply chain control. Open, accurate information flow across the value chain allows for rapid, informed adjustments to be made in response to disruptions (Barratt and Barratt 2011). • Redundancy involves the holding of additional resources in reserve that can be deployed during disruptions to ensure business continuity (Sheffi and Rice 2005). Through diverse sourcing, buffer capacities in manufacturing facilities and across the various SC echelons, or emergency stockpiles at warehouses, consistent service level can be maintained while recovery efforts are conducted in disrupted parts of the system (Mackay et al. 2020). • Collaboration between stakeholders throughout the system is an enabler for the other resilience drivers (Cao et al. 2010). As disruptions can easily propagate across the supply chain (Ripple Effect), there is incentive for stakeholders to collaborate even if limited direct risks are perceived (Dolgui et al. 2018). Process integration, the signing of contracts or joint economic incentives, and government policies are some mechanisms that can foster deeper collaborations within systems (Jain et al. 2017). • Effective demand planning processes supported by strong collaborations and high visibility across the value chain allow stakeholders to meet customer demand in the most efficient way possible (Ashayeri and Lemmes 2006). Many planning decisions, such as production quantities or product delivery schedules, are made prior to the point when customer demand is known (Kilger and Wagner 2008). Thus, demand forecasting needs to shift from solely a marketing department’s responsibility to data-driven technologies employing machine learning (ML) and artificial intelligence (AI).

3

Performance and Structural Metrics for Resilience Assessment

In this section, quantitative indicators for measuring the resilience of an energy system are discussed. There are two general indicators in the literature: performance metrics and structural metrics (Essuman et al. 2020). Performance metrics measure the performance of a system after disruption. Structural metrics, on the other hand, utilize inherent system characteristics to ad hoc assess the supply chain resilience. Some metrics are general and can be applied to any system, while others are specifically tailored to energy systems. This section will discuss both types of metrics as well as their advantages and drawbacks.

414

3.1

N. J. Chrisandina et al.

Performance Metrics

Performance metrics aim to quantify the change in performance over time as a system experiences disruption and recovers from it. This performance function typically reflects a measurable output of the system or infrastructure quality. This function takes a constant value under nominal operation. After a disruption event, the performance function drops until it reaches a minimum value to represent the degradation of the system, after which it starts increasing again as the system recovers. Depending on the recovery efforts implemented, the performance level might plateau to a point that is lower than the previous nominal operation. It could also reach the same point, or even a higher state which is desirable. This profile of the performance function is often called the “resilience triangle” (Bruneau et al. 2003), as illustrated in Fig. 1. Three key performance indicators (KPIs) that are used to describe the behavior of this performance function that can be derived from the resilience triangle are the time required for performance to reach steady-state post-disruption (time-torecover or TTR), the difference between pre-disruption nominal performance and post-disruption performance at a specified time point (recovery level or RL), and the cumulative lost performance during the recovery period (lost performance during recovery or LPR) (Behzadi et al. 2020) (Table 2). Some metrics are entirely based on one of the aforementioned KPIs, while others create aggregate metrics to track two or more of the aspects. Other relevant KPIs include lead time ratio and meantime-to-survival (MTTS).

Fig. 1 The resilience triangle (Bruneau et al. 2003)

Table 2 Taxonomy of resilience performance metrics in recent publications

Publication Bruneau et al. (2003) Reed et al. (2009) Afgan and Veziroglu (2012) Mu et al. (2021) Taghizadeh et al. (2021) Pant et al. (2014) Carvalho et al. (2012) Sawik (2017) Behzadi et al. (2020) Francis and Bekera (2014) Shukla et al. (2011)

TTR

RL

LPR

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

415

Fig. 2 The bathtub curve, also known as the resilience trapezoid

The resilience triangle assumes that the effect of a disruption on the system is instantaneous; that is, the system performance declines from pre-disruption level to its minimum performance instantaneously. In reality, there may be a slow decline of performance before the system reaches its minimum performance. This results in a bathtub curve (as illustrated by Fig. 2), sometimes also regarded as resilience trapezoid, that accounts for a gradual decline in performance (Taleb-Berrouane and Khan 2019). The performance curve models discussed so far implicitly assume that system performance returns to the pre-disruption level after the recovery period. However, recovery efforts may result in post-disruption performance not quite reaching pre-disruption performance levels or, in some cases, even exceeding them, thus providing significant corporate competitive advantage. Figure 3 illustrates the life cycle stages of resilience following a disruption, which is a generalization of the bathtub curve, and captures multiple potential post-disruption behaviors: repurposed stability, original performance, and a fail-safe behavior (see Fig. 3) (El-Halwagi et al. 2020). While performance metrics explicitly capture the response of a system under disruptions, they do not identify the system properties that directly or indirectly contribute to its performance. Furthermore, “good” resilience performance against one set of disruptions does not necessarily guarantee the same level of performance against a different set of disruptions. Bruneau et al. (2003), in the context of resilience assessment during seismic disruption events, provided one of the first quantitative definitions of system resilience, by defining the cumulative loss of resilience (R B ) as R = B

t1

[100 − QB (t)]dt,

(1)

t0

where QB is the performance quality function over the recovery period. Reed et al. (2009) developed two metrics to assess the resilience against natural hazards, with a focus on power distribution systems against hurricanes. The first is fragility, which is the percentage of power line outages per customer zones. The second metric is system resilience:

416

N. J. Chrisandina et al.

Fig. 3 The generalized bathtub curve (El-Halwagi et al. 2020)

t1 R = R

t0

QR (t)dt

t1 − t0

,

(2)

where R R is the resilience of the system, QR the quality function, and (t1 − t0 ) the duration of the recovery efforts. These metrics were utilized as part of a simulation framework to assess the fragility and resilience of the power system after Hurricane Katrina. Afgan and Veziroglu (2012) proposed a resilience index that is composed of several sub-indicators: economic, environmental, technological, and social resilience elements. Each indicator qi has a specific weight coefficient wi capturing its relative importance to the decision-maker. The total resilience coefficient is then a weighted sum of the cumulative lost performance during recovery for every performance indicator (Afgan and Veziroglu 2012): Rj =

i

tf

wi

(100 − qi )dt.

(3)

ti

Mu et al. (2021) assumed a bathtub curve performance curve (see Fig. 2), where the decline in performance is not instantaneous but rather has a noninfinite downward slope. Subsequently, the authors estimated the resilience of a supply chain against multiple shock events h by calculating the area under the

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

417

curve (E[AI Ah ]), weighted by the occurrence rate (v h ) and intensity (q h ) of each shock event. The resilience is then normalized against the total performance (T P ) assuming no disruptions to yield the following indicator: R=

TP −

H

vh

qh

E[AI Ah (q h )]φ h (q h )

TP

.

(4)

Taghizadeh et al. (2021) used the trapezoid rule to approximate the area underneath the performance curve during recovery. The authors define a resilience level for each facility zn , as a ratio of lost performance to steady-state performance normalized by the time taken to recover T : T zn =

Lost performance t=0 (1 − Steady state performance )

T

.

(5)

Pant et al. (2014) proposed two different temporal metrics: the time to total system restoration and the time to full system service resilience. These metrics are stochastic measures accounting for uncertainty in the time measurements. Time to total system restoration measures the total time span from the point when recovery activities commence (ts ), to the time when all recovery activities are completed. The j recovery time for component i after disruption event j (Ui ) is given by TT (ej ) =

j

(6)

Ui .

W

On the other hand, the time to full system resilience captures the total time span from the point when recovery activities commence (ts ), to the time when system service is completely restored (tf ) as per Eq. (7): Tϕ(t0 ) (e ) = j

Ah

max j

Ui ∈Ah

j [Ui ]

.

(7)

Carvalho et al. (2012) introduced the lead time ratio for each supply chain node stakeholder k on time period t, corresponding to the ratio between the promised lead time to the actually realized one. This indicator was then used in a simulation model to select the most resilient SC design. Jkt Actual LT(k, t)j /Jkt × 100, Lead Time Ratio(k, t) = Promised LT(k, t)j

(8)

j −1

where Jkt is the total number of orders delivered from node k directly to their customers during time period t.

418

N. J. Chrisandina et al.

Sawik (2017) utilized the expected service level for every disruption scenario as an objective, formulated as ES =

Ps xts /D,

(9)

s∈S t∈T

where ES is the expected service level, Ps is the probability of disruption scenario s occurring, xts represents the production level in period t under scenario s, and D is the total demand for product. This metric can be employed in sourcing decisions of the network and for scheduling over the planning horizon for each scenario to optimize service level. Behzadi et al. (2020) proposed an integrated metric, the net present value of loss profit (NPV-LP), which is a time weighted function of lost profit. The metric is 1 defined using a discount factor β = 1+r based on interest rate r for a particular resilience strategy m: NP V − LP (m) = f (m)zm +

T

β t−1 LP (m, t); ∀m ∈ θ,

(10)

t=1

where f (m) is the fixed cost of resilience strategy m and LP (m, t) is the cumulative lost profit at time period t under strategy m. Francis and Bekera (2014) devised a resilience metric (ρ) that explicitly accounts for the possibility that post-disruption steady-state performance may not be the same as the pre-disruption one, employing the following ratios: Fr Fd , Fo Fo tδ Sp = ∗ exp[−a(tr − tr∗ )] for tr ≥ tr∗ , tr ρ = Sp

Sp =

tδ otherwise, tr∗

(11) (12) (13)

where Fo is the original stable system performance level; Fd the performance level immediately post-disruption at time td ; Fr∗ the performance level after an initial post-disruption equilibrium state has been achieved at time tr∗ ; and Fr the performance at a new stable level after all recovery efforts have been exhausted at time tr . Additionally, Sp represents a speed recovery factor that measures the speed at which full recovery takes place. tδ is defined as slack time, which is the maximum amount of time post-disaster that is acceptable before recovery efforts begin. The speed recovery factor takes into account both the speed at which initial recovery efforts are undertaken compared to the allowable slack time along with a delay factor exp[−a(tr − tr∗ )] for the time taken to reach the final post-disruption state.

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

419

Shukla et al. (2011) proposed an optimization framework to address facility and link failure issues on network performance (focusing jointly on efficiency and robustness). Furthermore, the authors present metrics for the efficiency and robustness based on operational cost (OC) (including infrastructure, material handling, and transportation costs) and expected disruption cost (EDC), respectively: OCmax − OC , OCmax − OCmin

(14)

EDCmax − EDC , EDCmax − EDCmin

(15)

ηE =

ηR =

where OCmax is the OC of the most robust supply chain and OCmin is the OC of the most efficient supply chain and EDCmin is the EDC of the most robust supply chain and EDCmax is the EDC of the most efficient supply chain. Subsequently, the metrics are included in a multi-objective optimization problem (see Eq. (22)) to maximize survival based on the weighted average of the two indicators.

3.2

Structural Metrics

Structural metrics aim to ascertain the preparedness of a system to handle unknown disruptive events through measuring properties that contribute to the resilience drivers (including flexibility, data-driven control, redundancy, collaboration, and demand planning) (Essuman et al. 2020). These types of metrics have not been widely utilized for energy systems, but have been applied to the traditional supply chain literature and practice. Therefore, herein we first present a brief survey into structural metrics that have been proposed in the general supply chain resilience literature. These metrics could be tailored to energy systems to elucidate structural elements that can contribute to their resilience. Structural metrics can provide guidance on which design elements to incorporate in the system. However, it can still be challenging to identify the specific properties that are the strongest predictors of SC resilience. It is also noteworthy that structural metrics become increasingly significant in the context of “black swan” (low probability and high impact) events where the time of occurrence and the magnitude of disruptions are unknown (“unknown-unknown” events). Graph theory has been foundational for advancing the understanding of structural metrics for supply chain resilience; therefore, a brief explanation of relevant terms is given here. A supply chain can be represented as a graph consisting of nodes (facilities) and arcs that connect nodes (transportation modes) (Kim et al. 2015). Formally, a graph G = (N, A) consists of a set of nodes N and a set of arcs A. The total number of nodes in a graph is |N|, and the total number of arcs is |A|. Each arc in set A connects two nodes from set N, called endpoints. To represent the direction of flow of goods in a supply chain, a directed graph (or digraph) is used where all

420

N. J. Chrisandina et al.

arcs have an origin node (a tail) and a destination node (a head). Supplier nodes that do not receive any product are called source nodes, and demand nodes that do not deliver any product are called sink nodes. Falasca et al. (2008) developed a simulation-based framework to assess supply chain resilience. The authors defined three determinants of resilience (supply chain density, complexity, and node criticality) and correlated their effect on system resilience. These determinants are formulated as follows: Supply chain density =

|N| , and average inter-node distance

Supply chain complexity =|N | + |A|forward flow + |A|backward flow + |A|within-tier flow .

(16)

(17)

Node criticality for an individual node n is defined as a sum of the number of non-redundant inbound and outbound arcs to that node weighted by the relative importance of the node. The relative importance score is the estimated based on whether the node is responsible for critical components within the supply chain or large amounts of throughput. Tan et al. (2019) developed a graph-based model of a multi-stage supply chain network and identified which plants are critical to the network. The set of nonredundant plants Pnr is defined as Pnr =

P (m) if |P (m)| = 1, ∅ otherwise,

(18)

m∈MH

where MH is the set of all materials that flow in the supply chain and P (m) is the set of all plants that produce material m. If only one plant produces material m, then that plant is non-redundant. The set of critical plants Pcrit represents the plants that are shared by all possible paths towards the end product G within a supply chain G: Pcrit =

P.

(19)

G∈G

The authors then introduced network redundancy nscn , which represents the number of alternative paths towards the end product within a supply chain G as nscn = |G|.

(20)

The proposed model is capable of representing structural redundancy across multiple stages of the supply chain network and can be used to evaluate the effectiveness of alternative strategies for designing resilient supply chains.

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

421

Ahmadian et al. (2020) argued that since network resilience is constrained by the resilience of its weakest component, the characteristics of the latter can serve as a proxy for the resilience of the overall network. This is one of the few research efforts that explicitly link network-level resilience to node-level and arc-level resilience. The component resilience index (CRI), applicable to both nodes and arcs, is defined as follows: CRI = 1 −

total demand loss due to component disruption total demand in the network system during recovery period

(21)

Venkatasubramanian et al. (2004), using principles of evolutionary adaptation, presented a theory investigating the dependence of the structure and organization of a network on its survival (or objectives). The authors further propose that in order to maximize the overall survival fitness, composed of short- and long-term goals, a complex system needs to adapt and reconfigure its network structure. In order to balance the trade-offs between the two aspects, an optimization problem can be formulated where the objective function is given by max G = αηE + (1 − α)ηR ,

(22)

where the α parameter weighs the environmental against the selection pressure on the network. If α = 0, the network favors long-term survival (robustness); conversely, when α = 0, the network prefers short-term survival (efficiency). Thus, by varying the value of α between 0 and 1, different selection pressures can be enforced on the network, and resulting configurations can be further analyzed. The constraints of the optimization problem include network structure constraints (to enable or restrict warehouse-customer matching), material balance constraints, demand fulfillment and capacity constraints, and single sourcing constraints (to ensure that every customer zone is being served by exactly 1 warehouse).

4

A Framework for Next-Generation Data-Driven Resilient Energy Supply Chains

A resilient and agile energy SC comes at a higher cost as it embraces a diversified portfolio of suppliers, spare manufacturing capacity, additional inventory, and surplus cash. Consequently, an emerging challenge in the domain is the development and implementation of resilience principles in a cost-competitive manner in order to ensure economic viability in the global market. One approach for cost-competitive resilience is through the use of a hybrid network that blends distributed supply chains with centralized supply chains. A hybrid network takes advantage of economies of scale and risk pooling that is possible with centralized supply chains while also maintaining the flexibility of distributed supply chains (Iakovou and White 2020).

422

N. J. Chrisandina et al.

In this contextual framework, specifically as per the energy sector, distributed energy systems (DESs) have been increasingly embedded into the portfolio of nextgeneration sustainable and resilient energy sources. Consisting of a set of smallscale modular processing plants located in close proximity to energy sources and/or consumers, a DES utilizes reconfigurable and repurposable modular units within each plant (Allen et al. 2018). The use of modular units provides DESs with the flexibility to bypass disrupted plants through rapid reconfiguration, allowing them to quickly adapt to changing circumstances. The small scale of the plants also implies that the effects of disruption are localized to the region, decreasing the propagation of disruptive events throughout the entire system (El-Halwagi et al. 2020). This advantage is illustrated in the use of microgrids as part of the power system, which have been shown to cope with major power disruption events due to their ability to effectively utilize renewables and other localized sources of energy (Hussain et al. 2019). Energy systems are favorably poised to benefit not only from supply chain reconfiguration strategies (adaptability and flexibility) but also from the various chemical and physical synergies that exist at the process/plant level. At the process level, targeted risk mitigation techniques can be implemented on the individual plant and transportation mode which could result in outsized gains towards costcompetitive energy supply chain resilience. To this effect, multiscale techniques are required to accurately capture spatial and temporal differences across many scales of operation (Floudas et al. 2016). A multiscale approach should engage engineering principles to unveil the impact of individual plants, unit operations, and molecular structures on overall supply chain resilience. The application of multiscale methodologies to the energy sector has been a research topic of interest in the literature for some time, especially to support the energy transition towards renewable energy sources (readers are referred to the “Hydrogen-Based Dense Energy Carriers in Energy Transition Solutions” chapter of this handbook for a discussion of a multiscale approach for the hydrogen energy system). Within the energy supply chain, four scales of operation need be considered (Chrisandina et al. 2022): • Molecular scale. To build in resilient reactions, strategies from the field of green chemistry can be implemented. Such strategies include the use of reliably sourced catalysts to reduce energy requirements, designing reaction pathways to occur at ambient conditions and substituting hazardous solvents with benign alternatives. • Unit operation scale. Inherently safer design principles can help build in redundancies within the unit operation level. Strategies such as installing multiple backup units and power sources for units, using components with high reliability, and substituting large units with intensified alternatives to reduce energy requirements can be implemented at this scale. • Process scale. On the process scale, strategies that exploit the modularity of energy systems are particularly valuable to increase the flexibility of the system. Utilizing modular process units allows an individual plant to be reconfigured and repurposed to meet fluctuating market demands or handle variability in supply.

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

423

• Supply chain scale. Strategies from the supply chain resilience literature can provide the flexibility and redundancy needed to achieve resilient energy supply chains, such as location and distribution of manufacturing facilities, diversification of suppliers, and hybrid (centralized-distributed) supply chains. To ensure business continuity and sustained competitiveness of next-generation energy SCs, risk management should be considered across the scales of operation of the distributed network. Moreover, access and employment of real-time data can greatly enhance risk mitigation strategies (such as identifying precursors) and allow the full realization of adaptability and flexibility inherent to a DES. The catalytic role of data-driven digital technologies in enhancing energy supply chain resilience has been widely underlined in the literature (Arghandeh et al. 2021). Through leveraging state-of-the-art data collection, information processing, and digital ecosystems, the decision-making process across the stages of disruption (pre, during, and post) can be significantly improved (Bechtsis et al. 2021). This will lead to a dynamically resilient data-driven energy supply chain which is capable of rapidly detecting, responding to, and recovering from disruptions by adjusting manufacturing capacity, as needed (Iakovou and White 2020). An integrative data-driven framework to support the design and management of next-generation resilient energy supply chains should further be tailored to include the following research thrust areas: • Energy supply chain visibility and transparency. End-to-end value chain visibility in real time is imperative for identifying all the stakeholders from “suppliers’ suppliers to customer’s customers” for every product/material involved in the process. Additionally, it plays a crucial role in energy supply chain mapping and developing a risk index for critical parts and components. Currently, such capabilities are significantly impeded by higher associated costs, contract privacy issues, and lack of integrated IT platforms and data aggregation frameworks to enable collaboration between energy supply chain actors. • Re-evaluation of supply base. This is hindered not only by the complex sourcing on a global scale due to the incorporation of renewable energy sources to support the energy transition and the persistent pressures for lower total costs and higher return-on-investment (with lengthy lead times), but also by severe supplier risks at each stage (data inaccuracies, supply shortages, labor-related issues, human errors, etc.). Real-time data on demand levels gives energy SCs the ability to deploy a mix of energy sources to fulfill customer demands. • Demand planning. New strategies focusing on more effective base demand forecasting (via customer collaboration) and the ability to react quickly to disturbances and information inaccuracies are in dire need. These should capture the identification of demand-risk factors and demand fluctuations (including the bullwhip effect). Advanced forecasting models based on machine learning (ML) and artificial intelligence (AI) provide more reliable predictive ability compared to traditional demand forecasting methods to plan for energy demand surges during disruptive events (Ivanov and Dolgui 2021).

424

N. J. Chrisandina et al.

• Continuous learning and improvement. After disruptions occur and recovery efforts have been completed, a mechanism to re-evaluate supply chain design and possibly re-engineer the network to protect against similar types of disruption is needed.

5

Conclusions and Future Research Directions

In this chapter, we have provided a taxonomy of resilience metrics and quantitative methods employed in the traditional supply chain resilience literature. We then argue for the need for tailored data-driven quantitative integrative approaches which leverage the multiscale nature of energy systems to increase their supply chain resilience. We present a first effort towards a conceptual framework where the advantages of a hybrid supply chain structure (centralized-distributed) are integrated with the multiscale nature of energy systems through the use of datadriven technologies (such as machine learning and AI) to enable real-time risk management, end-to-end visibility, supply base re-evaluation, and demand planning. The continued development of this framework will require its embellishment by building additional research capabilities to address the following issues: • Utilization of digital twins for energy supply chain risk management. A digital twin is a virtual counterpart to a physical system that exchanges information with its physical counterpart (Ivanov and Dolgui 2021). It can be utilized to perform analyses on the system, such as stress-testing or comparison of different potential improvements. The use of digital twins within energy supply chains has the potential to enhance predictive and reactive capabilities through utilization of supply chain visualization, historical disruption data analysis, and real-time disruption data. • Development of long-term, high-capacity energy storage solutions. A major issue with renewable energy sources (wind, solar, etc.) is their spatial and temporal intermittency; that is, the sources are only available at select times throughout the day or year and in specific locations. To maintain consistent service level and distribute energy equitably across populations, long-term and high-capacity energy storage solutions are in dire need. Deployment of energy storage units will support the resilience of the energy supply chain as a whole through providing much-needed redundancy in the system. • Intersection of energy with food and water systems. The interdependence of three critical systems – energy, food, and water – has gained increased attention in the research literature. The complexities of this food-energy-water nexus (FEWN) expose the system to more vulnerabilities and may lead to more dire consequences should a disruption occur. Therefore, the resilience of energy supply chains need to be examined in tandem with the food and water systems (readers are referred to the “The Food-Energy-Water Nexus in Sustainable Energy Systems Solutions” chapter of this handbook for a thorough discussion on the food-energy-water nexus).

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

425

We envision that data-driven integrative frameworks tailored to energy systems, as proposed herein, would add significant value towards the energy transition and the design of management of next-generation cost-competitive and sustainable energy supply chains under the new “normal.”

6

Cross-References

Hydrogen-based Dense Energy Carriers in Energy Transition Solutions The Food-Energy-Water Nexus in Sustainable Energy Systems Solutions

References N. Afgan, A. Veziroglu, Sustainable resilience of hydrogen energy system. Int. J. Hydrog. Energy 37(7), 5461–5467 (2012) N. Ahmadian, G.J. Lim, J. Cho, S. Bora, A quantitative approach for assessment and improvement of network resilience. Reliab. Eng. Syst. Saf. 200, 106977 (2020) A. Ali, A. Mahfouz, A. Arisha, Analysing supply chain resilience: integrating the constructs in a concept mapping framework via a systematic literature review. Supply Chain Manag. Int. J. 22(1), 16–39 (2017) R.C. Allen, D. Allaire, M.M. El-Halwagi, Capacity planning for modular and transportable infrastructure for shale gas production and processing. Ind. Eng. Chem. Res. 58(15), 5887–5897 (2018) R. Arghandeh, B. Uzunoglo, S. D’arco, E.E. Ozguven, Guest editorial: data-driven reliable and resilient energy system against disasters. IEEE Trans. Ind. Inform. 18(3), 2075–2077 (2021) J. Ashayeri, L. Lemmes, Economic value added of supply chain demand planning: a system dynamics simulation. Robot. Comput. Integr. Manuf. 22(5–6), 550–556 (2006) M. Barratt, R. Barratt, Exploring internal and external supply chain linkages: evidence from the field. J. Oper. Manag. 29(5), 514–528 (2011) D. Bechtsis, N. Tsolakis, E. Iakovou, D. Vlachos, Data-driven secure, resilient and sustainable supply chains: gaps, opportunities, and a new generalised data sharing and data monetisation framework. Int. J. Prod. Res. 60(14), 4397–4417 (2021) G. Behzadi, M.J. O’Sullivan, T.L. Olsen, On metrics for supply chain resilience. Eur. J. Oper. Res. 287(1), 145–158 (2020) M. Bruneau, S.E. Chang, R.T. Eguchi, G.C. Lee, T.D. O’Rourke, A.M. Reinhorn, M. Shinozuka, K. Tierney, W.A. Wallace, D. Von Winterfeldt, A framework to quantitatively assess and enhance the seismic resilience of communities. Earthq. Spectra 19(4), 733–752 (2003) M. Cao, M.A. Vonderembse, Q. Zhang, T. Ragu-Nathan, Supply chain collaboration: conceptualisation and instrument development. Int. J. Prod. Res. 48(22), 6613–6635 (2010) H. Carvalho, A.P. Barroso, V.H. Machado, S. Azevedo, V. Cruz-Machado, Supply chain redesign for resilience using simulation. Comput. Ind. Eng. 62(1), 329–341 (2012) N.J. Chrisandina, E. Iakovou, E.N. Pistikopoulos, M.M. El-Halwagi, Distributed Manufacturing for Enhanced Resilience and Sustainability of Energy Systems: Perspectives and Challenges. Manuscript in preparation (2022) D.J. Closs, E.F. McGarrell, Enhancing Security Throughout the Supply Chain, IBM Center for the Business of Government, Washington, D.C. (2004) N.R. Council et al., Disaster Resilience: A National Imperative (The National Academies Press, Washington, DC, 2012) A. Dolgui, D. Ivanov, B. Sokolov, Ripple effect in the supply chain: an analysis and recent literature. Int. J. Prod. Res. 56(1–2), 414–430 (2018)

426

N. J. Chrisandina et al.

O. Edenhofer, Climate Change 2014: Mitigation of Climate Change, vol. 3 (Cambridge University Press, Cambridge, 2015) M.M. El-Halwagi, D. Sengupta, E.N. Pistikopoulos, J. Sammons, F. Eljack, M.-K. Kazi, Disasterresilient design of manufacturing facilities through process integration: principal strategies, perspectives, and research challenges. Front. Sustain. 1, 8 (2020) D. Essuman, N. Boso, J. Annan, Operational resilience, disruption, and efficiency: conceptual and empirical analyses. Int. J. Prod. Econ. 229, 107762 (2020) M. Falasca, C.W. Zobel, D. Cook, A decision support framework to assess supply chain resilience, in Proceedings of the 5th International ISCRAM Conference (2008), pp. 596–605 C.A. Floudas, A.M. Niziolek, O. Onel, L.R. Matthews, Multi-scale Systems Engineering for Energy and the Environment: Challenges and Opportunities, AIChE Journal. (Wiley Online Library), 62(3), 602–623 (2016) R. Francis, B. Bekera, A metric and frameworks for resilience analysis of engineered and infrastructure systems. Reliab. Eng. Syst. Saf. 121, 90–103 (2014) L.L. Grigsby, Electric Power Generation, Transmission, and Distribution (CRC Press, Boca Raton, 2007) C.S. Holling, Resilience and stability of ecological systems. Ann. Rev. Ecol. Syst. 4(1), 1–23 (1973) A. Hussain, V.-H. Bui, H.-M. Kim, Microgrids as a resilience resource and strategies used by microgrids for enhancing resilience. Appl. Energy 240, 56–72 (2019) E. Iakovou, C. White III, How to build more secure, resilient, next-gen U.S. supply chains (2020). https://www.brookings.edu/techstream/how-to-build-more-secure-resilient-next-genu-s-supply-chains/ A. Iftikhar, L. Purvis, I. Giannoccaro, A meta-analytical review of antecedents and outcomes of firm resilience. J. Bus. Res. 135, 408–425 (2021) D. Ivanov, A. Dolgui, A digital supply chain twin for managing the disruption risks and resilience in the era of industry 4.0. Prod. Plan. Control 32(9), 775–788 (2021) V. Jain, S. Kumar, U. Soni, C. Chandra, Supply chain resilience: model development and empirical analysis. Int. J. Prod. Res. 55(22), 6779–6800 (2017) J. Jasi¯unas, P.D. Lund, J. Mikkola, Energy system resilience–a review. Renew. Sustain. Energy Rev. 150, 111476 (2021) M. Kamalahmadi, M.M. Parast, A review of the literature on the principles of enterprise and supply chain resilience: major findings and directions for future research. Int. J. Prod. Econ. 171, 116– 133 (2016) C. Kilger, M. Wagner, Demand planning, in Supply Chain Management and Advanced Planning (Springer, Berlin, 2008), pp. 133–160 Y. Kim, Y.-S. Chen, K. Linderman, Supply network disruption and resilience: a network structural perspective. J. Oper. Manag. 33, 43–59 (2015) H. Lund, P.A. Østergaard, D. Connolly, B.V. Mathiesen, Smart energy and smart energy systems. Energy 137, 556–565 (2017) J. Mackay, A. Munoz, M. Pepper, Conceptualising redundancy and flexibility towards supply chain robustness and resilience. J. Risk Res. 23(12), 1541–1561 (2020) W. Mu, E. van Asselt, H. Van der Fels-Klerx, Towards a resilient food supply chain in the context of food safety. Food Control 125, 107953 (2021) R. Pant, K. Barker, J.E. Ramirez-Marquez, C.M. Rocco, Stochastic measures of resilience and their application to container terminals. Comput. Ind. Eng. 70, 183–194 (2014) M.M. Parast, M. Shekarian, The impact of supply chain disruptions on organizational performance: a literature review. Rev. Supply Chain Risk (Springer, Cham, 2019), 7, 367–389 (2019) T.J. Pettit, J. Fiksel, K.L. Croxton, Ensuring supply chain resilience: development of a conceptual framework. J. Bus. Logist. 31(1), 1–21 (2010) D.A. Reed, K.C. Kapur, R.D. Christie, Methodology for assessing the resilience of networked infrastructure. IEEE Syst. J. 3(2), 174–180 (2009) T. Sawik, A portfolio approach to supply chain disruption management. Int. J. Prod. Res. 55(7), 1970–1991 (2017)

Quantitative Methods for Data-Driven Next-Generation Resilience of Energy. . .

427

Y. Sheffi, J.B. Rice Jr., A supply chain view of the resilient enterprise. MIT Sloan Manag. Rev. 47(1), 41 (2005) M. Shekarian, M. Mellat Parast, An integrative approach to supply chain disruption risk and resilience management: a literature review. Int. J. Log. Res. Appl. 1–29 (2020) A. Shukla, V. Agarwal Lalit, V. Venkatasubramanian, Optimizing efficiency-robustness trade-offs in supply chain design under uncertainty due to disruptions. Int. J. Phys. Distrib. Logist. Manag. 41(6), 623–647 (2011) E. Taghizadeh, S. Venkatachalam, R.B. Chinnam, Impact of deep-tier visibility on effective resilience assessment of supply networks. Int. J. Prod. Econ. 241, 108254 (2021) M. Taleb-Berrouane, F. Khan, Dynamic resilience modelling of process systems. Chem. Eng. 77, 313–318 (2019) W.J. Tan, A.N. Zhang, W. Cai, A graph-based model to measure structural redundancy for supply chain resilience. Int. J. Prod. Res. 57(20), 6385–6404 (2019) B.R. Tukamuhabwa, M. Stevenson, J. Busby, M. Zorzini, Supply chain resilience: definition, review and theoretical foundations for further study. Int. J. Prod. Res. 53(18), 5592–5623 (2015) V. Venkatasubramanian, S. Katare, P.R. Patkar, F.-P. Mu, Spontaneous emergence of complex optimal networks through evolutionary adaptation. Comput. Chem. Eng. 28(9), 1789–1798 (2004) Y. Yao, B. Meurier, Understanding the supply chain resilience: a dynamic capabilities approach (2012)

Hybrid Attack Modeling for Critical Energy Infrastructure Protection Maryna Zharikova, Volodymyr Sherstjuk, and Stefan Pickl

Contents 1 2 3 4 5

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Recent Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Energy Infrastructure Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hybrid Operation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Spatial Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Time Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Goal Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Hybrid Attack Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Hybrid Threat and Risk Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1 Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Scenario-Case Approach to Threat and Risk Assessment . . . . . . . . . . . . . . . . . . . . . 7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

430 432 435 436 437 438 438 439 440 441 442 442 442 445 446 446

M. Zharikova Bundeswehr University Munich, Neubiberg, Germany Kherson National Technical University, Kherson, Ukraine V. Sherstjuk Kherson National Technical University, Kherson, Ukraine S. Pickl Bundeswehr University Munich, Neubiberg, Germany e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_183

429

430

M. Zharikova et al.

Abstract

The chapter is aimed at the development of a novel spatially and temporally referenced model of hybrid attacks on critical energy infrastructure, based on the hierarchical spatial model, the time model, as well as the hierarchical energy infrastructure model. Since hybrid operations are multistage and multistep, the achievement of the strategic objective of the attacker is related to the gradual achievement of a certain sequence of smaller-level goals by elementary hybrid attacks or hybrid episodes. In this regard, a model of a goal hierarchy is proposed representing the sets of goals of different levels. Hybrid attacks are represented as scenarios composed of sequences of events restricted by time intervals. The past or simulated scenarios are accumulated in a case base to test observed sequences of events concerning their similarity to some known patterns. A case consists of a certain condition represented by the scenario of hybrid operation, as well as a solution that defines threatening critical nodes, a corresponding goal, and a set of possible effects. Cases can be accumulated by simulation on computer models. Using a scenario-case approach, threats and risks for critical energy infrastructure can be approximately assessed. The proposed approach can support dynamic decision-making whereby optimal decisions can be determined based on both generic information from the past provided by cases and rigorous anticipation about the future. It enables more effective use of information simulating the actual decision-making process in early warning systems. Keywords

Hybrid attack · Hybrid operation · Critical node · Energy infrastructure · Threat assessment · Risk assessment

1

Introduction

Social, political, and energy networks do not operate independently but are instead “nested” in one another. Each infrastructure depends on other infrastructures to function successfully. Disruptions in a single infrastructure can generate disturbances within other infrastructures (Energy Critical Infrastructure 2007). Energy is an integral part of all branches of the economy and social sphere, with a special role in ensuring the security of the development of modern society. Therefore, among all the infrastructures, energy infrastructure might be identified as the most crucial one due to the enabling functions they provide across all other infrastructure sectors, the main of which is to provide essential fuel (Wang et al. 2018; Energy Critical Infrastructure 2007). For example, water supply systems rely on electric power systems to operate their pump stations, information and telecommunication systems rely on power networks to carry out information transmission, and transportation systems rely on fuel networks to obtain power for the vehicles (Wang et al. 2018). Even highly resilient social and political networks cannot operate for long without fundamental energy and environmental networks (Maliarchuk et al. 2019).

Hybrid Attack Modeling for Critical Energy Infrastructure Protection

431

In turn, critical energy infrastructure (CEI) depends on many other critical infrastructure sectors, such as transportation, information technology, communications, finance, water, financial services, and government infrastructures (Energy SectorSpecific Plan 2015). It is especially necessary to emphasize cyber/IT dependencies that have increased dramatically over time. For example, electricity and natural gas suppliers rely heavily on data collection systems based on the information and communications technologies. These cyber/IT components are essential in monitoring and controlling the production and distribution of energy and help to create the highly reliable and flexible energy infrastructure; however, the reliance of energy infrastructure on cyber infrastructure increases its vulnerability (Energy Critical Infrastructure 2007). The dependence of CEI on other critical infrastructures (CI) increases its vulnerability. Disruptions in the energy system may transverse to other dependent infrastructure systems and possibly even back to itself, where the failure originated (Energy Critical Infrastructure 2007). This cascading and escalating characteristic of failure adds to energy network’s vulnerability. CEI is often a subject to hybrid attacks. Not only government structures of states, but also criminal and terrorist organizations have the opportunity to use both information and cyber technologies, as well as information and communication networks to achieve their goals. For the purposes of this chapter, hybrid attack (HA) to energy infrastructure is defined as the full-spectrum use of state and nonstate instruments to shift the stability and legitimacy of key systems and institutions in an energy sector (Maliarchuk et al. 2019). We assume that hybrid attacks are usually implemented sequentially as hybrid operations in the following way (Sherstjuk and Zharikova 2020): • Hostile state and nonstate actors strive to reach their strategic objectives using a wide range of activities (further considered as tools) to achieve a specific outcome, such as disinformation, espionage, cyber operations, etc. • To achieve a certain strategic objective, actors combine the use of different tools by distributing them in time, space, as well as among several targeted elements of CEI. • Actors try to carefully disguise used tools to remain under the detectability threshold and undermine the decision-making and response capabilities of their targets. • Actors aim at creating nonlinear cascading and butterfly effects using simultaneously attack vectors of different nature against different targeted elements of CEI sectors in a coordinated and strategic way. • Cascading or butterfly effects of hybrid attacks create ambiguities, which complicate the response and blur situational awareness including threat and risk assessments. Thus, HA needs to be identified on time at the earliest possible stage. The important task is to evaluate the degree of threat or risk carried by a certain HA or the sequence of HA for some targeted elements of CEI.

432

M. Zharikova et al.

Attackers often try to achieve a given strategic goal by attacking different elements in different ways using sequences of HAs. HA poses risk for critical nodes of energy network considered as vulnerable elements of CEI. Such risk can be defined as a function of consequences, vulnerabilities, and threats (Energy Critical Infrastructure 2007; Energy Sector-Specific Plan 2015). To assess threat or risk, it is necessary to understand which a critical node is targeted by the hybrid attack and what goal it pursues. The potential physical and cyber consequences of HA for critical nodes are measured as the range of loss or damage that can be expected based on a set of criteria that can be divided into four main categories such as human impact (e.g., fatalities and injuries), economic impact, impact on public confidence, and impact on government capability (Energy Critical Infrastructure 2007). The targeting of energy infrastructure by HAs is the potential to cripple large sectors of the economy and an effective way to increase the vulnerability of a state or society (Maliarchuk et al. 2019). The characteristic of CEI nodes or CEI itself that render them susceptible to destruction, incapacitation, or exploitation by HAs is called vulnerability. Vulnerability assessments identify areas of weakness that could result in consequences of concern, taking into account intrinsic structural weaknesses, protective measures, resiliency, and redundancies (Energy Critical Infrastructure 2007). Threat component of risk analysis is calculated based on the likelihood that an asset will be disrupted or attacked. Such information is essential for conducting meaningful vulnerability and risk assessments. Based on the foregoing, it can be concluded that modern CEI almost completely depends on the state of cyber and information security. This gives rise to a need to protect energy supplies and their vulnerabilities from HAs. The provision of the cyber and information security of the energy infrastructures became a crucial condition for ensuring the infrastructure protection capability (Maliarchuk et al. 2019).

2

Recent Works

Numerous existing works investigate the safety of CEI by developing its models, the models of HAs and HA’s risk analysis. Since the energy infrastructure is closely related to other infrastructures, an emerging number of researchers focus on the study of interdependencies within the energy infrastructure sectors and across other CI sectors (Wang et al. 2018). To model interactions between interdependent systems in the modeling and simulation of energy infrastructures the actor-based and agent-based approaches are widely used (Wang et al. 2018). Actor-based model is composed of actors that can make local decisions, create more actors, send messages, and determine how to respond to messages received. This approach is used in the interdependent energy infrastructure simulation system (IEISS) (Toole and McCown 2008) developed as modeling, simulation, and analysis tool designed to understand interdependent energy infrastructures. In this system, the actors realistically simulate the dynamic interactions within each of the

Hybrid Attack Modeling for Critical Energy Infrastructure Protection

433

infrastructures, with a specialization in simulating the interdependent electric power and natural gas infrastructures (Wang et al. 2018). Agent-based models consist of dynamically interacting rule-based agents. Agentbased approach is used in (Casalicchio et al. 2010) to model a system composed of a power grid and a communication network with agents representing the entire infrastructure, its subsystems, and the humans involved in the scenario. Li et al. (2016) models the integrated energy system providing electricity and natural gas. The authors build a two-hierarchy smart agent model as the basis for the system reliability analysis. The lower hierarchy are the component smart agents representing the power lines, transformers, and electricity loads, the higher hierarchy are the zone agents representing the system topology (Wang et al. 2018). The advantages of agent-based approach are as following: it can capture complicated interdependencies by simulating physical or economic flows among different infrastructures enabling the study of large-scale problems by avoiding complicated theoretical analysis; it allows analyzing the behavior of customers or decisionmakers by making certain rules. However, this approach still has limitations in that it is difficult to validate, and not all types of interdependencies can be included in one single model. Most existing agent-based models can only simulate one type of interdependencies such as the physical or logical interdependency (Zhang and Peeta 2011; Wang et al. 2018). Several approaches are proposed in the literature to simulate HAs to CEI, mainly network approach based on tree or graph models that represent critical energy infrastructures as real-world networks, and hybrid attacks within them as moving from one critical node to another (Vaiman et al. 2012; Arianos et al. 2008). In this type of models, the energy infrastructures are represented by a set of vertices connected by a set of edges, where the vertices represent nodes and the edges represent connections between the nodes. In (Chen et al. 2019), a detailed network model of an attack propagation in the power grid is represented. This model contains a preparation stage and an execution stage of the attack. The authors propose a novel HA model that combines probabilistic learning attacker, dynamic defender model, and a Markov chain model to simulate the planning and execution stages of a bad data injection attack in power grid. Network approach is also used in (Buldyrev et al. 2010) to demonstrate the cascading fault evolving between power system and communication system. In (Page et al. 2013), this approach is used to simplify the energy network model using clusters that are aggregations of network nodes to build a less detailed model (Wang et al. 2018). Network models are easy to analyze due to their high level of abstraction and simplification. The probabilistic approach is widely used to model HAs to CEI. The probabilistic algorithms are necessarily applied to capture the uncertain characteristics of the system failure. In (Rao et al. 2016), the game-theoretic models for the estimation of attack probabilities are proposed. The authors analyze the interactions between attacker and defender in a grid network and apply game theory in order to implement a probabilistic Boolean attack-defense model. The authors assume that

434

M. Zharikova et al.

the attacker obtained information about the target system prior to the attack. In this way, the probabilistic model returns success rates for both the attacker and defender. In the attack scenario, both parties assume certain action after calculating the opponent’s success rate. This approach requires knowledge about the capabilities of the infrastructure, incidental degradation, incurred costs, defender strategies, and defense priorities (Vallant et al. 2021). Many models adopt Monte Carlo simulation method, as well as Markov chains using repeated sampling to determine the properties of HAs within CEI (Wang 2019). The work from (Gao et al. 2019) applies a Monte Carlo method based on Markov chains for operation reliability in power systems. For state prediction, the authors focus on time-dependent state probabilities. A system state is defined in terms of a specific situation of its composite states, the probability of which is depicted with corresponding state models. The resulting multistate Markov model relies on a state transition matrix in order to derive the rates of failure and repair for a specific state in the system (Vallant et al. 2021). In Vallant et al. (2021), the authors provide a methodology for systematic risk assessment of cyberattacks in smart grid (SG) systems (electrical power systems that incorporate increased information processing and efficient technological solutions). The authors develop a nondeterministic system model as Markov decision process that incorporates the system architecture, the attacker’s behavior, and the existing vulnerabilities of the system with identified exploitation probabilities. The required input elements for formal verification are the system model, together with the identified attack properties. Several models have been proposed for multistep HAs (Aguessy et al. 2016; Haji et al. 2019), which represent all known paths that an attacker can follow by a tree, a graph, or a network. However, the main drawback of such models is their static nature, while the mixture of HAs is always principally dynamic taking into account the results of previous attacks and the need to achieve low detectability. Another noticed drawback of such models is associated with a lack of representability of potential cycles in the paths by tree or graph structures. Finally, one more limitation is their focus on well-known possible paths, therefore they do not consider all uncertainties and ambiguities that arise during HAs. There are many works devoted not only to modeling HAs to CEI, but also to analyzing the risk from them. For the analysis of risk of HAs to CEI the attack graphs are widely used that group all the paths an attacker may follow in an information system. Their use is attractive because they leverage already available information (vulnerability scans and network topology). However, attack graphs are static and do not contain detections or attack status and thus are not fitted for dynamic risk assessment. Several extensions of static risk assessment models have been proposed in the literature to accommodate dynamic risk assessment, but they suffer from common limitations, such as existing cycles (François-Xavier Aguessy et al. 2016). Stochastic models are proposed in the literature to analyze risk of HAs for CEI that assess distribution of specific variables reflecting vulnerability of critical nodes (Vallant et al. 2021). For example, in (Langer et al. 2015), a cyber security

Hybrid Attack Modeling for Critical Energy Infrastructure Protection

435

risk assessment for attacks in smart grids is performed analyzing the impact and likelihood of cyberattacks against smart grids (Vallant et al. 2021). Jauhar et al. (2015) introduces a model-based approach to assess security risks for failure scenarios in SGs. The authors analyze several of such scenarios, each of which is formalized in a structured model containing information about concrete vulnerabilities and attacker characteristics. These models are used in order to reason about failures in every scenario (Vallant et al. 2021). Lee (2015) provides a theoretical analysis on risk assessment and failure scenario ranking in the electric sector. The work analyzes vulnerabilities as well as the resulting consequences and possible mitigation mechanisms. These vulnerabilities are listed and suggestions are given to define their naming conventions and classification (Vallant et al. 2021). Through this literature review, we could see that researchers have analyzed hybrid attacks to critical energy infrastructure, as well as risk from them but there exists a research gap to address all the issues under a single framework. Developing effective comprehensive approaches that can emulate HAs to energy systems and analyze risk from them is still a critical challenge, maybe because of their high complexity. Thus, HA models focused on the threat and risk assessment need further research. Based on the literature review, we can conclude that complex network theory, developed from graph theory, in combination with a probabilistic and scenario-case approaches, provides an attractive tool to reveal the hidden interdependencies of CEI affected by HAs. This approach can facilitate the development of the model of cascading effect among interconnected critical nodes. We assume that a hybrid threat corresponds to a certain series of HAs. Therefore, we need to put together a picture of a hybrid operation from a mosaic of various unrelated (at the first glance) events, each of which reflects a certain HA. Fortunately, such events are implicitly connected in time and location, which enables an opportunity to identify their connections by targets and goals. Besides, we assume that hybrid operations are quite repetitive with respect to their tools and goals, so we propose to use a scenario-case approach to assess hybrid threats (Zharikova et al. 2022). Therefore, the aim of the chapter is to develop a novel comprehensive approach to analyze threat of HAs to CEI and correspondent risk, based on graph theory and probabilistic approach.

3

Definitions

Let us introduce some definitions which will be used later on. Critical energy infrastructure – the organizational structure enabling the largescale transportation, directing and managing of energy flow from producer to consumer, the incapacity or destruction of which would negatively affect national economic security, national public health or safety, or any combination of those matters.

436

M. Zharikova et al.

Critical node – the key vulnerable elements of critical energy infrastructure that are crucial to its effective functioning, which, when disrupted, lead to a breakdown in network functionality and potentially spark cascading failures across networks. Hybrid attack to critical energy infrastructure – a combination of tools that target (threaten) a specific critical node within critical energy infrastructure using multiple ways such as conventional and unconventional, as well as military and nonmilitary (Giannopoulos et al. 2021). The consequences of a hybrid attack – human impact, economic impact, impact on public confidence, and impact on government capability (Energy Critical Infrastructure 2007). Threat – the likelihood that a critical node will be affected by hybrid attack and disrupted. Vulnerability – characteristics of a critical node that render it susceptible to destruction, incapacitation, or exploitation by hybrid attacks. Risk – a function of consequences, vulnerabilities, and threats.

4

Energy Infrastructure Model

The energy infrastructure is divided into interrelated domains (Maliarchuk et al. 2019), each of which contains a set of segments (sectors) connecting critical nodes in a network. Thus, domains can be represented as networks containing other networks. For example, fuel industry domain contains the following sectors: coal mining, gas, oil, peat, and chemical industries while oil industry sector contains such critical nodes as pumping stations, pipelines, refineries, etc. Electrical energy domain contains such sectors as the thermal and hydroelectric power stations, nuclear power plants, and alternative energy sources, each of which includes critical nodes of various types (wind power stations, solar power stations, etc.). The same goes for the generation infrastructure domain. Disruption in some critical node leads to a breakdown in domain functionality and potentially spark cascading failures across domain (Maliarchuk et al. 2019). The energy infrastructure model should reflect the interconnections between domains, sectors, and critical nodes within each sector. Let be an energy infrastructure including a set of domains = {δi }ni=1 . m Suppose each domain δ i contains a set of sectors, ςij j =1 ∈ δi , and each sector ςij l contains a set of critical nodes, oij k k=1 ∈ ςij . The energy infrastructure model can be defined by a certain graph , in which vertices ϕij ∈ represent critical nodes of various types as well as edges φ ik ∈ represent connections between the critical nodes (Aksoy et al. 2018; Fig. 1). Clearly, the graph can change over time, so its nodes might appear or disappear, and connections might become stronger or weaker. The nodes and edges within the graph should be referenced both spatially and temporally. To consider the variety of nodes and their connections, we introduce classes of nodes as well as classes of edges and define the corresponding hierarchies. Suppose

Hybrid Attack Modeling for Critical Energy Infrastructure Protection

437

Fig. 1 Domain graph containing subgraphs of sectors

Cϕ = {cϕ1 , . . . cϕm } is a set of node classes and Cφ = {cφ 1 , . . . cφ u } is a set of edge classes. Thus, the node hierarchy can be represented by ϕ = cϕ⊥ , Cϕ, := f ∈ Rp : f > 0 = int R≥

(2)

It is worth noting that because of the contradictory nature of the studied objective functions, finding a solution which is optimal for all objectives, concurrently, will be impossible. In other words, the multi-objective optimization problems are illdefined.

2.4

Cone

The subset C of Rp is a cone if for d ∈ C and a ≥ 0, a ∈ R, ad ∈ C.

2.5

Pointed Cone

The cone C is a pointed cone if C ∩ (−C) = {0}. In other words, if d ∈ C and

d = 0, then −d ∈ C. The cone C is convex if for all d2 ∈ C, there is d so that

d + d2 ∈ C. The objective search space of single-objective optimization problems is unidimensional and the optimal solution for a minimization problem is the one having the least value of the objective function. On the contrary, the objective search space of multi-objective optimization problems is associated with more than one dimension. In this regard, the decision maker seeks to find a solution which is the best one according to the priorities of the problem and preferences of the decision maker.

1484

A. Esmaeel Nezhad and P. H. J. Nardelli

Fig. 2 A representation of dominated and nondominated solutions along with a Pareto frontier

2.6

Efficient Solutions

ˆ A feasible solution kˆ ∈ K is efficient if there is no k ∈ K such that Z(k) ≤ Z(k). 2

ˆ ˆ ˆ If k is efficient then Z(k)would be a nondominated solution. If k, k ∈ K&Z(k ) ≤ Z k2 , k would dominate k2 and accordingly Z(k ) would dominate Z(k2 ). Figure 2 shows a schematic representation of nondominated and dominated solutions along with a Pareto frontier.

2.7

Ideal Point

f I = f1I , . . . , fpI is an ideal solution and defined as: fgI := min Z (k) = min fg , k∈K

2.8

f ∈F

g = 1, 2, . . . , p

(3)

Nadir Point

f N = f1N , . . . , fpN is a Nadir point and defined as follows:

fgN := max Zg (k) = max fg , k∈KE

f ∈FN

g = 1, 2, . . . , p

(4)

where the set of efficient points and the set of nondominated solutions are depicted by KE and FN , respectively. The ideal solution is the optimal solution, if it is feasible. Nevertheless, this condition is not true in most cases due to the contradictory behavior of objective functions. Deriving the ideal point would not be a difficult task to do as it is needed to solve p single-objective optimization problems. On the contrary, computing the Nadir point including optimization in the set of efficient points would be quite challenging. It is noteworthy that there

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1485

is not an efficient method to derive YN in multi-objective optimization problems. Accordingly, heuristic and metaheuristic methods are mainly used to obtain the Nadir point (Deb 2014).

2.9

Utopia Point

U If = ε = (ε1 , . . . , εp ) is a vector with small positive components, f U U U I f1 , . . . , fp , which is defined as fg := fg − εk , g = 1, 2, . . . p is a utopia point .

2.10

Optimal Solution

The feasible solution kˆ ∈ K is an optimal solution in single-objective optimization ˆ ≤ Z(k) ∀k ∈ K. problem if Z(k)

2.11

Strictly Optimal Solution

ˆ < Z (k) ∀k ∈ K. The feasible solution kˆ ∈ K is a strictly optimal solution if Z(k)

2.12

Scientometric Analysis

Analysis of the published documents based on the bibliometric analysis is regarded as one of the most useful tools to discuss a specific topic and its trend in the literature. Accordingly, this section investigates the published papers related to multi-objective optimization and energy systems. The Scopus database has been used to obtain the results. It is noteworthy that three keywords have been used to more accurately search the documents related to multi-objective optimization and energy systems. These keywords are “multi-objective,” “energy,” and “power.” Nevertheless, various combinations of keywords can be used to search the results. The search led to 4887 documents. Such information helps researchers about the rate of published papers, the subject area, and the interest in specific topics. The information can be used by researchers in their future projects. The number of published documents during the years 1978–2021 is shown in Fig. 3. As this figure depicts, the number of published documents has been increasing over these years, proving the significance of the topic and the increasing applications of the multiobjective optimization in energy systems. Besides, the subject areas have been shown in Fig. 4 showing that engineering in general and energy in particular have received the highest contributions.

1486

A. Esmaeel Nezhad and P. H. J. Nardelli

800 Number of documents

700 600 500 400 300 200 100 0 1970

1980

1990

2000 Year

2010

2020

2030

Fig. 3 The number of documents published during 1978–2021

Besides, an analysis has been done by using the data obtained from Scopus and the connections between the keywords have been network-visualized by using the VOS viewer software as shown in Fig. 5.

3

Multi-Objective Optimization Algorithms

There have been numerous multi-objective optimization algorithms proposed to tackle multi-objective optimization problems. These methods are mainly categorized into classical and heuristic and metaheuristic methods. This section presents a review on classical mathematical multi-objective optimization algorithms. These methods include epsilon-constraint technique and normal boundary intersection (NBI).

3.1

Epsilon-Constraint Technique

The epsilon-constraint approach has been well established in solving multi-objective optimization problems even with more than two objectives. The fundamental of this method is based on assigning one of the objective functions as the main or first objective function to the problem. For a minimization problem, afterward, the problem would be solved where the first objective function is minimized and other objectives would be added to the problem as extra constraints (Javadi and Esmaeel Nezhad 2019). The general expression of the epsilon-constraint approach can be observed in (5):

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1487

27 Engineering 24 Energy 11 30 112

78

131

88 98

235 260

136

27

Computer Science

10 4 2 1

Mathematics Environmental Science

1

Materials Science Physics and Astronomy 3141

806

Social Sciences Chemical Engineering

819

Business, Management and Accounting Decision Sciences Earth and Planetary Sciences

1383 2362

Chemistry Biochemistry, Genetics and Molecular Biology Agricultural and Biological Sciences Multidisciplinary Economics, Econometrics and Finance Medicine

Fig. 4 Subject areas of multi-objective optimization

Min f1 (x) subject to f2 (x) ≤ e2 f3 (x) ≤ e3 . . . fp (x) ≤ ep

(5)

It is worth mentioning that the number of objectives that should be optimized and decision variables vectors are indicated by p and x, respectively. In this specific case, all objectives are intended to be minimized. However, it is noted

1488

A. Esmaeel Nezhad and P. H. J. Nardelli

Fig. 5 The network visualization of subject areas

that the maximum and minimum values of p-1 objective functions should be accurately determined and assigned to the model for appropriately using the epsilonconstraint technique. These upper and lower bounds can be effectively specified by utilizing the pay-off table. The method starts with determining the individual optimal values of all objective functions fi and forming the pay-off table. It is noteworthy that fi∗ x ∗i and x ∗i depict the optimal values of fi and the decision variables that optimize these objective functions. Now the values of other objective functions, f1 , f2 , . . . , fi-1 , fi + 1 , . . . , fp , should be precisely specified by utilizing values are indicated the optimal point of the objective function fi. The obtained by f1 x ∗i , f2 x ∗i , . . . ,fi−1 x ∗i , fi+1 x ∗i , . . . , fp x ∗i . As the following pay off table illustrates the row i includes f1 x ∗i , f2 x ∗i , . . . , fi∗ x ∗i , . . . , fp x ∗i . Likewise, the remaining rows would be computed. f1∗ x ∗1 ⎜ .. ⎜ ⎜ . ∗ ⎜ = ⎜ f1 x i ⎜ .. ⎜ ⎝ . f1 x ∗p ⎛

⎞ · · · fi x ∗1 · · · fp x ∗1 ⎟ .. .. ⎟ . . ⎟ ∗ ∗ ⎟ ∗ · · · fi x i · · · fp x i ⎟ ⎟ .. .. ⎟ . . ⎠ · · · fi x ∗p · · · fp∗ x ∗p

(6)

The dimension of this pay-off table is p × p. The values calculated for the fj are given in the column j of the pay-off table where the smallest and greatest values show the lower and upper bounds of that constraint used by the epsilon-constraint technique. Some of the concepts employed by the epsilon-constrain technique are the Utopia point, Nadir point, and pseudo-Nadir point. In this respect, the Utopia point refers a particular point located outside the feasible space relating to all objective functions that are concurrently at their best points. The Utopia point can be mathematically represented as follows:

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

f U = f1U , . . . , fiU , . . . , fpU = f1∗ x ∗1 , . . . , fi∗ x ∗i , . . . , fp∗ x ∗p

1489

(7)

The Nadir point is another point located in the objective space at which all objectives are concurrently at their worst points. f N = f1N , . . . , fiN , . . . , fpN

(8)

where fiN = max fi (x) , x

subject to x ∈

(9)

The feasible space is indicated by . A point with a relatively similar concept is the pseudo-Nadir point mathematically shown as follows: f SN = f1SN , . . . , fiSN , . . . , fpSN

(10)

fiSN = max fi x ∗1 , . . . , fi∗ x ∗i , . . . , fi x ∗p

(11)

The abovementioned points are stated in the objective space which is a vector space and its dimensions are determined by the objective functions. By using the relation (12), the intervals of all objective functions are determined by utilizing the Utopia and pseudo-Nadir points denoted by superscripts U and SN, respectively. fiU ≤ fi (x) ≤ fiSN

(12)

Any multi-objective optimization problem is tackled aimed at deriving the ∗ optimal Pareto set. The Pareto optimal solution is shown by the point ∗ x ∈ , provided that there would not be x ∈ in a way that fi (x) ≤ fi x for i = 1, 2, . . . , p, associated with a minimum one strict inequality. In the next stage, the epsilon-constraint approach splits the interval of objective functions fi for i = 2, . . . , p, to equally separate distances shown by q2 , . . . , qp . This is done by using (q2 –1), . . . , (qp –1) intermediate points. Accordingly, there would be (q2 + 1), . . . , (qp + 1) grid points in total for fi for i = 2, . . . , p, respectively, accounting for the lower and upper limits of each objective function. Consequently, the total p number of subproblems to tackle would be (qi + 1) in which the subproblem i=2

(n2, . . . , np) would be mathematically represented as follows: Min f1 (x) subject to f2 (x) ≤ e2,n2 , · · · , fp (x) ≤ ep,np

(13)

1490

A. Esmaeel Nezhad and P. H. J. Nardelli

e2,n2 =

f2SN

e2,np =

fpSN

− −

f2SN − f2U q2 fpSN − fpU qp

× n2

n2 = 0, 1, . . . , q2

(14)

× np

np = 0, 1, . . . , qp

(15)

The constraints of the original problem are all considered in the abovementioned optimization subproblems. The members of the Pareto optimal set are obtained by solving these subproblems, while infeasible solutions are avoided. This method is superior compared to the weighted-sum method in terms of the following items (Javadi et al. 2020): • Weighted sum method leads to efficient extreme solutions for linear programming models while nonextreme efficient solutions are produced by using the epsilon-constraint approach. • Epsilon-constraint technique is capable of producing unsupported efficient solutions for multi-objective mixed-integer linear programming problems and multiobjective linear programming problems. • On the contrary to the weighted-sum method, the epsilon-constraint approach does not need any scaling of objective functions. However, there are two main demerits with the epsilon-constraint approach. • The interval of each objective in the efficient set is not optimized. Thus, the lexicographic optimization should be used along with the epsilon-constraint technique. • Some dominated or inefficient solutions might be produced by this method, while the augmented epsilon-constraint method can overcome this issue (Nezhad et al. 2014). One important issue that should be taken into account when forming the pay-off table is to check if the solutions are all Pareto optimal solutions. In case any other optimal solution is found, the previous one would not be a Pareto optimal solution. To this end, the lexicographic optimization can be used when forming the pay-off table. The lexicographic optimization is on the basis of optimizing the first objective function of the total number of objective functions and afterward, among the possible alternative optima, optimize for the next objective function, etc. The lexicographic optimization would be briefly stated and defined as: by optimizing the first objective function, its optimal value would be f1 = z1 *. When optimizing the next objective function, the optimal value of the first objective function is considered and added to the optimization problem as a constraint. As a result, the optimal solution of the second objective function would be f2 = z2 * while f1 = z1 *. The remaining objective function would also be optimized in the same way to form the pay-off table. It is noteworthy that the solutions reported by the

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1491

lexicographic optimization are nondominated solutions. The second demerit of the epsilon-constraint method can be effectively solved by converting the constraints added to the original problem due to objective functions to equalities by employing a slack variable method (Nezhad et al. 2014). The mathematical expression of this technique is: ⎧ P ⎪ si ⎪ ⎪ Min f − r (x) 1 1 ⎪ ri ⎪ ⎨ i=2 subject to ⎪ ⎪ ⎪ fi (x) + si,ni = ei,ni , i = 2, . . . , p ⎪ ⎪ ⎩ x∈

(16) &

si,ni ∈ R +

It is noteworthy that the slack variables are shown by s2 , . . . , sp added for the constraints (13). Besides, the second item of the objective function in (16) is utilized to avoid any scaling issue. The range of the objective function i is obtained as ri = fiSN −fiU . Indeed, the slack variables si relating to the objective function fi would be scaled to the range of the first objective. The mentioned technique is the augmented epsilon-constraint technique.

3.1.1 Applications This technique has largely been employed to solve energy systems’ problems. In this respect, this section reviews some of the applications of the epsilon-constraint technique. In this respect, Ref. (Yang et al. 2021) tackled the economic scheduling of a combined cooling, heat, and power (CCHP) microgrid while the peak load demand is also intended to be minimized. The problem has been formulated as a MILP multi-objective optimization problem tackled by the augmented epsilonconstraint approach. The epsilon-constraint approach was used in Ref. (Gazijahani et al. 2020) to solve the joint reserve and energy scheduling problem of a microgrid resourced by renewable energies and demand response programs. The augmented epsilon-constraint approach was applied to tackle the long-term planning problem of power systems to determine the best generation mix and best generation sites as well as the best options to reinforce the transmission system (Hamidpour et al. 2021). A multi-objective optimization-based decision-making framework was designed in Ref. (Hajiamoosha et al. 2021) for the energy management of a microgrid equipped with renewable energies. A peer-to-peer (P2P) market was also designed by using a multi-objective optimization model in Ref. (Chernova and Gryazina 2021) while taking into account the operational preferences of the end users.

3.2

Normal Boundary Intersection

The NBI approach is regarded as an efficient multi-objective optimization method with high capabilities. In general, this method is associated with two important merits compared to the epsilon-constraint technique (Simab et al. 2018):

1492

A. Esmaeel Nezhad and P. H. J. Nardelli

• Scale-free performance where the size of the objective functions would not impact the performance of the NBI approach • Producing evenly distributed Pareto front Similarly to the epsilon-constraint technique, the algorithm starts with forming the pay-off table where the formulations for obtaining the pay-off table with respect to the Utopia, Nadir, and pseudo-Nadir points are the same. After that, the solutions would be normalized as follows: f i (x) =

fi (x) − fiU

(17)

fiSN − fiU

Accordingly, the resulting space is a dimension-free space. A significant concept in the NBI method is the convex hull of individual minima known as CHIM and derived as a combination of the points in each row of the computed pay-off table. P(β 1 , β 2 , . . . , β p ) shows the points in the normalized space that is specified as follows: ⎡

⎤ β1 φ 11 + · · · + βp φ 1p ⎢. ⎥ ⎥ ⎢ ⎢ ⎥ P β1 , β2 , . . . , βp = ⎢ . ⎥ ⎢ ⎥ ⎣. ⎦ βp φ p1 + · · · + βp φ pp It is noteworthy that β i shows a positive value so that

p

(18)

βi = 1. Consequently,

i=1

the original multi-objective optimization problem is converted into p optimization subproblems while all of them are single objective in which the objective function of the subproblem D is defined as maximization of the distance measured between the CHIM and the Pareto front. This problem is mathematically stated by using the following expression: Maximize D s.t : φ β + D nˆ = F (x) where : { x ∈ R | g(x) ≤ 0, h(x) = 0}

(19)

The point where the boundary and normal of the feasible space nearest to the origin intersect would be regarded as the optimal solution of the subproblem. A point-wise representation of the Pareto set is acquired by tackling the problem (19) and assigning various values of β to the problem necessitating several runs. One significant merit of this method is that the size of the Pareto set can be effectively managed by appropriately specifying the values of β. Nevertheless, the higher number of Pareto optimal solutions leads to dramatically increasing the

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1493

computational load of the problem. Hence, the decision maker should decide on the desired size of the Pareto set (Ahmadi et al. 2020).

3.2.1 Applications The NBI method has been applied to different energy systems to tackle the multiobjective optimization problems. Ref. (Ahmadi et al. 2015a) has used the NBI approach to tackle the short-term hydrothermal self-scheduling problem taking into account the operating cost of the system and total environmental emissions. The problem was formulated by using a MILP model and the results derived from simulation were compared to those reported by other methods. The hydrothermal scheduling problem has been solved by using the NBI method and a MILP model in Ref. (Ahmadi et al. 2015b). A Bender’s decomposition-based multi-objective decision-making model has been developed in Ref. (Charwand et al. 2015) for an electricity retailer which is an active participant in the energy market. The NBI method has been used in Ref. (Fang and Xu 2020) for scheduling a renewable energy-based shipboard microgrid through a robust optimization technique. The long-term generation and transmission expansion planning problem has been tackled in Ref. (Mavalizadeh et al. 2018) by using the NBI method and robust optimization.

4

Multi-Criteria Decision-Making Techniques

4.1

Decision-Making Criteria

As mentioned above, the results would be desired if the decision made is based on different criteria. For instance, in the day-ahead power system operation, the simultaneous minimization of costs, minimization of power losses, and maximization of voltage security are examples of the decision-making criteria. In classical operations research problems, only one criterion is studied. In the MCDM, if decision making is on the basis of multiple objectives, then it is called multi-objective decision making.

4.2

Measurement Criteria

As previously mentioned, some criteria are quantitative and some are qualitative. Moreover, each criterion has its own measurement scales making the comparison between the criteria very difficult. However, they should be measured individually and converted to a comparable quantity. Measurement scales are categorized into quantitative and qualitative scales. The qualitative scales include nominal and ordinal ones. In addition, the quantitative scales comprise interval and ration ones. The nominal scales include categorization based on gender, nationality, etc., measured by counting. The ordinal scales involve categorization based on a feature while the most important one (regarding that specific feature) is ranked first while the least important one is ranked last. Interval scales consider the real intervals,

1494

A. Esmaeel Nezhad and P. H. J. Nardelli

besides ranking. This type of scales includes the arbitrary zero-like temperature. Ratio scales are like interval scales but with absolute zero.

4.3

MCDM Models

If the sets of solutions of the problem are countable, the model is discrete or multiattribute such as selecting a technology among different available technologies. If the sets of solutions of the problem are uncountable, the model is continuous or multi-objective like determining the optimal life of a transformer, battery, or generator such that the cost is minimized and reliability is maximized. Furthermore, if any shortfall in any of the criteria is compensated by other criteria, the model is compensatory like better quality at a higher cost. Otherwise, the model is noncompensatory like the criteria to obtain a driving license. If the decision making is done based on the observations of one person, the model is individual, otherwise it is a group model. The MCDM methods have largely been deployed to deal with power system problems. Accordingly, this section reviews some well-known mathematical optimization techniques, and their applications to tackle multi-objective optimization problems of power system are investigated.

4.4

Scientometric Analysis

This section provides a scientometric analysis on the applications of the MCDM methods in energy systems, mainly focusing on power system problems. In this respect, the Scopus database has been used and three keywords as “Multi-Criteria Decision Making AND Energy AND POWER” have been searched and the obtained results have been analyzed. The search returned 892 results published during the years 1996–2021 as depicted in Fig. 6. Furthermore, the subject areas of the search have been depicted in Fig. 7 indicating that energy and engineering have received the highest contributions.

4.5

MCDM Methods

This section provides a review on some well-known MCDM approaches that have largely been employed to cope with energy systems problems. In this regard, the fuzzy satisfying method, the VIseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) decision maker, the AHP, the ANP, and TOPSIS are introduced and a review is provided on their application in power systems problems.

4.5.1 Fuzzy Satisfying Method By deploying the multi-objective optimization techniques, the set of optimal solutions, called “Pareto optimal front,” is derived. In this stage, the decision maker faces multiple optimal solutions that each is desirable. However, one of the

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1495

180 Number of Documents

160 140 120 100 80 60 40 20 0 1995

2000

2005

2010 Year

2015

2020

Fig. 6 The number of documents published during 1996–2021

solutions should be picked and taken as the most desired strategy. The fuzzy decision maker is associated with proven capabilities in selecting the most desired solutions according to the preferences of the decision maker and priorities of the problem (Reza Norouzi et al. 2014). This method performs on the basis of defining linear membership functions for each objective function and each member of the Pareto set. By using these linear membership functions, the distance of each member of the set with respect to the maximum and minimum values of the sets and the values of other solutions is determined. It is noteworthy that there are two different linear membership functions defined for the minimization and maximization problems as represented in (20) and (21), respectively.

μrn n=2

μrn n=1

=

=

⎧ ⎪ ⎨ ⎪ ⎩

⎧ ⎪ ⎨ ⎪ ⎩

1 fnSN −fnr fnSN −fnU

fnr ≤ fnU fnU ≤ fnr ≤ fnSN ≥

0

fnr

0

fnr ≤ fnSN

fnr −fnSN fnU −fnSN

1

fnSN ≤ fnr ≤ fnU fnr

≥

(20)

fnSN

(21)

fnU

In the next stage after specifying the value of the membership of each objective function for each member of the Pareto set, the total membership value of each Pareto optimal solution should be determined accordingly. This total membership function which is also linear and defined as (22):

1496

A. Esmaeel Nezhad and P. H. J. Nardelli

Energy Engineering Environmental Science 21

8

5

13

3

26 30 37

Computer Science

2 1

Mathematics

1

27 27

Business, Management and Accounting

33

Social Sciences 61

516 Decision Sciences

64

Physics and Astronomy

125

Earth and Planetary Sciences 158

Agricultural and Biological Sciences Materials Science 405 276

Chemical Engineering Economics, Econometrics and Finance Chemistry Medicine Multidisciplinary Biochemistry, Genetics and Molecular Biology

Fig. 7 Subject areas of MCDM

p

μ = r

wn .μrn

n=1 p

n=1

(22) wn

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1497

where wn is a weighting factor indicates that to what extent each objective function in the decision-making process is significant. These weighting factors are specified and assigned to the model by the decision maker. The solution with the maximum value of the total membership is the most compromise solutions selected as the best strategy to implement (Mansouri et al. 2021). Applications The fuzzy satisfying method as a well-established MCDM method has been used in Ref. (Nojavan et al. 2017) and Ref. (Majidi et al. 2017) to select the best operational strategy among those obtained by utilizing the epsilon-constraint approach and weighted sum method, respectively, to operate a grid-tied microgrid resourced by photovoltaic (PV) panels, fuel cells, and batteries. The robust scheduling problem of electric vehicle (EV) aggregators was tackled in Ref. (Ahmadi-Nezamabad et al. 2019) by deploying the epsilon-constraint method and the best Pareto optimal solution is picked by employing the fuzzy decision maker. The interactive fuzzy satisfying technique and an evolutionary optimization algorithm were employed in Ref. (Basu 2004) to handle the multi-objective short-term hydrothermal scheduling problem. A hybrid stochastic-interval optimization method has been presented in Ref. (Jamalzadeh et al. 2020) to model the multi-objective operation of an energy hub. In this respect, the weighted sum method was put into use to tackle the problem and the best Pareto solution is selected by deploying the Fuzzy satisfying technique. Ref. (Pourghasem et al. 2019) proposed a hybrid optimization method based on the exchange market algorithm and weighted sum method to solve the multi-objective scheduling of combined heat and power (CHP) island microgrids. In this respect, the most operational strategy is selected among the Pareto optimal solutions by utilizing the fuzzy satisfying technique.

4.5.2 VIKOR Decision Maker The VIKOR decision maker as an effective method was first designed by Opricovic to tackle MCDM problems. By utilizing this approach, a positive ideal value and a negative ideal value would be defined to measure the relative distance of every solution from the optimal Pareto frontier (Javadi et al. 2021). Afterward, a compromise ranking is made to specify the significance of every member of the Pareto set with P members, depicted by xj as below: 1. Presenting the rating functions denoted by fij indicating the value of the ith objective function for the optimum shown by xj and determining the upper and lower bounds of the objectives depicted by fi+ and fi− . If the criterion is associated with a merit, it is positive and the related values would be calculated as below:

# " fi+ = max fij |j = 1, 2, . . . , m

(23)

1498

A. Esmaeel Nezhad and P. H. J. Nardelli

# " fi− = min fij |j = 1, 2, . . . , m

(24)

2. The group utility measure depicted by Sj and the individual regret measure depicted by Rj would be derived by using (25) and (26), respectively. It is noted that these values are obtained for all members of the Pareto set. Besides, the weighting factor of each criterion is shown by wi . Finally, Qj would be calculated as (27) while the Pareto solution with the least value of Qj is the most compromise one.

+ fi − fij Sj = wi + − f − f i i i=1 n $

+ & fi − fij wi + fi − fi−

(25)

% Rj = maxi ' Qj = w j

( ( ' Rj − R + Sj − S + + 1 − w , j = 1, 2, . . . . , P j S− − S+ R− − R+

(26)

(27)

where S + = Min

# " Sj |j = 1, 2, . . . , P

(28)

S − = Max

# " Sj |j = 1, 2, . . . , P

(29)

R + = Min

# " Rj |j = 1, 2, . . . , P

(30)

R − = Max

# " Rj |j = 1, 2, . . . , P

(31)

wj are the weights utilized for the maximum group utility point of view, and (1 − wj ) are utilized in the support of the individual regret. wj is generally set to 0.5 when in case of two objective functions with equal significance. 3. By deploying Sj and Rj and Qj for ranking Pareto optimal solutions, three sets of ranking would be obtained to pick the best Pareto optimal solution with respect to the priorities of decision making. Accordingly, the best Pareto optimal solution is picked by balancing the maximum Sj depicted by Min (Sj ) and the minimum Rj. depicted by Max (Rj ). Hence, the solution with the least value of Qj would be the intended result.

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1499

Applications The VIKOR decision maker was deployed in Ref. (Javadi et al. 2021) to select the best strategy for a residential end user to operate the smart home appliances while connected to the grid and equipped with the solar power generation. In this respect, the user discomfort index has also been taken into consideration as one of the objective functions and the other objective function is the total cost including the daily electricity bill of the customer. The VIKOR decision maker has been proposed in Ref. (San Cristóbal 2011) for choosing the renewable energy project in Spain. In this regard, this method was used together with the AHP to select the best strategy. Ref. (Katal and Fazelpour 2018) tried to specify and prove the compatibility of the current power plants in Iran by utilizing observational data and VIKOR decision maker. Ref. (Ahmadi et al. 2020) utilized the VIKOR decision making method to select the optimal solution obtained by using the NBI technique for the security-constrained unit commitment problem in the presence of renewable energies and electric vehicles. The VIKOR decision maker has been deployed in Ref. (Wu et al. 2016) under linguistic information to investigate the uncertainty of nuclear plants potential supplier selection. Ref. (Hongzhan and Ting 2011) presented a combinatorial decision-making framework on the basis of entropy weight and VIKOR for the long-term transmission expansion planning problem. The extension of VIKOR decision maker has been utilized in Ref. (Ci et al. 2012) for the expansion planning of power transmission systems.

4.5.3 TOPSIS Selecting an option among the Pareto optimal set is indeed a posteriori approach and the technique used for this relation should be sufficiently effective and efficient. In general, multi-attribute decision-making methods would be employed to this end. The TOPSIS is known as one of the well-established posteriori decision-making methods and widely applied to various multi-objective decision-making problems. This approach is on the basis of ranking the options and picking the best optimal solution among the finite set of Pareto optimal solutions taking into account multiple conflicting attributes. It is noteworthy that the selected option would be associated with the maximum performance and desirability for all criteria. The fundamental of this method is on the basis of selecting the closest solution to the ideal solution, and farthest from the negative ideal strategy (Ahmadi et al. 2015c). The normalized decision-making matrix would be obtained as follows (Ahmadi et al. 2015c): DMij P =1 DM Pj

NDM ij = n

(32)

It is noteworthy that DM = {DMij | i = 1, 2, . . . , n; j = 1, 2, . . . , m} is a matrix with n and m as its dimensions where the number of objective functions is indicated by m and the number of objective functions is shown by n, respectively. Besides, the performance rating of the option Xj is illustrated by DMij taking into consideration the criterion ATi . The entropy value would also be utilized to assess the decision data as follows:

1500

A. Esmaeel Nezhad and P. H. J. Nardelli

−1 $ NDM ij ln NDM ij ln n n

EV j =

(33)

i=1

It is noted that the divergence degree of the average intrinsic information shown by Di associated with every criterion ATi for j = 1, 2, . . . , m would be derived as below: Dj = 1 − EV j

(34)

Accordingly, OWij which is the objective weighted normalized value would be obtained as: OW ij = wi × NDM ij

(35)

The TOPSIS decision maker utilizes two parameters, AT+ and AT− , which are the positive and negative ideal values, respectively, to produce a total performance matrix defined for any of the Pareto set members as follows: AT + = (max (OW i1 )

max (OW i2 )

...

+ + max (OW im )) = OW + 1 , OW 2 , . . . , OW m (36)

AT − = (min (OW i1 )

min (OW i2 )

...

− − min (OW im )) = OW − 1 , OW 2 , . . . , OW m (37)

It is noteworthy that the n-dimensional Euclidean distance would be employed to evaluate the interval between the solutions. The separation of any of the solutions from the ideal solutions is mathematically computed as: )

Dj+

m $ 2 = OW j i − OW + i

*1/2 , j = 1, 2, . . . , n

(38)

, j = 1, 2, . . . , n

(39)

i=1

)

m $ 2 Dj− = OW j i − OW − i

*1/2

i=1

The relative interval between the solution and the ideal solution considering AT+ would be derived as below: RCj =

Dj− Dj+ + Dj−

,

j = 1, 2, . . . , n

(40)

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1501

Remember that as Dj− ≥ 0 and Dj+ ≥ 0, it can be concluded that RCj ∈ [0, 1]. The best compromise Pareto optimal solution would be introduced as the one with the highest value of RCj .

Applications The TOPSIS decision-making method was utilized in different areas of study to select the most compromise solution among the Pareto optimal solutions. In this regard, Ref. (Ahmadi et al. 2015c) has used the TOSIS decision maker to pick the best operational strategy derived by using the NBI method to operate a power system with combined heat and power (CHP) units. A combinatorial method with TOSIS and fuzzy-AHP was utilized in Ref. (Afsordegan et al. 2016) for the uncertain decision making in energy systems planning. Ref. (Sengül ¸ et al. 2015) utilized the fuzzy-TOPSIS decision-making method for the sake of ranking the renewable energy-based supply system in Turkey. Furthermore, the energy system planning has been carried out in Ref. (Kaya and Kahraman 2011) by deploying the modified fuzzy-TOPSIS technique. Moreover, an enhanced fuzzy-TOPSIS method was developed in Ref. (Rani et al. 2020) to choose renewable energy sources. The thermal energy storage in the concentrated solar systems has been evaluated by using the fuzzy-TOPSIS method in Ref. (Cavallaro 2010). Ref. (Nazari et al. 2018) utilized the TOPSIS method to find the best location for solar farms.

4.5.4 AHP One of the most well-known approaches in MCDM is absolutely AHP , enabling the decision maker to pick the most compromise solution by utilizing relevant information or by ranking the Pareto solutions (Singh and Nachtnebel 2016). There are different stages associated with this approach to select the most wanted Pareto optimal solution as selecting the decision alternatives and assessment criteria, deriving the performance indexes for the assessment matrix, converting into proportionate units, assigning weights to the criteria, putting in order and scoring the alternatives, carrying out a sensitivity analysis, and ultimately making the best strategy. As expected, the decision maker would assign the satisfying result as a target showing the desired or expected outcomes. It is noteworthy that the alternatives would be provided in various ways such as negotiation, gathering data from the alliance partners, field investigation, studying and conferencing, reports, innovative activities, and many other methods to gather information. It is of great significance to hierarchically structure different aspects in criteria and targets since the start of the project (Ganoulis 2008). Moreover, all criteria, subcriteria, and targets should be precisely investigated by the manager. In this respect, the AHP has extensively been utilized in MCDM problems performing on the basis of multiple groups of decision makers included while groups are conflicting. The fundamentals of AHP are in accordance with the axiomatic foundation producing the theoretical information on which the approach is established. In general, the mentioned axiomatic foundations would be stated as follows:

1502

A. Esmaeel Nezhad and P. H. J. Nardelli

• The reciprocal feature which is regarded as the basic item in making pairwise comparison. Assume DM(EW,ER) a pairwise comparison of items W and R considering their parent, item Y, showing that how many times more the item W owns a feature than that of item R, then DM(ER,EW) = 1/PC(EW,ER). If W is 6 times greater than item R, then item R would be one-sixth as large as W. • The homogeneity axiom describes that the items which are compared must not differ by a large extent, otherwise there would tendency toward great errors in the conclusion. Homogeneity is regarded as the feature of a person’s ability to make pairwise comparisons between different items which are relatively similar taking into consideration a common feature and thus, the requirement to put them in order showing the hierarchy. • Dependence of a lower level on the adjacent higher level. This synthesis axiom declares any assessment of the priorities of the items within a hierarchical manner without any dependency on the lower-level items. It is needed for the fundamental of the hierarchical composition to assign, and obviously means that the significance of the higher-level objectives must not rely on the priorities or weighting factors of the lower-level items. • The assumption that any result would merely show the expectations once the latter are well shown in the hierarchical manner. The AHP starts with estimating the priority of the weightings of a set of criteria or options from a square matrix of paired comparison A = [aij ] that will be positive and in case the pairwise comparison is thoroughly consistent it would be mutual which means that aij = 1/ aij for all ij = 1, . . . , n. Accordingly, the ultimate normalized weighting of the item i indicated by wi would be obtained as: wi =

aij n akj

∀i = 1, . . . , n

(41)

k=1

It should be noted that in practice, any error may occur. The proposed eigenvalue technique calculates the weighting as the principal right eigenvalue of A or w meeting the system of n linear equations is presented below: A w = λ max w, in which λmax shows the highest eigenvalue of A giving the following relation:

n

wi =

aij wj

j =1 n

∀i = 1, . . . , n

(42)

akj

k=1

In this regard, the natural measure of inconsistency or deviation from consistency, named “consistency index,” shown by CI would be obtained as follows:

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1503

Table 1 Random index Matrix order RI

1 0

2 0

3 0.58

4 0.90

CI =

5 1.12

λmax − n n−1

6 1.24

7 1.32

8 1.41

9 1.45

10 1.49

(43)

CI of a reciprocal matrix that is generated in a random manner from scale 1–9 with reciprocals forced for every dimension of the matrix, named the random index shown by RI, are included in Table 1. Another concept is the consistency ratio shown by CR and obtained as CR = CI RI in which RI would be a function of the dimension of the matrix, and CR < 0.01 shows a relatively desired bound; else, it is required to be modified and tuned appropriately. Another thing in the hierarchy is forming judgments all across the hierarchy to derive the priorities of the solutions taking into consideration the goal. The weightings would change by adding the priority of the item based on a specific criterion to the weighting of that criterion. More detailed explanations are available in Ref. (Ganoulis 2008). Applications The AHP decision maker has been applied to different energy systems–related problems to choose the best strategy. Accordingly, Ref. (Mastrocinque et al. 2020) employed the AHP decision maker for sustainable supply chain development in the renewable energy sector, mainly focusing on the solar PV panels as the case study. As renewable energies have largely been penetrated into power systems and have brought severe challenges to the system such as power quality issues due to their uncertain power output, storage systems must be used together with renewable energies to damp the oscillations in the power output. To this end, Ref. (Barin et al. 2009) deployed the fuzzy-AHP decision-making technique to choose the best option among the storage systems for operating jointly with the renewable energies. The difficulties in expanding renewable energies penetration in the Nepal’s power grid has been assessed by using the AHP method in Ref. (Ghimire and Kim 2018). The conditions of Korea to be an active player in the hydrogen industry for producing energy to meet the ever-growing load demand has been evaluated in Ref. (Lee et al. 2008) by using the AHP decision maker. A hybrid fuzzy-VIKOR and AHP decisionmaking strategy has been deployed in Ref. (Kaya and Kahraman 2010) for planning renewable energies in Turkey.

4.5.5 ANP By using the ANP, the complicated relationships existing between the elements of decisions would be determined by replacing a hierarchical process with a network process (Kheybari et al. 2020). By taking into consideration the benefits of AHP in the ANP, implementation of this method is straightforward and flexible, while it is associated with the concurrent employment of the quantitative and qualitative

1504

A. Esmaeel Nezhad and P. H. J. Nardelli

criteria and review of consistency in assessments. This technique evaluates each issue through a network of criteria, subcriteria, and objectives providing the mutual interaction between the items in the network. That is to say, it is possible to have feedback and interconnection between clusters within a network. The procedure of the ANP would be briefly stated as below: • Constructing a model and transforming the intended problem to a network design. In this respect, a problem should be converted to a logical system, for example, a network directly. It is noteworthy that the network structure would be derived by brainstorming or other techniques like the nominal group method. Accordingly, an element can interact with other elements. • Modeling a pairwise matrix of comparison and specifying the priority vectors. Like the pairwise comparison made in the AHP, the elements of decision in every cluster would be compared in a pairwise manner. Besides, the clusters would be compared according to their function and impact on attaining the goals and also based on the interdependencies between criteria of every cluster. The eigenvector w can be used to determine the impact of criteria on each other. Furthermore, the Saaty’s nine-point ratio scale would be utilized to specify the relative significance of the elements. In this respect, “1” shows “equally significant” while “9” shows “extremely more significant.” Accordingly, the internal significance vector depicting the relative significance of elements or clusters would be obtained as below: Aw = λmax w

(44)

The matrix of pairwise comparison of criteria is denoted by A and the largest eigenvector is shown by λmax . It is noted that the geometric mean approximation is generally employed to determine w. • Reducing a super matrix and transforming it into a weighted one. In order to attain the total priorities with interactions, the internal significance vectors should be fed to some particular columns of a matrix named “super matrix,” which is indeed a partition matrix depicting how two clusters of a system are related. The super matrix can be stated as below: ⎡

⎤ 0 0 0 wh = ⎣ w21 0 0 ⎦ 0 w32 I

(45)

It is noteworthy that w21 depicts a vector specifying the impacts of the goal on criteria. w32 indicates the impacts of criteria on solutions and I depicts the unit matrix. In case there are interactions between the criteria, the AHP would be

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1505

transformed into the ANP. w21 shows the interactions in the matrix wn , named “primary super matrix,” and an unweighted super matrix would be derived by substituting the internal priorities, elements, and matrices in the primary super matrix. ⎡

⎤ 0 0 0 wh = ⎣ w21 w22 0 ⎦ 0 w32 I

(46)

The weighted super matrix would be calculated by multiplication of the unweighted super matrix values in the cluster matrix in the last stage by utilizing: lim W k

k→∞

(47)

• Choosing the best solution. In this regard, the ultimate weight of studied solutions would be available from the solutions’ column in the limited super matrix. A solution with the maximum weight in this matrix would be picked as the best option. Applications One of the methods widely been applied to energy systems problems is the ANP decision maker. This method has been used in Ref. (Atmaca and Basar 2012) to choose the best generation mix for future expansion of the generation sector in the Turkish power system, and also the suitability of the existing power plants. Different renewable energy sources have been ranked in the Turkish power system to make the system greener (Öztay¸si et al. 2013). To this end, the hybrid fuzzy-ANP method has been utilized to determine the most compromise expansion plans. Ref. (Köne and Büke 2007) has done a comprehensive study on the best generation mix for the future Turkish power system including both nuclear power plants and renewable energies while it has been found that the current generation sector of the Turkish power system is far behind the sustainable development requirement.

5

Conclusion

Most of the problems are decision-making-related problems. Energy systems are not exceptions and in line with the sustainable development goals, the energy systems should adapt to the new conditions. In this regard, real-world decisionmaking problems are indeed MCDM problems that need special attention regarding the efficient method, particularly with uncertain data. In this regard, this chapter presented a comprehensive review on some of the well-established mathematical multi-objective optimization methods such as the weighted sum method, epsilon-

1506

A. Esmaeel Nezhad and P. H. J. Nardelli

constraint technique, and NBI method, as well as some MCDM methods such as the fuzzy satisfying method, VIKOR, TOPSIS, AHP, and ANP, together with their applications to the energy systems.

References A. Afsordegan, M. Sánchez, N. Agell, S. Zahedi, L.V. Cremades, Decision making under uncertainty using a qualitative TOPSIS method for selecting sustainable energy alternatives. Int. J. Environ. Sci. Technol. 13(6), 1419–1432 (2016). https://doi.org/10.1007/s13762-016-0982-7 A. Ahmadi, A. Kaymanesh, P. Siano, M. Janghorbani, A.E. Nezhad, D. Sarno, Evaluating the effectiveness of normal boundary intersection method for short-term environmental/economic hydrothermal self-scheduling. Electr. Power Syst. Res. 123, 192–204 (2015a). https://doi.org/10.1016/j.epsr.2015.02.007 A. Ahmadi, M. Sharafi Masouleh, M. Janghorbani, N. Yadollahi Ghasemi Manjili, A.M. Sharaf, A.E. Nezhad, Short term multi-objective hydrothermal scheduling. Electr. Power Syst. Res. 121, 357–367 (2015b). https://doi.org/10.1016/j.epsr.2014.11.015 A. Ahmadi, H. Moghimi, A.E. Nezhad, V.G. Agelidis, A.M. Sharaf, Multi-objective economic emission dispatch considering combined heat and power by normal boundary intersection method. Electr. Power Syst. Res. 129, 32–43 (2015c). https://doi.org/10.1016/j.epsr.2015.07.011 A. Ahmadi, A.E. Nezhad, P. Siano, B. Hredzak, S. Saha, Information-gap decision theory for robust security-constrained unit commitment of joint renewable energy and Gridable vehicles. IEEE Trans. Ind. Informatics 16(5), 3064–3075 (2020). https://doi.org/10.1109/TII.2019.2908834 H. Ahmadi-Nezamabad, M. Zand, A. Alizadeh, M. Vosoogh, S. Nojavan, Multi-objective optimization based robust scheduling of electric vehicles aggregator. Sustain. Cities Soc. 47, 101494 (2019). https://doi.org/10.1016/j.scs.2019.101494 E. Atmaca, H.B. Basar, Evaluation of power plants in Turkey using Analytic Network Process (ANP). Energy 44(1), 555–563 (2012). https://doi.org/10.1016/j.energy.2012.05.046 A. Barin, L.N. Canha, A. da Rosa Abaide, K.F. Magnago, Selection of storage energy technologies in a power quality scenario – The AHP and the fuzzy logic, in 2009 35th Annual Conference of IEEE Industrial Electronics, (2009), pp. 3615–3620. https://doi.org/10.1109/IECON.2009.5415150 M. Basu, An interactive fuzzy satisfying method based on evolutionary programming technique for multiobjective short-term hydrothermal scheduling. Electr. Power Syst. Res. 69(2), 277–285 (2004). https://doi.org/10.1016/j.epsr.2003.10.003 F. Cavallaro, Fuzzy TOPSIS approach for assessing thermal-energy storage in concentrated solar power (CSP) systems. Appl. Energy 87(2), 496–503 (2010). https://doi.org/10.1016/j.apenergy.2009.07.009 M. Charwand, A. Ahmadi, A.R. Heidari, A.E. Nezhad, Benders decomposition and Normal boundary intersection method for multiobjective decision making framework for an electricity retailer in energy markets. IEEE Syst. J. 9(4), 1475–1484 (2015). https://doi.org/10.1109/JSYST.2014.2331322 T. Chernova, E. Gryazina, Peer-to-peer market with network constraints, user preferences and network charges. Int. J. Electr. Power Energy Syst. 131, 106981 (2021). https://doi.org/10.1016/j.ijepes.2021.106981 T.-J. Ci, K. Kang, S.-T. Wan, Application of E-VIKOR method in power transmission network planning decision. J. North China Electr. Power Univ. (Natural Sci.) Ed, 5 (2012) K. Deb, Multi-objective optimization, in Search Methodologies, (Springer, 2014), pp. 403–449 A. Esmaeel Nezhad, A. Ahmadi, M.S. Javadi, M. Janghorbani, Multi-objective decision-making framework for an electricity retailer in energy markets using lexicographic optimization and augmented epsilon-constraint. Int. Trans. Electr. Energy Syst. 25(12), 3660–3680 (2015). https://doi.org/10.1002/etep.2059

Multiple-Criteria Decision-Making (MCDM) Applications in Optimizing Multi-. . .

1507

S. Fang, Y. Xu, Multi-objective robust energy management for all-electric shipboard microgrid under uncertain wind and wave. Int. J. Electr. Power Energy Syst. 117, 105600 (2020). https://doi.org/10.1016/j.ijepes.2019.105600 J. Ganoulis, Multicriterion Decision Analysis (MCDA) for conflict resolution in sharing groundwater resources, in Overexploitation and Contamination of Shared Groundwater Resources, (Springer, 2008), pp. 375–392 F.S. Gazijahani, A. Ajoulabadi, S.N. Ravadanegh, J. Salehi, Joint energy and reserve scheduling of renewable powered microgrids accommodating price responsive demand by scenario: A risk-based augmented epsilon-constraint approach. J. Clean. Prod. 262, 121365 (2020). https://doi.org/10.1016/j.jclepro.2020.121365 L.P. Ghimire, Y. Kim, An analysis on barriers to renewable energy development in the context of Nepal using AHP. Renew. Energy 129, 446–456 (2018). https://doi.org/10.1016/j.renene.2018.06.011 P. Hajiamoosha, A. Rastgou, S. Bahramara, S.M. Bagher Sadati, Stochastic energy management in a renewable energy-based microgrid considering demand response program. Int. J. Electr. Power Energy Syst. 129, 106791 (2021). https://doi.org/10.1016/j.ijepes.2021.106791 H. Hamidpour, S. Pirouzi, S. Safaee, M. Norouzi, M. Lehtonen, Multi-objective resilientconstrained generation and transmission expansion planning against natural disasters. Int. J. Electr. Power Energy Syst. 132, 107193 (2021). https://doi.org/10.1016/j.ijepes.2021.107193 N. Hongzhan, Y. Ting, Comprehensive evaluation for transmission network planning scheme based on entropy weight method and VIKOR method, in 2011 International Conference on Mechatronic Science, Electric Engineering and Computer (MEC), (2011), pp. 278–281. https://doi.org/10.1109/MEC.2011.6025455 F. Jamalzadeh, A. Hajiseyed Mirzahosseini, F. Faghihi, M. Panahi, Optimal operation of energy hub system using hybrid stochastic-interval optimization approach. Sustain. Cities Soc. 54, 101998 (2020). https://doi.org/10.1016/j.scs.2019.101998 M.S. Javadi, A. Esmaeel Nezhad, Multi-objective, multi-year dynamic generation and transmission expansion planning- renewable energy sources integration for Iran’s National Power Grid. Int. Trans. Electr. Energy Syst.29(4) (2019). https://doi.org/10.1002/etep.2810 M. Javadi et al., A multi-objective model for home energy management system self-scheduling using the Epsilon-Constraint Method, in 2020 IEEE 14th International Conference on Compatibility, Power Electronics and Power Engineering (CPE-POWERENG), vol. 1, (2020), pp. 175–180. https://doi.org/10.1109/CPE-POWERENG48600.2020.9161526 M.S. Javadi et al., Self-scheduling model for home energy management systems considering the end-users discomfort index within price-based demand response programs. Sustain. Cities Soc. 68, 102792 (2021). https://doi.org/10.1016/j.scs.2021.102792 F. Katal, F. Fazelpour, Multi-criteria evaluation and priority analysis of different types of existing power plants in Iran: An optimized energy planning system. Renew. Energy 120, 163–177 (2018). https://doi.org/10.1016/j.renene.2017.12.061 T. Kaya, C. Kahraman, Multicriteria renewable energy planning using an integrated fuzzy VIKOR & AHP methodology: The case of Istanbul. Energy 35(6), 2517–2527 (2010). https://doi.org/10.1016/j.energy.2010.02.051 T. Kaya, C. Kahraman, Multicriteria decision making in energy planning using a modified fuzzy TOPSIS methodology. Expert Syst. Appl. 38(6), 6577–6585 (2011). https://doi.org/10.1016/j.eswa.2010.11.081 S. Kheybari, F.M. Rezaie, H. Farazmand, Analytic network process: An overview of applications. Appl. Math. Comput. 367, 124780 (2020). https://doi.org/10.1016/j.amc.2019.124780 A.Ç. Köne, T. Büke, An Analytical Network Process (ANP) evaluation of alternative fuels for electricity generation in Turkey. Energy Policy 35(10), 5220–5228 (2007). https://doi.org/10.1016/j.enpol.2007.05.014 S.K. Lee, G. Mogi, J.W. Kim, The competitiveness of Korea as a developer of hydrogen energy technology: The AHP approach. Energy Policy 36(4), 1284–1291 (2008). https://doi.org/10.1016/j.enpol.2007.12.003

1508

A. Esmaeel Nezhad and P. H. J. Nardelli

M. Majidi, S. Nojavan, N. Nourani Esfetanaj, A. Najafi-Ghalelou, K. Zare, A multi-objective model for optimal operation of a battery/PV/fuel cell/grid hybrid energy system using weighted sum technique and fuzzy satisfying approach considering responsible load management. Sol. Energy 144, 79–89 (2017). https://doi.org/10.1016/j.solener.2017.01.009 S.A. Mansouri, A. Ahmarinejad, E. Nematbakhsh, M.S. Javadi, A.R. Jordehi, J.P.S. Catalão, Energy management in microgrids including smart homes: A multi-objective approach. Sustain. Cities Soc. 69, 102852 (2021). https://doi.org/10.1016/j.scs.2021.102852 E. Mastrocinque, F.J. Ramírez, A. Honrubia-Escribano, D.T. Pham, An AHP-based multi-criteria model for sustainable supply chain development in the renewable energy sector. Expert Syst. Appl. 150, 113321 (2020). https://doi.org/10.1016/j.eswa.2020.113321 H. Mavalizadeh, A. Ahmadi, F.H. Gandoman, P. Siano, H.A. Shayanfar, Multiobjective robust power system expansion planning considering generation units retirement. IEEE Syst. J. 12(3), 2664–2675 (2018). https://doi.org/10.1109/JSYST.2017.2672694 M.A. Nazari, A. Aslani, R. Ghasempour, Analysis of solar farm site selection based on TOPSIS approach. Int. J. Soc. Ecol. Sustain. Dev. 9(1), 12–25 (2018) A.E. Nezhad, M.S. Javadi, E. Rahimi, Applying augmented ε-constraint approach and lexicographic optimization to solve multi-objective hydrothermal generation scheduling considering the impacts of pumped-storage units. Int. J. Electr. Power Energy Syst. 55, 195–204 (2014). https://doi.org/10.1016/j.ijepes.2013.09.006 S. Nojavan, M. Majidi, A. Najafi-Ghalelou, M. Ghahramani, K. Zare, A cost-emission model for fuel cell/PV/battery hybrid energy system in the presence of demand response program: ε-constraint method and fuzzy satisfying approach. Energy Convers. Manag. 138, 383–392 (2017). https://doi.org/10.1016/j.enconman.2017.02.003 B. Öztay¸si, S. U˘gurlu, C. Kahraman, Assessment of green energy alternatives using fuzzy ANP, in Assessment and Simulation Tools for Sustainable Energy Systems, (Springer, 2013), pp. 55–77 P. Pourghasem, F. Sohrabi, M. Abapour, B. Mohammadi-Ivatloo, Stochastic multi-objective dynamic dispatch of renewable and CHP-based islanded microgrids. Electr. Power Syst. Res. 173, 193–201 (2019). https://doi.org/10.1016/j.epsr.2019.04.021 P. Rani, A.R. Mishra, A. Mardani, F. Cavallaro, M. Alrasheedi, A. Alrashidi, A novel approach to extended fuzzy TOPSIS based on new divergence measures for renewable energy sources selection. J. Clean. Prod. 257, 120352 (2020). https://doi.org/10.1016/j.jclepro.2020.120352 M. Reza Norouzi, A. Ahmadi, A. Esmaeel Nezhad, A. Ghaedi, Mixed integer programming of multi-objective security-constrained hydro/thermal unit commitment. Renew. Sust. Energ. Rev. 29, 911–923 (2014). https://doi.org/10.1016/j.rser.2013.09.020 M. Sadegh Javadi, M. Saniei, H. Rajabi Mashhadi, G. Gutiérrez-Alcaraz, Multi-objective expansion planning approach: Distant wind farms and limited energy resources integration. IET Renew. Power Gener. 7(6), 652–668 (2013). https://doi.org/10.1049/iet-rpg.2012.0218 J.R. San Cristóbal, Multi-criteria decision-making in the selection of a renewable energy project in Spain: The Vikor method. Renew. Energy 36(2), 498–502 (2011). https://doi.org/10.1016/j.renene.2010.07.031 Ü. Sengül, ¸ M. Eren, S. Eslamian Shiraz, V. Gezder, A.B. Sengül, ¸ Fuzzy TOPSIS method for ranking renewable energy supply systems in Turkey. Renew. Energy 75, 617–625 (2015). https://doi.org/10.1016/j.renene.2014.10.045 M. Simab, M.S. Javadi, A.E. Nezhad, Multi-objective programming of pumped-hydro-thermal scheduling problem using normal boundary intersection and VIKOR. Energy 143, 854–866 (2018). https://doi.org/10.1016/j.energy.2017.09.144 R.P. Singh, H.P. Nachtnebel, Analytical hierarchy process (AHP) application for reinforcement of hydropower strategy in Nepal. Renew. Sust. Energ. Rev. 55, 43–58 (2016). https://doi.org/10.1016/j.rser.2015.10.138 Y. Wu, K. Chen, B. Zeng, H. Xu, Y. Yang, Supplier selection in nuclear power industry with extended VIKOR method under linguistic information. Appl. Soft Comput. 48, 444–457 (2016). https://doi.org/10.1016/j.asoc.2016.07.023 X. Yang et al., Multi-objective optimal scheduling for CCHP microgrids considering peakload reduction by augmented ε-constraint method. Renew. Energy 172, 408–423 (2021). https://doi.org/10.1016/j.renene.2021.02.165

Simulation Applications in Analyzing the Trade-Off Between Climate Change and Energy Consumption Amin Vahidi

Contents 1 2 3 4 5

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Simulation Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Global Warming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biodiversity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fossil Fuel Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Urbanization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Agricultural and Livestock Products Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Industrial Products Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Energy Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Deforestation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1 Public Awareness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Policies, Laws, Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1510 1511 1514 1516 1518 1518 1519 1519 1521 1524 1525 1528 1528 1530 1531 1533

Abstract

The United Nations estimates that 90% of the worst disasters occurred as a result of Climate Change. From an ecological point of view, Global Warming can cause far-reaching changes in the environment, biodiversity and human life. Today, scientists and experts in climatology and environmental sciences believe that one of the causes of global Climate Change in recent years has been the increase in Greenhouse Gases. Human emissions of carbon dioxide come from two main sources: burning fossil fuels and changing land use, such

A. Vahidi () Industrial Engineering, Shahid Beheshti University, Facility of Mechanics and Energy, Tehran, Iran © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_53

1509

1510

A. Vahidi

as deforestation. In this research System Dynamics methodology used. System Dynamics method is based on concept of feedback loops. If the loop amplify itself, it is a positive feedback or reinforcing loop. And if increasing in a variable cause itself to decrease so it is a negative or balancing feedback loop. The overall System Dynamics stock and flow model presented based on these concepts. Time horizon of this model simulation of this model is from 2020 to 2100. Simulation if current situation continues in 2050 and 2100, Global Warming will be 2.38 and 5.56 than 2020. It is catastrophic and make earth hardly habitable. Also the simulation shows in 2100 Fossil Fuel Consumption will be 2.1 times of 2020 in an exponential pattern. In 2100 vegetation decreased from 55% to 33%. So animals and plants will extinct with extremely high speed and earth will be a lifeless planet with current trend. So if Global Warming continues, it only take a decade to reach an irreversible point. If human race improve 30% in Public Awareness about Global Warming effects and Policies, Laws, Guidelines effectiveness on deforestation and forest growth, Global Warming will decrease to 41% in 2050 and 37% in 2100. So a global effort needed to save the earth and humankind. Keywords

Climate Change · Global Warming · Fossil Fuel Consumption · Greenhouse Gases · System Dynamics

1

Introduction

The growing aggravation of environmental problems such as Global Warming, depletion of water resources, and the destruction of biodiversity in the next century, has become one of the most important concerns of many countries and international organizations. These crises have spread to such an extent that there is no other way for human beings but to prevent the severity of these changes. If the amount of Greenhouse Gases in the atmosphere increases in the same way, the result will be much more than the extinction of species and rising sea levels. On the eve of the third millennium, the protection of the environment as the common ground of human rationality, thought and action to achieve sustainable development and a bright future in which the rights of future generations are guaranteed, is the most important duty and responsibility of governments. The slogan for World Environment Day, “Global Warming and melting ice,” shows the importance of paying attention to the fact that it is a global threat to all lands; because Climate Change affects all aspects of human life. What is called Global Warming these days is actually an increase in the average temperature of the earth near its surface (Buchwitz et al. 2015). The International Panel on Climate Change (IPCC), an authoritative body on Climate Change and effects of Global Warming, said in a report: “Most of the Global Warming since the middle of the twentieth century is due to Greenhouse Gases

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1511

Humans have produced.” The panel did not rule out the role of natural factors such as solar volcanoes, but said the effects could be traced back to the early 1950s, when the world was not yet industrialized. About 20% of the world’s Greenhouse Gases come from deforestation, as cleared forest land is often used for grazing livestock, raising animal feed or providing biofuels (Vetter et al. 2019). Global Warming has many unpleasant effects on the lives of humans and animals. As the earth warms, the polar ice caps melt, the sea level rises, and the seasons intensify. This means that winters will be colder than ever and summers will be warmer and drier. This has an adverse effect on agriculture, which is one of the key productive activities on Earth. Climate Change is increasing the average temperature of the earth, thus prolonging the growing season. In areas where summer heat is high under normal conditions and harmful to crop growth, rising temperatures make the situation even more difficult. As the temperature rises, the probability of drought increases and the rate of evaporation of soil moisture increases (Hughes et al. 2021). According to the Office of the National Climate Change Plan, the projected increase in temperature will reduce the fertility of rice seedlings, reduce the shelf life of corn, ripen wheat, and reduce the germination of potatoes. On the other hand, Global Warming changes the amount of rainfall in different regions. Changes in the amount and timing of precipitation affect soil erosion and moisture; Factors that are both very important in agriculture. It is predicted that if the heating trend increases in the same way, the amount of precipitation will increase in the northern offerings and decrease in the areas near the equator (Costantino et al. 2021). Increasing atmospheric carbon dioxide may make crops such as wheat and rice more fertile; On the other hand, water constraints and rising temperatures do not allow carbon dioxide to increase yields. Also, the increase of other pollutants, such as ozone in the atmospheric layers near the earth’s surface, which is caused by Global Warming, impairs the growth of agricultural products (Sovacool et al. 2021). The rapid emergence of Greenhouse Gases in the early Jurassic period (about 180 million years ago) also caused the average earth temperature to rise between 5 ◦ C and 9 ◦ C.

2

Simulation Methodology

It is suggested to use the System Dynamics model for holistic review of plans and decisions. The resulting qualitative and quantitative analysis can be a great practice guide for discovering and analyzing the complications of a system (Vahidi 2018). System Dynamics (SD) is a method of system modeling and policy analysis based on feedback systems theory. In the late 1950s, Jay Forester, a pioneer in computer engineering and design, coined the concept (Forrester 1994). Since then, System Dynamics has developed as a separate field from disciplines such as operations research and management science. The computer easily simulates these models due to their numerical nature (Vahidi et al. 2019).

1512

A. Vahidi

System Dynamics combines the behavioral and social sciences. It requires careful design and construction of innovative models with many interactive variables. Although the results and arguments of a good model will not be too vague and complex for decision makers (Sterman 2001). On the other hand, although the dynamic model of the system is an advanced model, it is still compressible enough to run on a laptop. This feature allows all hypotheses and scenarios to be tested quickly and thoroughly in interactive strategy development sessions. System Dynamics is used in systems that face complex and important decisions that seek to integrate insights into fundamental problems that may affect the results of years or decades to come. System Dynamics help these systems assess the pros and cons of policies they have considered or may consider later (Vahidi et al. 2018). An integrated and strategic vision is needed when different policies have multiple short- and long-term consequences (including the reaction of all affected parts and the reaction of those reactions). In such cases, routine planning methods (such as the use of spreadsheets, statistical or even regression models) will not work because they cannot predict possible embedded feedback loops behavior (Hayward 1961). In a dynamic model, one policy may seem like a good policy at first glance, but in the real world, it faces obstacles or resistances that counteract its effects. It may even make things worse instead of helping to solve the problem. Also, the policy that seems weak or expensive at first may work well in the end, and its limitations and disadvantages will be lessened because of its advantages. System Dynamics models focus not only on primary effects but also on secondary effects and feedback effects, thereby improving our ability to choose intelligent policies (which remain stable over time). System Dynamics models are custom built to solve a specific problem. The problem to be modeled must be precisely defined. The model designer must work with the client at all stages to determine the main actions and outcome variables. Determining the level of detail of actions and variables is also very important. A model is useful when it considers all the important variables and excludes unnecessary sub-variables. This will take into account all available and possible decision-making policies (even those that are unusual but worth considering) and eliminate ineffective policies as a result (Fig. 1). The concept of feedback is one of the most important parts of the System Dynamics method. Feedback loop diagrams and causal loops are tools for conceptualizing the structure of a complex system and linking model-based findings. A feedback loop occurs when information from an action or action circulates across a Fig. 1 System Dynamics loop sample

Global Population. Pop Growth Rate

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1513

system and eventually returns to the point of origin. If the loop tends to amplify the initial action, it is called a positive feedback loop or reinforcing loop (Vahidi and Aliahmadi 2019). In this case, if a variable increases (decreases), it will increase (decrease) again after a delay. If the loop tends to oppose the initial action, the loop is called a negative or balancing feedback loop. In this case, if the variable decreases, it will increase, and if it increases, it will decrease. The sign of the ring is called the “polarization of the ring” (Forrester 1995). No model can accurately and completely predict the future. However, some models are more reliable than others. Dynamic modeling of the system requires knowledge, art, and effort, and the slightest mistake or slip in it can cause problematic results. In designing a good model, on the one hand, historical data for the outcome variables should be followed as much as possible, and on the other hand, appropriate numerical evaluation should be done for the model input hypotheses. Also it could use fuzzy numbers (Vahidi 2019). A good model should be comprehensive and full of convincing evidence. In the System Dynamics model, after selecting the variables, we have to consider some hypotheses and proceed to the modeling process based on them. Not all hypotheses need to be complete and conclusive. This uncertainty of the hypotheses shows how important it is to test the sensitivity and validity of the hypotheses. The purpose of these experiments is to see if a change in the model hypotheses will affect the results of the relative effects of strategic decisions. In fact, the nature of the System Dynamics is such that the inaccuracy of the hypotheses does not matter much, because these hypotheses will be modified during the process. Here are some things to look for when selecting yours. • Search for additional data and information that can compensate for the uncertainty of the hypotheses. • Design the model with more details of the parameters related to the problem or issue. • Identify which parameters of the problem need to be further investigated to get a more definitive answer to the problem. The System Dynamics simulation includes the following steps: • Define the problem dynamically, based on graphs whose horizontal axis is the time axis. • Trying to achieve an endogenous behavioral insight into the dynamics of a system and focusing on the characteristics of the system. • Thinking about all the concepts in the real system as continuous values that are related to each other in the feedback loops and the causality loop. • Identify the levels of stocks in the system and the rate of their input and output flows. • Develop a behavioral model that has the ability to reproduce (reproduce). This model is a computer simulation that is usually expressed as nonlinear equations,

1514

A. Vahidi

but is sometimes expressed as a diagram based on the accumulation feedback structure and the flow / causal feedback structure of the system, rather than being quantified. • Understanding the concepts and executive policies of the resulting model. • Implement the changes derived from the findings and model understanding.

3

Global Warming

The amount of CO2 in the air is a measure of Greenhouse Gases. For example, before the industrialization of the world, the amount of CO2 in the air was 260– 280 PPM, but after industrialization this amount increased and now it has reached 360 PPM. 2005 is said to be the warmest year in the last century; this year, the amount of CO2 in the air reached 381 PPM. A group of French experts believe that the fight against rising temperatures should not be limited to reducing CO2 because methane, another gas whose greenhouse effect is very important in the short term, has been ignored. The experts also noted that reducing CO2 alone could not reduce global temperatures by 2◦ by 2050 (Nong et al. 2021). NASA scientists have found a large combination of global air pollution, including thick fog from smoke and summer chemical vapors that play a major role in Global Warming. In a global assessment, they calculated the effect of ozone on Global Warming, how ozone has changed temperatures in the lowest part of the atmosphere over the past 100 years. Using the best estimates available from the global emissions of ozone-producing gases, GISS computer sample research reveals how much of this air and greenhouse gas pollutant has contributed to heating certain areas of the world. The new results highlight the unexpected impact of global air pollution control effects. According to researchers, we now know that reducing ozone pollution can not only improve air health, but also facilitate Global Warming, especially in the Arctic. Global Warming affects all living and non-living things on Earth: Global Warming increases the intensity of evaporation and, consequently, the need for water for agricultural products. On the other hand, the amount of water for agricultural use is reduced and as a result of drought, food security is endangered. Drought causes villagers to migrate to cities, to be marginalized, to turn to false jobs, and to increase social anomalies (Kuramochi et al. 2021). Rising global temperatures are also affecting access to fresh water and drinking water. On the other hand, due to rising sea levels, saline water infiltrates coastal freshwater resources and changes its quality. Another problem is that surface evaporation due to Global Warming causes rivers to dry up and water quality to decline. This is especially evident in less watery areas such as deserts and semideserts. In addition, the increase in freshwater salts due to evaporation reduces its quality.

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1515

Rising global temperatures are also causing rapid melting of polar ice caps and floating icebergs, resulting in rising sea levels. According to the forecasts of international organizations in the next 100 years, the water level of the high seas will increase between 30 cm and 1.5 m, which in turn will cause the disappearance or depopulation of some islands, the advancement of water in coastal areas and Affects facilities and people’s lives in ports. Another devastating consequence of Global Warming is the increase in the number and strength of hurricanes and tornadoes that we are currently facing. The melting of the polar ice caps creates a stream of pure cold water that, due to being cold, slides under the warm waters of the oceans, creating large-scale ocean currents that create more storms, hurricanes, and more powerful hurricanes. As Global Warming increases, so is the likelihood of volcanoes erupting. Rising temperatures on the Earth’s surface in recent years have undergone far more profound changes than the melting of the polar ice caps, eventually leading to the melting of the earth’s inner layers, which emerge from volcanic mountains. With the disappearance of icebergs, the pressure exerted on the rocks beneath the ice sheets will be greatly reduced, resulting in the melting of the earth’s inner rocks and the formation of magma. As the ice melts, more magma will form inside the earth, which could set the stage for volcanic eruptions that could affect the entire world. As the pressure on the earth’s crust changes, so does the pressure of the geological layers in the earth’s crust. As a result, an explosion is likely to occur, as the weight of the ice sheets acts as a barrier to the melting of the Earth’s inner material and has protected the Earth from volcanic eruptions for years. Another devastating consequence of rising temperatures is the spread of tropical diseases, some fevers, and viral diseases. The gradual warming of the earth also increases the birth rate of rodents such as mice and, consequently, the spread of diseases transmitted by them. According to international organizations, the surface of the high seas will rise by 30–150 cm in the next 100 years due to the phenomenon of “Global Warming”. Surface evaporation caused by the “Global Warming” phenomenon causes rivers to dry up and water quality to decline (Bogaerts et al. 2017). Ozone (O3) is the earth’s protective layer against the sun’s harmful rays. Around 1970, cavities such as CFCs were used in the compressors of refrigerators and some other refrigerants to create cavities in this shield. Most of these materials are used in the production of foaming agents, solvents, cooling systems, and insulation. Due to the significant effect of these substances on ozone depletion, a global agreement (the Montreal Protocol) has been reached to restrict the use of these substances. Hydrofluorocarbons (HFCs) and perfluorocarbons (PFCs) have now replaced these materials, which are not ozone depleting but still have greenhouse effects. Therefore, research is currently underway to replace materials that do not increase the greenhouse effect and are compatible with the ozone layer. Human emissions of carbon dioxide come from two main sources: burning fossil fuels and changing land use, such as deforestation. Scientists’ research shows that enough solar energy reaches the earth’s surface. The energy received every 40 min can provide 100% of the total energy needed by the world for a year. Trapping only a

1516

A. Vahidi

Fig. 2 Global Warming causes tree

Fig. 3 Global Warming effects tree

small part of this solar energy can meet the entire electricity needs of a large country. On the other hand, we can also mention wind energy and geothermal energy. In our country, we have a lot of wind energy in the east and west, which can meet the needs of the cities in this region. In the following Figures, Global Warming Causes and Effects are presented in tree and loops (Figs. 2, 3, and 4): As you can see Global Warming will reproduce and amplify itself in an increasing behavior due to above positive feedback loop.

4

Biodiversity

Decreased biodiversity and adverse effects on plant and animal species are other consequences of Global Warming. Animal migration and vegetation change due to drought and water scarcity cause changes in the food chain and adverse effects on

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . . Fig. 4 Global Warming loop

Global warming Rate

Global warming.

greenhouse gas

Vegetaon

1517

Wildﬁres

deforestaon

the ecosystem of the region. This problem also leads to changes in the biodiversity of the aquatic ecosystem in the seas. A clear and tangible example of this is the whitening of corals on the shores of the Persian Gulf. According to scientists, corals die earlier and turn white due to the average increase in water temperature. Due to the importance of Climate Change, various studies on bird life have been conducted and interesting results have been obtained. In general, Climate Change has direct and indirect effects on the lives of birds, which can severely disrupt their life cycle. For example, as the weather warms, birds migrate to more northerly shores and migrate there. In this way, the birds travel longer distances than in the past during their migration, which can lead to the extinction of weaker species. An interesting study published by the University of Michigan in the United States shows that as the concentration of carbon dioxide in the atmosphere doubled and as a result of more Global Warming, the number of birds in certain areas decreased significantly and was not found at all in some areas. Different birds sometimes lay eggs about a week or two earlier, which makes the birds about a week or two older when migrating. In such cases, when returning or during the journey, the probability of extinction of birds increases and their population decreases day by day (Ramanathan and Feng 2009). On the other hand, due to Global Warming, the time of flowering and growth of plant species has changed compared to the past. Many animal and plant species are now extinct due to climatic effects and many species of animals are forced to migrate. As a result, the entire planet’s ecosystem is falling apart. More than half of all living things have been extinct several times in Earth history, but it has taken hundreds of thousands of years to recover (Fig. 5).

1518

A. Vahidi

Fig. 5 Extinction of plants and animals causes tree

5

Fossil Fuel Consumption

As explained before many variables cause Global Warming and Climate Change, but the main variables are as follows:

5.1

Urbanization

From ancient times to the present day, villagers have migrated to cities for various purposes, and this event has accelerated in the last two decades and has led to the phenomenon of widespread urbanization. It is obvious that the increase in urban population will lead to a peak in energy consumption, which can lead to the decline of natural resources. Urbanization will increase the demand for industrial goods (especially plastic). On the other hand, on average, per capita food consumption is higher in urban communities. These effects will increase energy demand, greenhouse gas emissions and Global Warming (Righini et al. 2020). Increasing urban population can also lead to challenges such as increasing air pollution, widening the income gap, and the deterioration of urban infrastructure. The most important action that managers and citizens can take to prevent resources from running out is to take approaches to stabilize as many cities as possible. What

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1519

we are looking at here is to examine the effect of urbanization on Global Warming. It should be noted that the current trend of increasing urbanization is 0.3% per year.

5.2

Agricultural and Livestock Products Demand

A group of scientists, and in particular a number of environmental NGOs, have pointed the finger at the agricultural and livestock industries as one of the main causes of Global Warming (Rajkhowa and Sarma 2021). According to the Food and Agriculture Organization of the United Nations (FAO), the livestock and animal husbandry industry ranks first in environmental threats. The report, entitled Livestock’s Long Shadow, states that the livestock industry has degraded land, polluted water resources, destroyed species diversity, and affected the greenhouse effect (Juniper 2021). Livestock activities lead to the production of Greenhouse Gases in two ways, direct and indirect. Consumption of fossil fuels and the production of carbon dioxide gas, on the one hand, and gases produced during the production of animal feed or in the process of their digestion – which are released from the intestine and mainly through feces – on the other hand, it has challenged environmental activists and advocates (Pfadt-Trilling and Fortier 2021). The most important greenhouse gas produced during animal feed is nitrogen dioxide. Wheat and soybean production produce large amounts of nitrogen dioxide, with some scientists claiming that about 65% of atmospheric nitrogen dioxide comes from such activities (Edmonds et al. 2003). The next pollutant is methane. The destructive effects of methane on the Global Warming process are 23 times greater than those of carbon dioxide. Thirty-seven percent of the methane in the atmosphere is produced by animal husbandry. In India alone, livestock industries produce 11.75 million tons of methane a year. UK Food, Environment and Agricultural Business Organization (Defra) examine the ground and work with Cranfield University on the environmental role of resources, equipment, and methods used in various branches of agriculture and animal husbandry and the impact of each on heating. The studies focused on 10 major activities including beef production, pork production, mutton, poultry, eggs, and milk production. According to an indicator called Global Warming potential in 100 years (GWP100), poultry and egg production have the least impact and beef production has the most impact on Global Warming. However, poultry-breeding ammonia makes up a large portion of the ammonia in acid rain (64%).

5.3

Industrial Products Demand

All sectors, from transportation and agriculture to industry, contribute to greenhouse gas emissions, but it must be acknowledged that industry has a great share in air pollution and greenhouse gas emissions (Bollen and Brink 2014).

1520

A. Vahidi

To prove this claim, we can refer to the daily human relationship with various types of industrial products. Industry, with its vast range of products, emits Greenhouse Gases in two direct and indirect ways. Meanwhile, the first part is created due to the existence of production facilities and the second part is created by supplying energy to produce products. In the discussion of the direct emission of Greenhouse Gases, we can refer to the emission of gases produced by fossil fuels. Fossil fuels, including gas and oil, which are mostly used in industries such as the production of chemicals, iron, steel, and cement, generate more Greenhouse Gases in industrial facilities than other fuels. But if we look at the issue of indirect greenhouse gas emissions, we come to power plants, which provide the energy needed to turn production lines and industrial machinery; this is because the energy needed to activate power plants is usually provided by fossil fuels (Betz et al. 2015). In this regard, the US Environmental Protection Agency (IPIA) in a report examine the share of various industries in the production of Greenhouse Gases. According to the report, the total Greenhouse Gases produced in the United States are estimated at 6637 million tons of carbon dioxide, while the industry with 21% of the total Greenhouse Gases in this country, ranks first in gas production and If we consider the production and use of electricity as effective in generating Greenhouse Gases, the share of industry in total Greenhouse Gases this year will increase to a Figure between 29% and 31%. It was the only industrialized country in the world that has not ratified the Kyoto Protocol because US officials believe that reducing greenhouse gas emissions under the Kyoto Protocol will lead to a slowdown in economic growth. After the United States, China is the next largest producer of Greenhouse Gases with an annual production of 6018 million tons of Greenhouse Gases, but given the amount of production and activity of various industries in China, the production of this amount of polluting gases is not surprising. With a population of over one billion, it is the world’s largest population and uses a huge amount of fossil fuels in cities, the food industry, and the transportation industry (Tuckett 2021). Various industries are increasingly involved in the production of Greenhouse Gases. Oil and gas production, which produces gases such as carbon dioxide and methane, accounts for 3.6% of the world’s greenhouse gas emissions. The cement industry also emits 4% of the world’s Greenhouse Gases by emitting carbon dioxide, because the demand for cement is very high due to the boom in construction worldwide. Cement production, which involves extracting limestone and then processing it at very high temperatures, requires a lot of energy (Tokioka 1995). In addition, the aviation industry accounts for 3.5% of the world’s greenhouse gas emissions through the production of carbon dioxide, water vapor, nitrous oxide and particulate matter. According to the Intergovernmental Panel on Climate Change (IPC), the emissions of water vapor, nitrogen oxides, particulate matter, and carbon dioxide from aircraft are 2–4 times higher than carbon dioxide emissions alone. The iron and steel industry also emits 2.3% of the world’s Greenhouse Gases by producing carbon dioxide. The iron and steel industry has one of the largest carbon footprints of any single industrial sector (Dunne et al. 2013). The sheer size of the industry and the processes of mining, transportation, smelting, and production that

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1521

require tremendous energy are key factors in this regard. Waste also has a 3% share in greenhouse gas emissions by producing carbon dioxide and methane; Landfills are considered sources of methane production due to the corruption of various materials. If waste is incinerated, it also releases carbon dioxide. Today agricultural products waste is about 45–50% and animal (protein) products waste is about 30%. The definition of the Kyoto Protocol (Kyoto Protocol) states that Kyoto is an international treaty signed in 1997 to reduce greenhouse gas emissions between countries around the world (Fattorini 2021). The agreement, designed to complement previous ones, was intended to allow industrialized nations to reduce their greenhouse gas emissions by 5% by 10 years and to provide financial assistance to developing countries to increase the penetration of renewable energy sources such as solar energy (Dutta 2020). The agreement was signed with the aim of obliging the countries of the world to reduce the effects of Greenhouse Gases and the negative consequences of Global Warming. The treaty was drafted in Kyoto, Japan, in December 1997 and submitted to various countries for signature in March 1998 and was finalized on March 15, 1999. The agreement specifies how many years and by what percentage greenhouse gas emissions will be reduced. The gases covered by this treaty are: carbon dioxide, methane, nitrous oxide and sulfur hexafluoride. Following the pact, 10,000 delegates from 195 countries gathered at a meeting in Paris on November 30 to jointly fight Climate Change and Global Warming. The main part of the meeting will focus on the implementation of an alternative to the Kyoto Protocol to reduce greenhouse gas emissions. The treaty should reduce the world’s Fossil Fuel Consumption to such an extent that Global Warming does not rise by more than 2◦ by the end of the current century since the beginning of global industrialization. For the first time, the pact binds industrialized as well as developing countries to reduce carbon dioxide (fossil fuel gas) emissions, and is expected to replace the Kyoto Protocol by 2020. This time, the leaders of the world, especially the industrialized countries, all believe that they must find a common solution to combat Global Warming. For example, the United States, China, Germany, and Russia, the world’s largest emitters of Greenhouse Gases, can play an important role in global efforts to reduce emissions. Today 14% of total energy is renewable (alternative) energy and it’s increasing 1.1% per year.

5.4

Energy Efficiency

Energy efficiency means using less energy for services. For example, a fluorescent lamp is more efficient than traditional incandescent bulbs. Because it consumes less electricity than producing the same amount of light. Similarly, an efficient boiler uses less fuel to heat a home at a certain temperature than a less fuel efficient boiler. The term energy efficiency is often a term used to describe any type of energy storage. Technically, though, it should be distinct from energy savings – a broader term that can include regardless of whether a device changes the efficiency of whatever it produces. Examples of energy savings include lowering the thermostat in the winter or walking instead of driving to a point.

1522

A. Vahidi

Increasing energy efficiency often comes at a cost, but in most cases this cost will be recouped by reducing energy costs in the short term. This makes productivity development an attractive starting point for reducing carbon emissions. The scope of storage and the techniques required depend on the situation. For homes in cold countries such as the UK, the most effective measures are to increase insulation, install good quality double glazed windows, and use more efficient tools and energysaving lamps and LEDs. The Climate Change Committee estimates that the improvement could reduce annual carbon dioxide emissions from British homes by about 17 million tons by 2020 (Dunne et al. 2013). In contrast, increasing productivity in non-residential buildings often means focusing on ventilation and air conditioning, in addition to focusing on light, heat, and appliances. Many of these buildings gained about 25% of their energy savings after doing so for energy efficiency. High-energy industries, such as the iron, steel, and cement industries, have become more efficient over time due to the use of new equipment and the reuse of heat lost over time. For example, a tube that is exposed to heat and contains chemicals that need to be cooled can be used to heat other chemicals. Engines are widely used in industry for a variety of tasks such as pumping, mixing, and moving conveyors. Installing the right and efficient size of motors can help save up to 25% of energy. Energy efficiency in cars has also increased in recent decades. This increase has been achieved by modifying factors such as engines, headlights, and better aerodynamic design. There is potential for future development, and in the EU, the average emission of new cars is targeted to be reduced from 150 to 95 grams of carbon dioxide per kilometer by 2020. This increase in energy efficiency in cars and vans could reduce carbon dioxide emissions in the UK by 12.3 million tons by 2020, which is about 10% of total surface transport in 2008. Improving energy efficiency does not necessarily mean changes to reduce crane emissions. If the energy comes from fossil fuels, such as gasoline in cars or electricity from coal-fired power plants, it will cut off emission efficiency improvements. But if energy from a low-carbon source, such as generating electricity from nuclear or renewable power plants, improving efficiency may have little effect on emissions. When comparing electrical and non-electrical appliances, it is important to consider energy efficiency: a change from 90% gas boiler efficiency to 100 electric heater efficiency, if electricity is supplied from fossil fuel power plants that are heavily self-sustaining. They are inefficient, will increase energy consumption and emissions. Energy efficiency is always a good idea. Whether or not it leads to energy savings depends on what we do with the money we save. In some cases, productivity savings can be offset by a change in user behavior, called a reflection effect. An example

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1523

would be that insulating a house might make it more economical for residents to maintain a higher temperature. Nevertheless, improving energy efficiency is a key tool to reduce carbon dioxide emissions. Comes with energy saving and low carbon energy sources such as renewable energy. Today total Energy Efficiency is about 65% (Figs. 6 and 7).

Fig. 6 Goods Demand Causes Tree

Fig. 7 Goods Demand Effects Tree

1524

A. Vahidi

As you can see the main cause of Goods demand is increasing population and urbanization.

6

Deforestation

In addition to factors such as increased demand for plant and livestock products, which were mentioned in detail above, widespread fires are one of the causes of deforestation in the world. The effects of Global Warming on air temperature, rainfall, and soil moisture have engulfed many of the planet’s forests over the past decades. Humidity and rainfall change as climates warm; in this way, the wet areas are getting wetter day by day and the dry areas are getting drier day by day. Higher temperatures in the spring and summer and the early melting of snow in the spring cause the soil to stay dry longer, resulting in longer periods of fire risk in pastures and forests, especially in areas such as the western United States, Central Africa, and Australia are growing dramatically. These hot and dry climates make forests very vulnerable to factors such as lightning and human error; in such a way that the slightest spark can cause a great catastrophe. The first serious forest fire alarm was sounded in the mid-1980s; Because from this date onwards, the number of fires as well as their duration have increased, and as a result, the volume of burned forests has increased. According to a statistical study conducted in the western forests of the United States of America, the number of fires and their duration lasted from 1986 to 2003 compared to 1970 and 1986, respectively, 6 and 5 times. Periods of natural drought, volcanic eruptions, human activities, such as land use change, and human impact on Climate Change are some of the factors that can increase the likelihood of fires. Two examples that confirm the impact of Climate Change on fire are Yosemite National Park and the northern part of the Rocky Mountains, where land use change has been prevented and therefore the main cause of the increase in fires in these two areas is likely Climate Change. Natural phenomena such as Hurricane El Nino can also affect changes in drought and fire potential in the area by affecting air humidity and rainfall. However, since long-term meteorological forecasts show significant changes in temperature and rainfall during the twenty-first century, we seem to potentially experience more fires during the current century (Tuckett 2021). The period of the year when there is a possibility of pasture and forest fires is called the fire season. This season varies between 6 and 8 months in hot and dry areas. Researchers believe that with increasing Climate Change and Global Warming, we will probably see an increase in the duration of the fire season to 12 months (the whole year) in many parts of the world! It should be noted that in such circumstances, the probability of fire due to factors such as human error and lightning will increase more than before.

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1525

Fig. 8 Deforestation causes tree

Fig. 9 Deforestation effects tree

Also, in the coming decades, the forests of temperate and humid regions will become warmer, and with the loss of moisture, they will become relatively arid regions, which as a result of these changes will increase the possibility of fires in such forests, which were previously low. Surprisingly, grasslands and pastures in arid areas may be less at risk of fire in the future; because due to lack of water, they will become semi-desert areas and there will be practically nothing to catch fire in these areas! Today earth (lands) vegetation is about 55% and Deforestation rate is about 0.25% per year. Also forests growth rate is about 0.125% per year (Figs. 8 and 9).

7

Simulation

The overall System Dynamics stock and flow model that contain all variables and their relationship based on the above descriptions is as follows (Fig. 10). The simulation (from 2020 to 2100) shows if the current situation continues main variables of the models will behave as the following figures (Fig. 11). This charts says in 2050 and 2100, earth will be 2.38 and 5.56 more hot than 2020. This will be devastating and presents catastrophic Climate Change tackles.

1526

A. Vahidi

Pop Growth Rate

Industrial products demand

Global Population.

the extinction of plants and animals rate

agricultural products demand forests growth

Livestock products demand

Vegetation deforestation.. widespread droughts rate fossil fuels consumption

Wildfires

greenhouse gas Global warming.

Energy Efficiency

Global warming Rate Industrial products Waste Agri. Food Waste

Policies, Laws, Guidelines

alternative energy sources

Animal (protein) Food Waste

public awareness

Urbanization. Urbanization Rate

Maximum Urbanization

Fig. 10 Overall model

Fig. 11 Global Warming normal scenario

Climate Change models designed by the IPCC show that between 1990 and 2100, the average surface temperature rose between around 6.4 ◦ C. Also this study indicates that Global Warming will be about 6.26 ◦ C (from 1900 to 2100) (Fig. 12). As the simulation indicates, Fossil Fuel Consumption will be 1.3 and 2.1 times of 2020 consumption. It is mainly because of population increase. Also the chart shows an exponential pattern that make it more and more dangerous (Fig. 13). Many researchers have claimed that if urban populations continue to grow at the current rate, by 2050 the urbanization rate will rise from the current 55% to about 65%, which the model simulation confirms.

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1527

Fig. 12 Fossil Fuel Consumption normal scenario

Fig. 13 Urbanization normal scenario

Also it is shown that in 2100 earth urban population is about 80% of total population live on earth. This will be new dawn to human species lifestyle (Fig. 14). The above charts shows a worrying exponential pattern in earth (land) vegetation. In 2050 and 2100, vegetation decreased from 55% to 50% and 33%. This trend shows if human do not tackle Global Warming, he/she should live in a world with spars vegetation (Fig. 15). Simulation shows that droughts will be increased 5.8 and 12.1 times more in 2050 and 2100. Also animals and plants will extinct with same speed. It confirms that human will live in arid lifeless planet, if he/she don’t reduce Global Warming.

1528

A. Vahidi

Fig. 14 Vegetation normal scenario

Fig. 15 Droughts normal scenario

8

Solutions

8.1

Public Awareness

In this part, we will examine the effect of main applicable policies to reduce Global Warming. The first policy is increasing “Public Awareness”. Public Awareness cause tree is as follows (Fig. 16). As it is shown “Public Awareness” will decrease “Agri. Food Waste,” “Animal (protein) Food Waste,” and “Industrial products Waste”. Also it will increase Energy Efficiency due to education and public worrying about human race destination (Fig. 17).

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1529

Fig. 16 Public Awareness effects tree

Fig. 17 Public Awareness scenario

As the above chart show, increasing 10, 20, and 30% in “Public Awareness” will decrease temperature about 0.5, 0.9, and 1.2◦ in 2050. Also in 2100, increasing 10, 20, and 30% in “Public Awareness” will decrease temperature about 1.2, 2.1, and 2.8◦ . The chart shows a negative exponential pattern, so the 10% scenario has the most slope.

1530

8.2

A. Vahidi

Policies, Laws, Guidelines

“Policies, Laws, Guidelines” effects on deforestation and could decrease the causes of deforestation. Also it could increase forest growth by giving incentive to forest preservation and Arboriculture programs. Also, Policies, Laws, Guidelines will oblige industries to use less energy and increase their “Energy Efficiency” in their production (Figs. 18 and 19). “Policies, Laws, Guidelines” scenario is increasing its effect by 10, 20, and 30%. These scenarios will decrease temperature about 0.24, 0.44, and 0.61◦ in 2050. Also in 2100, increasing 10, 20, and 30% in “Policies, Laws, Guidelines” effectiveness will decrease temperature about 0.69, 1.24, and 1.7◦ . The chart shows a negative exponential pattern, so the 10% scenario has the most slope.

Fig. 18 Policies, Laws, Guidelines effects tree

Fig. 19 Policies, Laws, Guidelines scenario

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

9

1531

Conclusion

Do we have the right to consume enough electricity to warm the world, causes floods, destroy forests, because we want to make chandeliers? Scientists at a conference in Britain say Global Warming is fast approaching a point where it can no longer be stopped. Scientists involved in the “European Project,” which is studying the amount of carbon dioxide by sampling from the depths of the Antarctic ice sheet, described their findings at a conference in Norwich. They have shown that carbon dioxide levels are now much higher than at any time in the past 8,00,000 years, and they conclude that if Global Warming continues at its current rate, it may only take a decade for this trend to reach an irreversible point. Eric Wolff, of the British Center for Antarctic Studies, said the most frightening thing about their findings was the speed with which Climate Change was taking place. According to a new report released by the FAO, industrial animal husbandry emits 18% more carbon dioxide than the transportation system. In addition, it is an important factor in the destruction of water and land resources. “Industrial animal husbandry is an important factor in environmental crises today, and alternatives must be urgently considered,” said Henning Steinfield, FAO’s head of animal husbandry information and author of the report. As welfare increases, people consume more meat and dairy products each year. Global meat production is projected to more than double from 229 million tons in 2001–2009 to 465 tons in 2050, and industrial milk and dairy production from 580 tons to 1043 tons. The livestock industry is expanding more than any other branch of agriculture in the world, and this growing rate is followed by enormous environmental losses. Livestock is estimated to produce 9% of the carbon dioxide released from human activities, but the industry also produces other Greenhouse Gases that have more harmful environmental effects: 65% of all nitro oxide gas produced by human activities belongs to industrial livestock. 296 heating units (GWP) have more heating potential than carbon dioxide. Most of this gas is released from animal manure. In addition, it is estimated that 37% of the total methane gas produced by human activities is released from industrial livestock (Which has 23 times the heating potential of carbon dioxide!). This gas is mainly produced in the digestive tract of ruminants – and of course 64% ammonia, which plays a major role in acid rain. Today, industrial animal husbandry controls 30% of the entire surface of the earth. About 20% of the total live mass on earth is estimated to belong to meat and dairy cattle. The presence of so many livestock over a large area of the earth and their growing need for food is also a serious and critical factor in the destruction of biodiversity. Fifteen ecosystems out of 24 important ecosystems on earth are known to be endangered, the main culprit being industrial animal husbandry.

1532

A. Vahidi

Fig. 20 Combined scenario

Table 1 Global Warming (degree increase) Time (year) Combine 30 Combine 20 Combine 10 Policies, Laws, Guidelines 30 Policies, Laws, Guidelines 20 Policies, Laws, Guidelines 10 Public Awareness 30 Public Awareness 20 Public Awareness 10 Normal

2050 0.9816 1.307 1.749 1.773

2100 2.047 2.812 3.905 3.865

2050 compare to normal 41% 55% 73% 74%

2100 compare to normal 37% 55% 73% 74%

1.941

4.321

82%

81%

2.141

4.876

90%

90%

1.239 1.553 1.927 2.381

2.699 3.463 4.399 5.564

52% 65% 81% 100%

52% 65% 81% 100%

As current simulation shows, increasing “Public Awareness” and “Policies, Laws, Guidelines” could reduce Global Warming. These policies have effects on deforestation and could decrease the causes of deforestation. Also they could increase forest growth by giving incentive to forest preservation and Arboriculture programs. Also Policies, Laws, Guidelines will oblige industries to use less energy and increase their “Energy Efficiency” in their production. But if two solutions used together, better results will be provided. As the below Fig. 20 shows if combined scenario used Global Warming will decrease between 73% and 41% in 2050 and between 73% and 37% in 2100 (Table 1).

Simulation Applications in Analyzing the Trade-Off Between Climate Change. . .

1533

References M.R. Betz, M.D. Partridge, M. Farren, L. Lobao, Coal mining, economic development, and the natural resources curse. Energy Econ. 50, 105–116 (2015). https://doi.org/10.1016/j.eneco.2015.04.005 M. Bogaerts, L. Cirhigiri, I. Robinson, M. Rodkin, R. Hajjar, C. Costa Junior, P. Newton, Climate change mitigation through intensified pasture management: Estimating greenhouse gas emissions on cattle farms in the Brazilian Amazon. J. Clean. Prod. 162, 1539–1550 (2017). https://doi.org/10.1016/J.JCLEPRO.2017.06.130 J. Bollen, C. Brink, Air pollution policy in Europe: Quantifying the interaction with greenhouse gases and climate change policies. Energy Econ. 46, 202–215 (2014). https://doi.org/10.1016/J.ENECO.2014.08.028 M. Buchwitz, M. Reuter, O. Schneising, H. Boesch, S. Guerlet, B. Dils, I. Aben, R. Armante, P. Bergamaschi, T. Blumenstock, H. Bovensmann, D. Brunner, B. Buchmann, J.P. Burrows, A. Butz, A. Chédin, F. Chevallier, C.D. Crevoisier, N.M. Deutscher, et al., The Greenhouse Gas Climate Change Initiative (GHG-CCI): Comparison and quality assessment of near-surfacesensitive satellite-derived CO2 and CH4 global data sets. Remote Sens. Environ. 162, 344–362 (2015). https://doi.org/10.1016/J.RSE.2013.04.024 A. Costantino, L. Comba, G. Sicardi, M. Bariani, E. Fabrizio, Energy performance and climate control in mechanically ventilated greenhouses: A dynamic modelling-based assessment and investigation. Appl. Energy 288 (2021). https://doi.org/10.1016/J.APENERGY.2021.116583 J.A. Dunne, S.C. Jackson, J. Harte, Greenhouse effect, in Encyclopedia of Biodiversity, 2nd edn., (2013), pp. 18–32. https://doi.org/10.1016/B978-0-12-384719-5.00068-X R. Dutta, Use of clean, renewable and alternative energies in mitigation of greenhouse gases, in Encyclopedia of Renewable and Sustainable Materials, (2020), pp. 821–834. https://doi.org/10.1016/B978-0-12-803581-8.11048-3 J. Edmonds, J. Clarke, J. Dooley, S. Kim, S. Smith, Modeling greenhouse gas energy technology responses to climate change. Greenhouse Gas Control Technologies – 6th International Conference, 863–868 (2003). https://doi.org/10.1016/B978-008044276-1/50136-7 S. Fattorini, Climate change and extinction events, in Encyclopedia of Geology, (2021), pp. 585– 595. https://doi.org/10.1016/B978-0-12-409548-9.12116-5 J.W. Forrester, System dynamics, systems thinking, and soft OR. Syst. Dyn. Rev. 10(2–3), 245–256 (1994). https://doi.org/10.1002/sdr.4260100211 J.W. Forrester, The beginning of system dynamics. McKinsey Q. 4, 4–16 (1995). https://doi.org/10.3401/poms.1080.0022 J. Hayward, Introduction to system dynamics, in System Dynamics, (1961), pp. 1–18. https://doi.org/10.1049/ep.1968.0042 K.A. Hughes, P. Convey, J. Turner, Developing resilience to climate change impacts in Antarctica: An evaluation of Antarctic Treaty System protected area policy. Environ Sci. Policy 124, 12–22 (2021). https://doi.org/10.1016/J.ENVSCI.2021.05.023 T. Juniper, A food system fit for the future, in Rethinking Food and Agriculture, (2021), pp. 135– 148. https://doi.org/10.1016/B978-0-12-816410-5.00007-4 T. Kuramochi, L. Nascimento, M. Moisio, M. den Elzen, N. Forsell, H. van Soest, P. Tanguy, S. Gonzales, F. Hans, M.L. Jeffery, H. Fekete, T. Schiefer, M.J. de Villafranca Casas, G. De Vivero-Serrano, I. Dafnomilis, M. Roelfsema, N. Höhne, Greenhouse gas emission scenarios in nine key non-G20 countries: An assessment of progress toward 2030 climate targets. Environ Sci. Policy 123, 67–81 (2021). https://doi.org/10.1016/J.ENVSCI.2021.04.015 D. Nong, P. Simshauser, D.B. Nguyen, Greenhouse gas emissions vs CO2 emissions: Comparative analysis of a global carbon tax. Appl. Energy 298 (2021). https://doi.org/10.1016/J.APENERGY.2021.117223

1534

A. Vahidi

A.R. Pfadt-Trilling, M.-O.P. Fortier, Greenwashed energy transitions: Are US cities accounting for the life cycle greenhouse gas emissions of energy resources in climate action plans? in Energy and Climate Change, vol. 2, (2021), p. 100020. https://doi.org/10.1016/J.EGYCC.2020.100020 S. Rajkhowa, J. Sarma, Climate change and flood risk, global climate change, in Global Climate Change, (2021), pp. 321–339. https://doi.org/10.1016/B978-0-12-822928-6.00012-5 V. Ramanathan, Y. Feng, Air pollution, greenhouse gases and climate change: Global and regional perspectives. Atmos. Environ. 43(1), 37–50 (2009). https://doi.org/10.1016/J.ATMOSENV. 2008.09.063 I. Righini, B. Vanthoor, M.J. Verheul, M. Naseer, H. Maessen, T. Persson, C. Stanghellini, A greenhouse climate-yield model focussing on additional light, heat harvesting and its validation. Biosyst. Eng. 194, 1–15 (2020). https://doi.org/10.1016/J.BIOSYSTEMSENG.2020.03.009 B.K. Sovacool, S. Griffiths, J. Kim, M. Bazilian, Climate change and industrial F-gases: A critical and systematic review of developments, sociotechnical systems and policy options for reducing synthetic greenhouse gas emissions. Renew. Sust. Energ. Rev. 141 (2021). https://doi.org/10.1016/J.RSER.2021.110759 J.D. Sterman, System dynamics modeling: Tools for learning in a complex world. Calif. Manag. Rev. 43(4), 8–25 (2001). https://doi.org/10.1111/j.1526-4637.2011.01127.x T. Tokioka, Climate changes predicted by climate models for the increase of greenhouse gases. Prog. Nucl. Energy 29(SUPPL), 151–158 (1995). https://doi.org/10.1016/0149-1970(95) 00038-L R. Tuckett, Greenhouse gases and the emerging climate emergency. Climate Change, 19–45 (2021). https://doi.org/10.1016/B978-0-12-821575-3.00002-5 Vahidi, A. (2018). Management Cybernetics Evolution A. Vahidi, A new defuzzification method for solving fuzzy mathematical programming problems. Hacettepe J. Math. Stat. 48(3), 845–858 (2019). https://doi.org/10.15672/HJMS.2018.612 A. Vahidi, A. Aliahmadi, Describing the necessity of multi-methodological approach for viable system model: Case study of viable system model and system dynamics multi-methodology. Syst. Pract. Action Res. (2019). https://doi.org/10.1007/s11213-018-9452-0 A. Vahidi, A. Aliahmadi, E. Teimoury, Researches status and trends of management cybernetics and viable system model. Kybernetes 48(5), 1011–1044 (2018). https://doi.org/10.1108/K11-2017-0433 A. Vahidi, A. Aliahmad, E. Teimouri, Evolution of management cybernetics and viable system model. Syst. Pract. Action Res. 32(3), 297–314 (2019). https://doi.org/10.1007/S11213019-9478-Y S.H. Vetter, T.B. Sapkota, J. Hillier, C.M. Stirling, J.I. Macdiarmid, L. Aleksandrowicz, R. Green, E.J.M. Joy, A.D. Dangour, P. Smith, Corrigendum to “Greenhouse gas emissions from agricultural food production to supply Indian diets: Implications for climate change mitigation” [Agric. Ecosyst. Environ. 237 (2017) 234–241] (S0167880916306065)(10.1016/j.agee.2016.12.024). Agric. Ecosyst. Environ. 272, 83–85 (2019). https://doi.org/10.1016/J.AGEE.2018.11.012

Analysis of Energy Transition Pertaining to the Future Energy Systems Engin Deniz and Melih Soner Çelikta¸s

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Boundary and Framework of the Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 An Overview of Energy Transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Trends on Energy Management in the Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 A Brief Overview on Energy Storage Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Strengths, Challenges, and Future Research Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Conclusion and Future Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1536 1538 1538 1541 1545 1550 1551 1551

Abstract

The increase in electricity demand and how it can be supplied are one of the world’s most serious challenges and risks. As the energy transition continues, the array of alternatives and paradigm changes in the energy system’s non-fuel sources become more prominent. The electric power grid is being modernized to make it easy to use modern, uninterrupted electricity while the ongoing global transition of the energy system is taking on new facets, especially as the amount of electricity produced from intermittent sources such as renewable energy increases. Increased integration of intermittent renewable energy sources necessitates the construction of energy storage to store surplus electricity and improve system reliability, all while contributing greatly to the advancement of

E. Deniz Hanwha Q CELLS GmbH, Berlin, Germany Beuth University of Applied Sciences Berlin (Beuth Hochschule für Technik Berlin), Berlin, Germany M. S. Çelikta¸s () Solar Energy Institute, Ege University Bornova, Izmir, Turkey e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_92

1535

1536

E. Deniz and M. S. Çelikta¸s

emerging technology. It is becoming an important tool in the design of energy systems that are applicable to the advancement of existing energy technologies. This section analyzes numerous emerging developments and applications for modernized power systems while illustrating emerging technical innovations that are useful for resource effective power grids by undertaking a literature review with a focus on various advanced energy system technologies. Keywords

Future energy systems · Integration of renewable energies · Energy storage systems · Energy system analysis · Solar energy · Energy transition

1

Introduction

The world is experiencing a paradigm shift phase regarding the current energy generation methods that have been used to date in terms of energy diversity (Junne et al. 2019). Energy business is a strategic field qualified as a vital importance for countries’ development policies because population increase and industrialization and urbanization concepts around the world and the commercial opportunities increasing due to globalization increase the demand for natural resources and energy every day (Junne et al. 2019). Conspicuously, the new manufacturing and distribution mechanism based on fossil fuels is not sustainable and is exacerbating global environmental issues such as global warming and greenhouse gases (GHG) (Fouquet 2010). Due to the macroeconomic and environmental consequences of using fossil fuels, countries are largely aiming to develop existing technology to strive for a sustainable model, and they are seeking to increase the share of renewable energy in the next generation model (Xiao et al. 2019). The increasing importance of renewable energy in recent years is a result of conventional energy sources that have detrimental effects on the environment. To increase the diversity of existing renewable energy resources, fossil-based power plants are planned to close gradually. By regulating existing policies which is the key factor of the development for energy technologies, directing market transformation, and technological development, the energy transition is attempted to be characterized in a multidimensional manner; a new economic system is tried to be conceptualized with sustainable energy resources and technologies (Gürtler et al. 2021). The majority of the studies outline pathways to phase out technologies that are unsustainable while implementing sustainable renewable energy sources that will use less energy and meet increasing energy demands (Xiao et al. 2019; Gürtler et al. 2021). Digitalization, decentralization, and electrification are poised to create new economic models, change consumer behavior, and radically shape traditional economies, aided by innovative policy processes and business technologies (Mikova et al. 2019). With a huge expansion of low-cost renewables, a smarter and much more flexible power grid, and dramatic improvements in the number of vehicles and other products and processes that run on electricity, the future energy system

Analysis of Energy Transition Pertaining to the Future Energy Systems

1537

might look very different than what we have today. Supporting sustainable growth and combating environmental improvements are critical components of dealing with rising energy demand (Gürtler et al. 2021). Regulations and technological developments form the course and momentum of energy transition and have long-term consequences for the stability of energy systems, and organizations are adjusting to this energy shifting, with the coming of new market form and players (Dambeck et al. 2021). Projections of US Energy Information Administration (EIA) depict that in case existing energy policies and energy demand preferences are continued, the world’s primary energy demand will increase nearly 50% between the years of 2018 and 2050 (Energy Information Administration (EIA) 2018). This energy deficit can be filled by moving from traditional generation to renewable energy sources – a transition that currently providing 28% of the global electricity generation – opening the way for decarbonization and keeping pace with increasing energy demand (EWEA 2015). Hence, expanding clean energy sources is critical to reaching an environment-friendly global economy, whereas a sustainable society comprises energy needs that are mostly met by renewable energies, as well as expanded electrification in all industries (IEA 2020). Today, where electric energy consumption per person indicates development, the persistent depletion of primary energy supplies, environmental degradation, and also climatic equilibrium problems associated with the use of such fuels tend to be a problem that requires urgent attention (Junne et al. 2019; Xiao et al. 2019). One approach to the issue is to find renewable and new energy supplies through the adaptation of new technology, and another is to reduce demand to the bare minimum economically acceptable level (Mikova et al. 2019). Individual consolidation and privatization efforts in manufacturing, transmission, distribution, trading, and other companies are accelerating around the world, especially in European electricity markets that are liberalizing and restructuring in this direction in order to provide more reliable and functional electric markets. In spite of these challenges, transmission services in such a strong public monopolistic trend remain somewhat popular due to the high capital investment and substantial expropriation issues. It is the primary goal of reform programs in energy that consumers have choices on a competitive, transparent market that can be summarized in seven main topics: quality, affordable, uninterrupted, environmentally friendly electric production, transmission, and sales services (Energy Information Administration (EIA) 2018; Junne et al. 2019; Mikova et al. 2019). According to the International Renewable Energy Agency, the world economy’s energy transition from nonrenewable to renewable energy is progressing toward decarbonization and can reach 90% or more of the needed CO2 reductions. Renewable energy is an essential component of energy supply since it accounts for a major portion of the supply, balances marketing discrepancies, and preserves the environment. As a result, the development of renewable energy sources has become a critical component in the transition to a low-carbon economy. Switching from nonrenewable to renewable energy sources can help bridge the gap between energy demand and supply, lowering carbon dioxide emissions (IRENA 2020a).

1538

E. Deniz and M. S. Çelikta¸s

Considering the global political and technological transition, the energy transition problem is no longer restricted to individual countries. To achieve the energy transition goals, it is necessary to promote interaction and strategizing on a multilevel basis. The energy transition must be examined and analyzed from a comparative angle in other words. Our review aims to provide an overview to examine current trends in the energy transition, energy storage, and technology, along with what lessons can be learned from previous assessments.

2

Boundary and Framework of the Review

There is an inextricable link between energy and carbon emissions. Energy efficiency and energy transition have contributed to the development of quantitative measures of carbon emissions. Energy transition has traditionally been viewed as the process of replacing and electrifying energy, incorporating renewable energy as an indirect measure. Research has been conducted on various aspects of energy transition, energy policy, and evidence-based assessment indicators that may be used both internationally and domestically. Our systematic review has focused mainly on energy- and storage-related scientific publications found in various sources. The review focused on energyrelated books and journal articles published between 1990 and 2021 to capture the most recent debate. We excluded conference papers and book chapters to ensure replicability and ease of access.

3

An Overview of Energy Transition

Among the commitments of the Kyoto Protocol adopted in 1992 was that industrialized countries must reduce and developing economies must stabilize greenhouse gas emissions at 1990 levels by 2000. A quarter century later, the Paris Agreement came into force (Gürtler et al. 2021). Decarbonization of the energy mix has been a policy goal of governments since the Kyoto Protocol. Even though these policies relatively failed to reduce carbon emissions, the production of renewable energy has increased dramatically (Keyßer and Lenzen 2021). In relation to nuclear energy, renewable energy has reached a similar level of penetration as it has. The renewable energy transition, unlike nuclear energy, is expected to continue at an unprecedented pace, say energy institutions and industry. In 2040, this share of renewables will range between 25% and 30%, surpassing previous energy shifts, according to diverse energy scenarios from different institutions (IRENA 2020a; IEA 2021; Fig. 1). To reduce global warming emissions and to increase the environmental sustainability of every nation, the energy sector must adapt and implement structural reforms that will result in reduced greenhouse gas emissions (Xiao et al. 2019; Keyßer and Lenzen 2021). Almost all fossil fuel-based nonrenewable energy sources burn fossil fuels, causing hydrocarbons to break down and releasing carbon dioxide into the atmosphere. Thus, if we use these forms of energy in greater

Analysis of Energy Transition Pertaining to the Future Energy Systems

1539

Fig. 1 Renewable energy supply in selected global energy scenarios. Source: Ghasemian et al. (2020)

numbers, it may be environmentally harmful and a threat to the environment. Changing energy consumption is critical to changing the way we use energy in the future as well as achieving long-term growth. An important source of concern for economic growth is the large investment required for renewable energy generation. Economic development can be influenced in a variety of ways by the creation of renewable energy sources (Hess 2014). Energy transitions can maintain their effectiveness if they are successful in local transitions. The transition to low-carbon and sustainable energy is predicted to play a critical role in improving human well-being, safety, and sustainability while mitigating the consequences of global warming (Selvakkumaran and Ahlgren 2020). Due to increasing attention being paid to renewable energy sources as an effective means of decarbonizing the global energy system and preventing global warming, working on linking the major sectors of the energy sector is underway in an effort to capitalize on synergies. Numerous assessments, mostly focusing on the power industry, underline the problems of incorporating additional renewable energy. However, by increasing the flexibility of the energy system, cross-sector integration may allow for greater proportions of renewables to be incorporated. Several studies show the possibility of integrating the power, heat, transportation, and industrial sectors so that 100% of future energy needs may be met with renewable sources (Kazan et al. 2015). Electrical integration is fundamental to the advancement of sectoral integration because of the given fact that transportation, industrial, and thermal sectors are all incorporated in a smart energy system by means of electrification. The demand for energy has been driven by changes in production technologies, decentralization of generation, and use of renewable

1540

E. Deniz and M. S. Çelikta¸s

energy. Renewable resources, such as solar, wind, hydro, and bioenergy, provide a crucial tool for reducing carbon emissions. Over the last few years, solar and wind energy have become the preferred source of energy as costs have declined rapidly (Mousavi Baygi and Sadrameli 2018). Sustainable development can be achieved by expanding the use of renewable energy, notably solar, wind, and biomass. Renewable energies can supply 25% of primary energy in 2030, according to the International Renewable Energy Agency (IRENA) REmap scenario, and with consumption efficiency measures, the share of renewable energy can reach 30% (IRENA 2020a). Carbon capture and storage (CCS) technology is identified in several studies as a crucial component of a 2 ◦ C scenario. The adoption of CCS might cut emissions from the industrial sector while also eliminating emissions from biofuel burning. To reduce environmental, economic, and social burdens, the energy system should also become more electrified as much as possible. It is possible to substitute renewable synthetic hydrocarbons in a carbon capture and utilization process for any remaining energy services that cannot be provided by electrification, storage services for seasonal, or processes requiring hydrocarbons. It is important to recognize that the carbon budget that remains once fossil CCS has been implemented is subject to considerable uncertainties (Aghaie et al. 2018). Majority of GHG emissions come from electricity, heat, transportation, and industrial sources, and the rest is coming from agriculture and land use. Decarbonizing the electricity sector seems to be the simplest and most influential way to decarbonize while also the most important for the rest of the energy industry (Lewis et al. 2018). In conjunction with the development of an electrified power sector and close sector coupling, decarbonized energy will be available for heat and transport. There are several approaches for lowering carbon emissions from heating networks through the use of heat pumps and heat storage. The drop in energy supply prices can be linked to increased storage and warehousing capacity, with market-clearing prices strongly affecting the outcome (Li et al. 2021). Cogeneration, or combined power and heat generation, is an efficient way to increase energy efficiency. Considering the basis of biomass power as a source of renewable energy generation and applying appropriate sustainability criteria for its use should be a key part of designing the energy system for the coming years (Coban et al. 2012; Uyan et al. 2020). On the other hand, scientists, as well as the International Energy Agency (IEA), recognized the hydrogen role in this progression from the beginning. In the early twenty-first century, hydrogen was strongly backed as an important part of energy transition. Hydrogen, with its zero emissions of carbon dioxide and greenhouse gases, has consistently provoked curiosity and hope for future development as one of the most unique types of energy storage. Even though development of a useful and stable technology took a long time and costs money, hydrogen proved to be worth the risk. As a consequence, hydrogen is today an essential element of energy storage. When applied to a wide range of technologies, electrolysis is recognized as one of the most powerful methods to decarbonize technology and build a carbon neutral society. In order to decrease the production costs, research and development must

Analysis of Energy Transition Pertaining to the Future Energy Systems

1541

60 50 40

Coal

30

Fossil

20

Renewables

10

Wind and solar

0

Fig. 2 % share of electricity production in EU-27. Source: Redl et al. (2021)

still play a fundamental role, since the main obstacle to development remains the financial aspects. As history demonstrates, a new technology’s inertia will allow it to be widely adopted by inertia if its techno-economic implications will be favorable (IEA 2019). Energy systems are transitioning from fossil fuels to renewable energy as a result of the switch from fossil fuels to renewable energy. The amount of electricity produced by renewable energy systems has surpassed the amount generated by fossil fuels by 2020 in the EU as well as in the UK (NRG_IND_REN 2021; Fig. 2). Approximately 34% of the power generated by the EU-27 came from renewable sources, while the rest of the power came from fossil fuels in 2019 (Redl et al. 2021). In some part, the drop in power use can be attributed to the COVID-19 pandemic, which dropped the power use by 7%, though it was not the major factor. The market share of fossil fuels has decreased as a result of rising renewable energy production (Fig. 3). Rather than technology improvements, policies are driving the energy transition. The energy transition represented here is different from previous energy transitions, and it has significant implications. Each country implements policies differently, and the mix of energy a country uses, the level of energy supplied, and the price difference are all factors that determine whether a country’s decarbonization level is the same as another’s. Decarbonizing the economy in the coming years is more necessary than ever due to climate change challenges (Ipakchi and Albuyeh 2009).

4

Trends on Energy Management in the Power System

The future of sustainable energy will be marked by smart, adaptive, and renewable systems backed by a healthy mix of participation from all sectors. A major increase in the usage of renewable energy sources is required as part of the energy transition.

1542

E. Deniz and M. S. Çelikta¸s

120% 100% 80% 60% 40% 20% 0%

Fig. 3 Share of energy from renewable sources in EU-27. Source: NRG_IND_REN (2021)

At the moment, there are still barriers to boosting renewable energy use. Because current clean energy technologies have reached commercial maturity, the majority of the remaining impediments to increased renewable energy deployment are connected to various national and regional policies (IRENA 2019). It is a challenge to find suitable ways to incorporate renewable energy sources into the energy system on a large scale. Renewable resources such as solar and wind power are intermittent in nature, which can be accommodated by a variety of flexible technologies, particularly energy storage. As a result, there is a clear need for more research into demand response and energy storage technologies, as well as mathematical models that examine the impact of various technical choices and their contribution to energy system decarbonization (Loorbach 2010). Analyzing the total amount of materials needed to shift to a sustainable global energy system is also critical. Global energy transitions have the potential to significantly change mineral and energy resource flows throughout time (Meadowcroft 2009). As a result, capturing the impact of large-scale energy transitions on resource flows in a comprehensive and dynamic manner is becoming increasingly important. The role of the source-energy nexus in the energy transition is an area of research that needs to be explored further, because a future energy system with a higher proportion of renewable energy will need to decrease losses. Moreover, it is necessary, taking into consideration operational flexibility and compatibility of the energy sector with the inherent intermittence of changeable renewable resources, to study the chain of sustainable source and energy provision further. Transitioning to a renewable energy-based electricity system is a tough task for many places throughout the world. There are regions with significant wind and solar resources, but there are other locations with harsh continental temperatures and an energy-intensive economy based on fossil fuels. In such places, cross-sectoral

Analysis of Energy Transition Pertaining to the Future Energy Systems

1543

integration and the deployment of smart and innovative energy technology should be prioritized. In an integrated energy system, energy can be moved from one system to another, and high demands and outputs can move between vectors, allowing the entire system to be flexible (Connolly et al. 2011). The shaping and restricting of future growth by accumulated precedent is referred to as route dependency, and due to its route dependency, energy system is a tremendous complexity, interdependent, and long-lived infrastructure with a low turnover rate. During the twentieth century, technological and economic advantages to scale contributed to the physical and economic creation of a highly centralized energy system (Zheng and Kammen 2014). Although there exists a growing technological and business case for decentralization, driven by systematic forces such as liberalization of markets, environmental policies and regulations, technological breakthroughs in renewables and storage, and the convergence of information technology and digital control systems with energy infrastructure (Kern and Smith 2008). As such, this contradiction is a fundamental problem for future energy development and has important implications for technological innovation and deployment, the political and regulatory frameworks, corporate and investment strategies, and social acceptability and commitment. The views of diverse operators and consumers of energy systems concerning the future and its impact on the management and operation of energy technology and infrastructure will define the future of the energy systems. There is the possibility of changing the roles, relationships, and responsibilities of cotorie each other within a distributed energy system, such as power producers, distributors, transmission operators, and end consumers (Burke and Stephens 2018). It is possible that new actors may emerge as a result of the integration of energy systems. Also energy firms employing distributed energy resources, cities, and utilities offering flexible and intelligent home services will be among those participating (Campbell 2018). Since 2014, solar energy has the same levelized cost of energy (LCOE) as electricity production. There was continuous, sometimes bruising debate on the costs of integration of variable renewables like wind or solar power into energy networks. The costs of generating renewable energy have decreased so significantly that, in certain situations, the average costs of electricity generated by renewable and traditional sources no longer vary. The PV prospect is particularly pertinent to decision-makers and investors concerned about climate change. According to some studies, it is anticipated that by the middle of the century, the installed solar system will exceed 15 TW with a significant rise in demand for solar power (Ipakchi and Albuyeh 2009). Other observers argue that photovoltaic development is limited by the use of land and the flexibility of the underlying grid (Al-Shahri et al. 2021; Fig. 4). Globally, the installed capacity of wind turbines exceeded 743 GW at the end of 2020, a result of adding over 93 GW of new capacity with a 52% year-onyear growth. Wind energy is the essential source of renewable energy in the

1544

E. Deniz and M. S. Çelikta¸s

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05

CSP LCOE Onshore wind LCOE Oﬀshore wind LCOE

Hydro LCOE Bioenergy LCOE

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

0

Solar LCOE Geothermal LCOE

Fig. 4 Levelized cost of electricity by technology. Source: IRENA (2020b)

energy transition process while onshore and offshore installations are increasing cumulatively (Ozay and Celiktas 2016). Offshore wind farms have grown quickly as a result of the demand for net-zero carbon electricity. Additionally, their maintenance, repairs, and electric grids are more complex and expensive, and they need observation-based data for optimization (Jung and Broadwater 2014). However, even though offshore wind farms typically have higher building and maintenance costs than onshore wind farms, their greater capacity factor, which is often offset by the strength of offshore wind sources, typically compensates for these costs. On the surface, offshore wind farms generate electricity longer than their onshore counterparts based on the relative strength of offshore wind resources. Although offshore wind farms cost more to build and to maintain than onshore farms, they are partially offset by higher capacity resulting from the strength of offshore wind resources. The lifetime cost of energy (LCOE) for onshore wind has decreased by 36% globally from 2014 to 2019. Based on a variety of recent estimates from the EU and global sources, offshore wind’s lifetime cost of energy has fallen by 28% from 2014 to 2019 (Carrasco et al. 2006). Markets are shifting and environmental regulations are strengthening, suggesting a moment of transformative changes for the auto industry. Over the past few

Analysis of Energy Transition Pertaining to the Future Energy Systems

1545

decades, electric vehicles (EVs), in particular, have grown in popularity and expanded worldwide. Globally, renewable energy sources can supply enough electrical demand for electric vehicles, according to the International Energy Agency (IEA) (IEA n.d.). As an effective storage source for flexible renewable energy sources, electric mobility leads to frequent productivity increases as well as decreasing carbon dioxide emissions and other air pollution (Fouquet 2016). Energy transition and technology choices often aim to approach the same problem from two distinct perspectives, macroeconomic and microeconomic, with substantial and diverse implications for investments and business strategy. It is widely acknowledged that existing energy systems are unsustainable, and many countries have already begun transitioning to renewable and sustainable energy sources. As the climate changes, energy security issues arise, energy poverty occurs, and troubled hybrid energy systems are managed, the entire energy system must undergo a fundamental restructuring, and therefore, the economy will have to change (Lund et al. 2016).

5

A Brief Overview on Energy Storage Systems

There are various methods of energy storage technology which can be categorized for storing the energy in different way such as mechanical, electrical, chemical, electromechanical, as well as thermal energy storage. A balance between supply and demand is always hard to maintain. When significant quantities of renewable energy sources are integrated whose production changes during the day because of weather conditions, the network balance is increasingly problematic. Although batteries for large-scale energy storage were not utilized frequently because of their high costs, they are being employed for energy applications. The necessary battery voltage and current can be achieved by connecting the cells in parallel and in series. The capacity of the battery is determined via energy and power measurement. A variety of mechanical energy storage systems, including pumped hydroelectric systems, flywheels, and compressed air energy systems, have the biggest installed capacity to convert electricity into mechanical energy. Mechanical energy storage uses a variety of mechanical energies, depending on the requirements of electrical systems, as a medium for storing and releasing energy (Larcher and Tarascon 2015). The pumped hydroelectric energy storage system generates electricity by pushing water higher and then releasing it through a turbine (Hanley et al. 2016). Traditional pumped hydroelectric energy storage systems employ vertically divided water reservoirs as storage tanks. During off-peak hours, pumps move water from the lower reservoir to the upper reservoir. When necessary, energy is generated by altering the flow of water (Emmanouil et al. 2021). Pumped hydro is a hydro system in which water is stored and then distributed which can be used at floating PV plants or wind power plants as a hybrid choice in order to gain more efficiency and sustainable energy. Currently, pumped hydroelectric energy storage systems coupled with wind or solar power generation are being developed. A network of isolated or distributed renewable resources may benefit from this. Since pumped

1546

E. Deniz and M. S. Çelikta¸s

Fig. 5 A hybrid hydro-wind-solar system with pumped storage system. Source: Simão and Ramos (2020)

hydroelectric energy storage system operates as time shifters, frequency controllers, nonspinning reserves, and supply reserves, their applications are generally closely associated with energy management. The construction time and cost of pumped hydroelectric energy storage systems are very high due to site selection limitations (Vicenzutti et al. 2021; Fig. 5). In compressed air energy systems, energy is stored in the form of compressed air in a vessel when it is not in use. Off-peak electricity is often used for compressors in these systems. Using compressed air to discharge into a wind turbine uses compressed air energy storage during periods of lower electric energy demand to support electrical network operation and then release the compressed air to drive an expander during periods of peak demand for electricity generation. Similar to pumped hydroelectric energy storage systems, this technology is an incredibly high power and energy storage system with a single unit, making it ideal for large-scale applications such as peak shaving, load shifting, and auxiliary services for grids. CAES possesses many benefits, including long energy storage, short response time and high efficiency, low environmental impacts, high efficiency, low cost, and long life of about 40 years (Budt et al. 2016). In fact, it is difficult to locate a largescale CAES system near mines and hydrocarbon fields due to the geographical limitations, which will result in high capital costs (Wicki and Hansen 2017; Fig. 6). Flywheels are electric energy storage devices that make use of cylinders with shafts attached to generators (Larcher and Tarascon 2015). The generator transforms electric energy into kinetic energy by increasing the rotational speed of the flywheel. This spinning mass acts as a mechanical battery by storing kinetic energy. The rotor is frequently housed in an evacuated cylinder, allowing it to travel at high speeds while using renewable or off-peak electricity and storing it as rotational energy. FES converts electrical energy into flywheel kinetic energy during off-peak hours by

Analysis of Energy Transition Pertaining to the Future Energy Systems

1547

Fig. 6 Simplified diagram of distributed compressed air energy storage. Source: Huang et al. (2018)

making the flywheel rotate rapidly. Among the many factors affecting FES energy density are the rotational velocity of the flywheel, energy loss caused by wind shear, and air resistance. The energy density of FES is thus enhanced by increasing the flywheel material strength or by placing the flywheel in a vacuum environment. Wind solo applications continue to suffer from storage costs as a major constraint (Chen et al. 2009; Fig. 7). A variety of stand-alone renewable energy systems employ batteries as their energy storage system. Throughout everyday life and business, rechargeable batteries are among the most commonly used storage devices. To provide a particular load with the requisite voltage and capacity, a battery is made by connecting electrochemical cells in series and parallel. Batteries come in a variety of varieties, many of which are still undergoing research and development. Many battery kinds lack separate power and energy capacities, which are decided during the battery’s construction. A battery’s efficiency, life duration, operating temperature, selfdischarge, and depth of discharge are all important factors to consider (Goodenough and Kim 2010). During the design of most battery types, power and energy capacity are fixed and do not fluctuate with power. Over time, a storage system fills up with energy, and the energy is released for its uses for the same amount of time, so battery applications often require short discharge and recharge times (Wang et al. 2014). For power applications, several cycles are needed per day. Storage of the batteries is not

1548

E. Deniz and M. S. Çelikta¸s

Fig. 7 Typical layout of a nonintegrated electromechanical flywheel energy storage system. Source: Buchroithner et al. (2018)

restricted to main batteries like lead-acid and lithium-ion. Other types of batteries are also appropriate for grid applications but need to be reengineered (Choi and Aurbach 2016). A lead-acid battery is the oldest and most common type of rechargeable battery, and it is used in almost all power systems (Larcher and Tarascon 2015). The use of lead-acid batteries on stationary devices permits both energy management and battery backup. Various technologies have been developed for energy production, hybrid cars, and electrical vehicles. Utility storage systems across the world, however, face numerous constraints due to their low cycling durations, low energy density, and high specific energy. In particular, lead-acid batteries facilitate renewable energy adoption in stand-alone energy systems due to their spill resistance, ease of use, and lower cost compared to other types of batteries (May et al. 2018). Despite their technical shortcomings, lead-acid batteries have a number of technical issues, including a short cycle life, shallow discharge, durability limitations, and poor charging and maintenance. Recent research has attempted to solve these problems by developing lead-acid batteries (Goodenough and Park 2013). The lithium-ion cell consists of a cathode, anode, separator, and electrolyte, just like an electrochemical battery (Soloveichik 2011; Celiktas and Alptekin 2019). However, a lithium-ion battery has a distinct internal structure and function than a normal battery. When the battery is charged, the lithium particles in the cathode transform into ions. These ions go through the electrolyte to the carbon anode, where they interact with outside electrons to create lithium atoms between the carbon layers. During discharge, similar procedures are utilized (Sasaki et al. 2013). Recent advancements in Li-ion battery technology have resulted in substantially more affordable large-scale storage options (Mossali et al. 2020). However, because of the high demand for lithium from portable devices, hybrid electric cars, and electric

Analysis of Energy Transition Pertaining to the Future Energy Systems

1549

vehicles, large-scale storage options may become prohibitively expensive. This is the most costly form of battery, but it also has the lowest cost per cycle, making it ideal for storing renewable energy. To be useful in diverse grid applications, this battery chemistry will need to continue to lower prices while also improving longevity and state of charge (Petrovic 2021). The nickel cadmium battery technology has reached maturity. An electrode is made of metallic cadmium and nickel hydroxide. The system is typically durable and generally dependable, with few maintenance requirements (Tarascon and Armand 2001). Nickel-cadmium batteries have a memory effect, which means they can lose a lot of capacity if drained and recharged incorrectly (Shukla et al. 2001). It provides good protection against voltage sags and standby power, making nickel-cadmium batteries appropriate for demanding settings. Because of their excellent temperature resistance, nickel-cadmium batteries are becoming increasingly attractive for storing solar energy. Aside from its high initial cost, this technique has numerous advantages; its main disadvantages are its poor efficiency rates and the risk of cadmium contamination (Stephan et al. 2021). Flow batteries are liquid electrolytes dissolved in one or more types of ions (Larcher and Tarascon 2015). Flow batteries are distinguished by the following characteristics: high power, extended duration, decoupled power and energy ratings, rechargeable electrolytes, quick charge and discharge rates, and poor efficiencies (Soloveichik 2011; Larcher and Tarascon 2015). Because separated electrolytes cannot react with one another, the system has no self-discharge. Li-ion batteries are also compact and lightweight, have a high energy density, and function at close to 100% efficiency, making them ideal for portable electronics (Chandran et al. 2021). Although Li-ion technology is promising, it has some disadvantages, including expensive prices and limited battery life owing to severe depletion. Flow batteries may be ideal for applications needing long-term storage because to their non-selfdischarge capabilities. Aside from the higher operating and capital expenses for a chemical plant requiring pump systems, flow control, and external storage, the flow battery system has a significant disadvantage in terms of capital and operating costs. When it comes to boosting power density, flow batteries confront the most severe hurdles (Wen et al. 2008). Sodium-sulfur batteries are well-suited for storage of electrical energy for utility-scale applications, especially for large-scale electric utilities (Ceraolo 2000). Sodium-sulfur batteries can be recharged at high temperatures and provide appealing options for a wide range of applications. This battery type has a long cycle life and was built using low-cost components, in addition to enabling high energy density and great charge efficiency (Soloveichik 2011). Managing traffic and energy quality, slashing peak loads, and integrating renewable energy are examples of applications. A main goal of research and development is to improve cell performance indices and reduce/eliminate the constraints associated with operating them at high temperatures (Dunn 2002). In the future, hydrogen energy is predicted to be an important source of energy (Edwards et al. 2008; Larcher and Tarascon 2015). Using key technologies such as hydrogen manufacturing, hydrogen storage, and hydrogen fuel delivery, hydrogen

1550

E. Deniz and M. S. Çelikta¸s

energy storage works by converting electrical power into hydrogen fuel (Jain 2009). There is considerable evidence that liquid hydrogen energy storage is suitable for a ESS of today’s high capacity, and there is enormous potential for its development, due to its many advantages, such as its high energy density, low operating and maintenance costs, and environmental impact (Abe et al. 2019). There are still issues with hydrogen storage systems, such as low conversion efficiency and high production costs (Widera 2020). Due to continuing advances in materials and fundamental technologies, the world looks to be on the cusp of huge breakthroughs in hydrogen energy storage (Evans et al. 2012). An electrical power storage system transforms electrical power from a distribution system into a form which can be stored and converted into electrical energy, if necessary, through an external interface. In distribution network storage systems, an electricity conversion system provides the electrical interface. It is possible for the electric energy system to decarbonize economically with solely renewable sources, a result of the development of energy storage technologies that provide prolonged storage without increasing storage prices (Lund et al. 2015).

6

Strengths, Challenges, and Future Research Needs

The present production pattern indicates that an energy transition is achievable in the medium term if existing policies are executed, but the transportation sector’s longterm viability is jeopardized. Currently, the debate is dominated by concerns about climate action; nevertheless, it is equally necessary to address policy gaps in order to manage the long-term risk of sustainable energy production and carbon containment routes. To close these gaps, a coordinated global effort is required to apply wellestablished local government regulations all over the world, increase the efficiency of existing energy technologies to reach a lower carbon intensity in industry, and improve efforts to create alternative choices. The relevance of the findings justifies more action, and study in this area is necessary. Encouraging long-term and lasting changes in energy demand may be more justified by energy transition policy goals. These changes may include a change in institutional frameworks, such as incorporating the benefits and market effects of greenhouse gas reductions into cost-effectiveness tests or defining a broader set of performance metrics that include renewable energy market transformation and energy efficiency. Utility companies can participate in the low-carbon transition by doing more than just generating electricity, for example, by increasing energy efficiency, decentralization, balancing systems, and enabling decentralized generation. On the other hand, dynamics of the energy markets are heavily impacted by structural changes in electricity markets around the world. Consumption and manufacturing processes are evolving in such a way that vertically integrated companies are no longer dominant on the market. A new generation of smaller businesses offering a variety of core services is disrupting the value chain. Changing the structure of the economy and society has significant economic and social repercussions.

Analysis of Energy Transition Pertaining to the Future Energy Systems

1551

The development in battery technologies, as well as the ability to be coupled to renewable energy systems, will accelerate the speed and dynamics of energy transition. Another development on the battery side is that increasing electric vehicles will play a key role in developing battery systems. It is recommended to carry out further studies on safety and price related to battery technologies. In addition, the efficiency and lifetime of energy storage systems and some renewable energy systems lead to high operating and maintenance costs in accelerating the energy transition. Based on the literature search over time, various advances are requested in the near future. Research and development of approaches that will reduce costs while increasing system efficiency and durability should be done. In addition, policy makers should develop measures that can bring renewable energy and advanced battery systems to today’s markets and encourage the development of integrated energy storage systems.

7

Conclusion and Future Perspectives

Clean energy has effectively repressed the environmental pollution concerns created by fossil fuels, and energy transformation has become a widespread occurrence. It is necessary to place a greater focus on research and innovation. More study is needed into energy storage technologies and sustainable energy alternatives. Because the global energy sector is so enormous, determining the appropriate technique for analyzing decarbonization implications is challenging. For thousands of years, fossil-based energy sources have been exploited, leading to a host of problems including resource depletion, pollution, and global warming. All these challenges require an energy transition from an existing fossil-based energy system to a new one based on renewable energy technologies and more efficient energy use. The role of renewable energy will become increasingly essential as LCOE costs fall, legislation changes, and acceptance of the energy transition knowledge grows. According to many study scenarios, solar and wind will play a major part in the energy transition as the two most evident key renewable energy technologies. This article shows the main technologies used to generate, re-electrify, and store energy that can be used to integrate renewable energy into the energy sector. On the other hand, this article highlights specific renewable energy system applications and also provides information on energy storage and technology. As the development and deployment of renewable energy systems increase, greater technical readiness will allow for the creation of the energy transition in the future.

References J.O. Abe et al., Hydrogen energy, economy and storage: Review and recommendation. Int. J. Hydrog. Energy (2019). https://doi.org/10.1016/j.ijhydene.2019.04.068 M. Aghaie, N. Rezaei, S. Zendehboudi, A systematic review on CO2 capture with ionic liquids: Current status and future prospects. Renew. Sust. Energ. Rev. (2018). https://doi.org/10.1016/j.rser.2018.07.004

1552

E. Deniz and M. S. Çelikta¸s

O.A. Al-Shahri et al., Solar photovoltaic energy optimization methods, challenges and issues: A comprehensive review. J. Clean. Prod. (2021). https://doi.org/10.1016/j.jclepro.2020.125465 A. Buchroithner et al., Design and experimental evaluation of a low-cost test rig for flywheel energy storage burst containment investigation. Appl. Sci. (Switzerland) 8(12) (2018). https://doi.org/10.3390/app8122622 M. Budt et al., A review on compressed air energy storage: Basic principles, past milestones and recent developments. Appl. Energy (2016). https://doi.org/10.1016/j.apenergy.2016.02.108 M.J. Burke, J.C. Stephens, Political power and renewable energy futures: A critical review. Energy Res. Soc. Sci. 35 (2018). https://doi.org/10.1016/j.erss.2017.10.018 R.J. Campbell, The smart grid: Status and outlook. Congressional Research Service (2018). https:/ /sgp.fas.org/crs/misc/R45156.pdf J.M. Carrasco et al., Power-electronic systems for the grid integration of renewable energy sources: A survey. IEEE Trans. Ind. Electron. (2006). https://doi.org/10.1109/TIE.2006.878356 M.S. Celiktas, F.M. Alptekin, Biorefinery concept: Current status and future prospects. in International Conference on Engineering Technologies, vol. 7–9(2017) (2019) M. Ceraolo, New dynamical models of lead-acid batteries. IEEE Trans. Power Syst. 15(4) (2000). https://doi.org/10.1109/59.898088 V. Chandran et al., Comprehensive review on recycling of spent lithium-ion batteries. Mater. Today: Proc. (2021). https://doi.org/10.1016/j.matpr.2021.03.744 H. Chen et al., Progress in electrical energy storage system: A critical review. Prog. Nat. Sci. (2009). https://doi.org/10.1016/j.pnsc.2008.07.014 J.W. Choi, D. Aurbach, Promise and reality of post-lithium-ion batteries with high energy densities. Nat. Rev. Mater. (2016). https://doi.org/10.1038/natrevmats.2016.13 I. Coban et al., Bioethanol production from raffinate phase of supercritical CO2 extracted Stevia rebaudiana leaves. Bioresour. Technol. 120 (2012). https://doi.org/10.1016/j.biortech.2012.06.057 D. Connolly et al., The first step towards a 100% renewable energy-system for Ireland. Appl. Energy 88(2) (2011). https://doi.org/10.1016/j.apenergy.2010.03.006 H. Dambeck, F. Ess, H. Falkenberg, A. Kemmler, A. Kirchner, M. Koepp, S. Kreidelmeyer, S. Lübbers, A. Piégsa, S. Scheffer, T. Spillmann, N. Thamling, A. Wünsch, M. Wünsch, I. Ziegenhagen, Towards a climate-neutral Germany by 2045, How Germany can reach its climate targets before 2050. Berlin (2021). Available at: https://static.agora-energiewende.de/fileadmin/ Projekte/2021/2021_04_KNDE45/A-EW_213_KNDE2045_Summary_EN_WEB.pdf S. Dunn, Hydrogen futures: Toward a sustainable energy system. Int. J. Hydrog. Energy 27(3) (2002). https://doi.org/10.1016/S0360-3199(01)00131-8 P.P. Edwards et al., Hydrogen and fuel cells: Towards a sustainable energy future. Energy Policy 36(12) (2008). https://doi.org/10.1016/j.enpol.2008.09.036 S. Emmanouil et al., Evaluating existing water supply reservoirs as small-scale pumped hydroelectric storage options – A case study in Connecticut. Energy 226 (2021). https://doi.org/10.1016/j.energy.2021.120354 Energy Information Administration (EIA), Energy information administration: Annual energy outlook 2018 with projections to 2050. J. Phys. A: Math. Theor. (2018). Available at: https:/ /www.eia.gov/outlooks/aeo/ A. Evans, V. Strezov, T.J. Evans, Assessment of utility energy storage options for increased renewable energy penetration. Renew. Sust. Energ. Rev. (2012). https://doi.org/10.1016/j.rser.2012.03.048 EWEA, Balancing responsibility and costs of wind power plants. EWEA – The European Wind Energy Association (2015) R. Fouquet, The slow search for solutions: Lessons from historical energy transitions by sector and service. Energy Policy 38(11), 6586–6596 (2010). https://doi.org/10.1016/j.enpol.2010.06.029 R. Fouquet, Historical energy transitions: Speed, prices and system transformation. Energy Res. Soc. Sci. 22 (2016). https://doi.org/10.1016/j.erss.2016.08.014 S. Ghasemian et al., An overview of global energy scenarios by 2040: Identifying the driving forces using cross-impact analysis method. Int. J. Environ. Sci. Technol. (2020). https://doi.org/10.1007/s13762-020-02738-5

Analysis of Energy Transition Pertaining to the Future Energy Systems

1553

J.B. Goodenough, Y. Kim, Challenges for rechargeable Li batteries. Chem. Mater. (2010). https://doi.org/10.1021/cm901452z J.B. Goodenough, K.S. Park, The Li-ion rechargeable battery: A perspective. J. Am. Chem. Soc. (2013). https://doi.org/10.1021/ja3091438 K. Gürtler, D. Löw Beer, J. Herberg, Scaling just transitions: Legitimation strategies in coal phase-out commissions in Canada and Germany. Polit. Geogr. 88 (2021). https://doi.org/10.1016/j.polgeo.2021.102406 E.S. Hanley, G. Amarandei, B.A. Glowacki, Potential of redox flow batteries and hydrogen as integrated storage for decentralized energy systems. Energy Fuels 30(2) (2016). https://doi.org/10.1021/acs.energyfuels.5b02805 D.J. Hess, Sustainability transitions: A political coalition perspective. Res. Policy 43(2) (2014). https://doi.org/10.1016/j.respol.2013.10.008 Y. Huang et al., Integration of compressed air energy storage with wind generation into the electricity grid, in IOP Conference Series: Earth and Environmental Science, (2018). https://doi.org/10.1088/1755-1315/188/1/012075 IEA, The future of hydrogen for G20. Report prepared by the IEA for the G20, Japan, 6(June) (2019) IEA, Global energy review 2020 – Analysis. IEA (IEA, Paris, 2020). Available at: https:// www.iea.org/reports/global-energy-review-2020 IEA, The role of critical minerals in clean energy transitions (2021). Available at: https:// www.iea.org/reports/the-role-of-critical-minerals-in-clean-energy-transitions IEA, Renewables 2020, Analysis and forecast to 2025 (n.d.). Available at: https://www.iea.org/ reports/renewables-2020 A. Ipakchi, F. Albuyeh, Grid of the future. IEEE Power Energy Mag. 7(2) (2009). https://doi.org/10.1109/MPE.2008.931384 IRENA, Innovation landscape for a renewable-powered future: Solutions to integrate variable renewables, International Renewable Energy Agency, Abu Dhabi. Abu Dhabi (2019). Available at: https://www.irena.org/publications/2019/Feb/Innovation-landscape-for-a-renewablepowered-future IRENA, Reaching zero with renewables: Eliminating CO2 emissions from industry and transport in line with the 1.5◦ C climate goal, IRENA. Abu Dhabi (2020a). Available at: https:// www.irena.org/publications/2020/Sep/Reaching-Zero-with-Renewables IRENA, Renewable power generation costs in 2019, International Renewable Energy Agency (2020b). Available at: https://www.irena.org/publications/2020/Jun/Renewable-Power-Costsin-2019 I.P. Jain, Hydrogen the fuel for 21st century. Int. J. Hydrog. Energy 34(17) (2009). https://doi.org/10.1016/j.ijhydene.2009.05.093 J. Jung, R.P. Broadwater, Current status and future advances for wind speed and power forecasting. Renew. Sust. Energ. Rev. (2014). https://doi.org/10.1016/j.rser.2013.12.054 T. Junne et al., How to assess the quality and transparency of energy scenarios: Results of a case study. Energ. Strat. Rev. 26 (2019). https://doi.org/10.1016/j.esr.2019.100380 A. Kazan et al., Bio-based fractions by hydrothermal treatment of olive pomace: Process optimization and evaluation. Energy Convers. Manag. 103 (2015). https://doi.org/10.1016/j.enconman.2015.06.084 F. Kern, A. Smith, Restructuring energy systems for sustainability? Energy transition policy in the Netherlands. Energy Policy 36(11) (2008). https://doi.org/10.1016/j.enpol.2008.06.018 L.T. Keyßer, M. Lenzen, 1.5 ◦ C degrowth scenarios suggest the need for new mitigation pathways. Nat. Commun. 12(1) (2021). https://doi.org/10.1038/s41467-021-22884-9 D. Larcher, J.M. Tarascon, Towards greener and more sustainable batteries for electrical energy storage. Nat. Chem. (2015). https://doi.org/10.1038/nchem.2085 E. Lewis et al., Intergovernmental panel on climate change. Sustainaspeak (2018). https://doi.org/10.4324/9781315270326-109 L. Li, X. Cao, P. Wang, Optimal coordination strategy for multiple distributed energy systems considering supply, demand, and price uncertainties. Energy 227 (2021). https://doi.org/10.1016/j.energy.2021.120460

1554 D.

E. Deniz and M. S. Çelikta¸s

Loorbach, Transition management for sustainable development: A prescriptive, complexity-based governance framework. Governance 23(1) (2010). https://doi.org/10.1111/j.1468-0491.2009.01471.x P.D. Lund et al., Review of energy system flexibility measures to enable high levels of variable renewable electricity. Renew. Sust. Energ. Rev. (2015). https://doi.org/10.1016/j.rser.2015.01.057 H. Lund et al., Energy storage and smart energy systems. Int. J. Sustain. Energy Plann. Manage. 11 (2016). https://doi.org/10.5278/ijsepm.2016.11.2 G.J. May, A. Davidson, B. Monahov, Lead batteries for utility energy storage: A review. J. Energy Storage (2018). https://doi.org/10.1016/j.est.2017.11.008 J. Meadowcroft, What about the politics? Sustainable development, transition management, and long term energy transitions. Policy. Sci. 42(4) (2009). https://doi.org/10.1007/s11077-009-9097-z N. Mikova, W. Eichhammer, B. Pfluger, Low-carbon energy scenarios 2050 in north-west European countries: Towards a more harmonised approach to achieve the EU targets. Energy Policy 130 (2019). https://doi.org/10.1016/j.enpol.2019.03.047 E. Mossali et al., Lithium-ion batteries towards circular economy: A literature review of opportunities and issues of recycling treatments. J. Environ. Manag. 264 (2020). https://doi.org/10.1016/j.jenvman.2020.110500 S.R. Mousavi Baygi, S.M. Sadrameli, Thermal management of photovoltaic solar cells using Polyethylene Glycol1000 (PEG1000) as a phase change material. Therm. Sci. Eng. Progr. 5, 405–411 (2018). https://doi.org/10.1016/j.tsep.2018.01.012. Elsevier Ltd NRG_IND_REN, Eurostat, Eurostat (2021). Available at: https://ec.europa.eu/eurostat/data browser/view/NRG_IND_REN/bookmark/bar?lang=en&bookmarkId=bdf882c7-1d2a-499b-ae 01-86ae5fab734c C. Ozay, M.S. Celiktas, Statistical analysis of wind speed using two-parameter Weibull distribution in AlaçatI region. Energy Convers. Manag. (2016). https://doi.org/10.1016/j.enconman.2016.05.026 S. Petrovic, Nickel–Cadmium Batteries. in Battery Technology Crash Course (2021). https:// doi.org/10.1007/978-3-030-57269-3_4 C. Redl et al., The European power sector in 2020: Up-to-date analysis on the electricity transition. Berlin (2021). Available at: https://www.agora-energiewende.de/en/press/news-archive/ renewables-overtake-gas-and-coal-and-coal-in-eu-electricity-generation-1/ T. Sasaki, Y. Ukyo, P. Novák, Memory effect in a lithium-ion battery. Nat. Mater. 12(6) (2013). https://doi.org/10.1038/nmat3623 S. Selvakkumaran, E.O. Ahlgren, Impacts of social innovation on local energy transitions: Diffusion of solar PV and alternative fuel vehicles in Sweden. Glob. Transit. 2 (2020). https://doi.org/10.1016/j.glt.2020.06.004 A.K. Shukla, S. Venugopalan, B. Hariprakash, Nickel-based rechargeable batteries. J. Power Sources 100(1–2) (2001). https://doi.org/10.1016/S0378-7753(01)00890-4 M. Simão, H.M. Ramos, Hybrid pumped hydro storage energy solutions towards wind and PV integration: Improvement on flexibility, reliability and energy costs. Water (Switzerland) 12(9) (2020). https://doi.org/10.3390/w12092457 G.L. Soloveichik, Battery technologies for large-scale stationary energy storage. Annu. Rev. Chem. Biomol. Eng. (2011). https://doi.org/10.1146/annurev-chembioeng-061010-114116 A. Stephan, L.D. Anadon, V.H. Hoffmann, How has external knowledge contributed to lithium-ion batteries for the energy transition? iScience 24(1) (2021). https://doi.org/10.1016/j.isci.2020.101995 J.M. Tarascon, M. Armand, Issues and challenges facing rechargeable lithium batteries. Nature (2001). https://doi.org/10.1038/35104644 M. Uyan et al., Bioconversion of hazelnut shell using near critical water pretreatment for second generation biofuel production. Fuel 273 (2020). https://doi.org/10.1016/j.fuel.2020.117641. Elsevier Ltd

Analysis of Energy Transition Pertaining to the Future Energy Systems

1555

A. Vicenzutti et al., Enhanced partial frequency variation starting of hydroelectric pumping units: Model based design and experimental validation. Int. J. Electr. Power Energy Syst. 131 (2021). https://doi.org/10.1016/j.ijepes.2021.107083 X. Wang et al., Flexible energy-storage devices: Design consideration and recent progress. Adv. Mater. (2014). https://doi.org/10.1002/adma.201400910 Z. Wen et al., Research on sodium sulfur battery for energy storage. Solid State Ionics 179(27–32) (2008). https://doi.org/10.1016/j.ssi.2008.01.070 S. Wicki, E.G. Hansen, Clean energy storage technology in the making: An innovation systems perspective on flywheel energy storage. J. Clean. Prod. 162 (2017). https://doi.org/10.1016/j.jclepro.2017.05.132 B. Widera, Renewable hydrogen implementations for combined energy storage, transportation and stationary applications. Therm. Sci. Eng. Progr. 16 (2020). https://doi.org/10.1016/j.tsep.2019.100460 M. Xiao, S. Simon, T. Pregger, Scenario analysis of energy system transition – A case study of two coastal metropolitan regions, eastern China. Energ. Strat. Rev. 26 (2019). https://doi.org/10.1016/j.esr.2019.100423 C. Zheng, D.M. Kammen, An innovation-focused roadmap for a sustainable global photovoltaic industry. Energy Policy 67 (2014). https://doi.org/10.1016/j.enpol.2013.12.006

Data Mining Applications in Smart Grid System (SGS) Mohammad Taghi Dehghan Nezhad and Mohammad mahdi Sarbishegi

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Data Mining Applications in SGS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Data Mining in Demand Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Data Mining in Transmission and Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Data Mining in Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Data Mining in Predictive Maintenance and Fault Management (PM&FM) . . . . . 3 Processing Software of Data Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Data Mining Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Data Visualization Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Challenges of Using Data Mining in SGS and Future Works . . . . . . . . . . . . . . . . . . . . . . 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1558 1559 1560 1562 1564 1565 1566 1567 1568 1568 1569 1569 1569

Abstract

Smart grid system (SGS) is a state-of-the-art technology that has been invented for the creation of a smart relation between consumers and producers then handled by a central center to control and manage the grid. Data mining is a useful tool, and it has been almost used in all industries and researches to including SGS and power industry. Data mining has profit methods such as classification, clustering regression, time series, and they are helping to SGS for performing the task to defined for its. This chapter introduces a M. T. Dehghan Nezhad () Sharif University, Tehran, Iran e-mail: [email protected] M. m. Sarbishegi Tehran University, Tehran, Iran e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_142

1557

1558

M. T. Dehghan Nezhad and M. m. Sarbishegi

comprehensive survey of applications of data mining in SGS. First, survey data mining applications in demand response, for example, load forecasting, generate forecasting, smoothing the consumption of power curve. Second, survey data mining applications in distribution and transmission of power in SGS such as detecting the power transmission line faults and islanding situation and power quality measurement. Third, survey data mining applications in the security of SGS such as finding anomaly, false data injection, and DoS (or DDoS), and then in fourth, survey data mining applications in predictive maintenance and fault management (PM &FM) such as an outage, equipment breakdown, electrical and arc faults. After this data mining applications, survey the best of data mining and visualization software and compare them with each other and introduce the strengths of every software in data mining methods. In the end, survey challenges of today’s data mining applications in SGS and then give some suggestion for future works. Keywords

Smart grid system · Data mining · Machine learning · Demand response · Security · Distribution

1

Introduction

Phuangpornpitak and Tia (2013) defines the SGS as: SGS is the key for an efficient use of distributed energy resources. According to the US Department of Energy (2009): “A smart grid uses digital technology to improve reliability, security, and efficiency (both economic and energy) of the electrical system from large generation, through the delivery systems to electricity consumers and a growing number of distributed-generation and storage resources.” Smart grid is a two-way dialogue where electricity and information can be exchanged between the utility and its customers. New data is always created in the SGS, and in addition to storing and transporting, analyze them is an important step to gain certain knowledge or facilitate certain decision-making. Mitchell (1997) defines data mining as using historical data to discover regularities and improve future decisions. Data mining is a process which discover patterns in data with statistic and machine learning tools (Chakrabarti 2006). Machine learning is a part of AI which study computer algorithms that can improve automatically through experience and by the use of data (Mitchell 1997). Machine learning is categorized in four categories: supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning. In supervised learning, both inputs and outputs are available (Russell and Norvig 2002) and the goal is to reach function which predicts the output associated with new inputs. Supervised applications contain classification and regression. In unsupervised learning, input data without output is given to the model, and the objective is to find structure in the data. A central application of unsupervised

Data Mining Applications in Smart Grid System (SGS)

1559

learning is in the field of density estimation in statistic (Jordan and Bishop 2004). Another important application of unsupervised learning is clustering (cluster analysis) and we will explain about it in the following sections. In semi-supervised learning, model trains based on small amount of labeled data and large amount of unlabeled data. Semi-supervised learning contrasts unsupervised learning for unlabeled or all data and supervised learning for labeled ones. Reinforcement learning is the problem faced by an agent that must learn behavior through trial-and-error interactions with a dynamic environment (Kaelbling et al. 1996). Over-fitting is one of the major problems in machine learning. Over-fitting is a situation in which training error is low, but it is failing to generalize to develop and test set. Many techniques have been developed to combat over-fitting such as: k-fold cross-validation, feature selection, dropout, and regularization (Ng 2017). Machine learning is an import tool in data mining and employs in data mining methods as unsupervised and supervised learning. In this chapter, applications of data mining in the SGS will be covered.

2

Data Mining Applications in SGS

Data mining or data analytics is a set of methods for uncovering the knowledge and patterns of data that lie on databases. Data mining has two important sections: preprocessing and processing data. In preprocessing data, methods such as data cleaning, data integration, and data reduction are used. Data cleaning: to detect and then change or remove incorrect or corrupted records from dataset (Wu 2013). Data integration: to combine and integrate data residing in different sources, databases, or files (Lenzerini 2002). Data reduction: to obtain reduced representation in volume but produces the same or similar analytical results. It may be in the form of dimensionality reduction (encoding), numerosity reduction, or data compression (Aung 2013). In reference French (1996), data processing is defined as: “the collection and manipulation of items of data to produce meaningful information.” Data processing is closely related to database management, machine learning, and statistics. The most common data process applications are clustering, regression, time series methods, and classification. Clustering: to find structure and organization of a dataset. Clustering is an unsupervised learning process which classes and labels are unknown. One of the important hyper-parameter of clustering algorithms is number of classes and groups. Some popular clustering algorithms are k-means, fuzzy c-means, expectation maximization, DBSCAN, BIRCH, and hierarchical clustering (Aung 2013). The clustering process mainly includes two steps: select clustering features and choose

1560

M. T. Dehghan Nezhad and M. m. Sarbishegi

the clustering algorithm (Dou 2018). K-means is an important clustering algorithm that cluster data to k groups based on the distance of each data from center of k groups. Regression: to predict value of data, in another way regression models estimate the relationship between inputs and outputs. Regression is a supervised learning algorithm. Some of the popular regression models are Gauss–Newton algorithm, logistic regression, linear regression, and k-nearest-neighbor. Deep learning models such as NN and CNN are also used as regression models, but they need more data and their complexity is high. Time series: to forecast future based on previous data. Time series models are supervised learning. There are many similarities between regression and time series analysis; in some references, time series models are a subset of regression models. But time series analysis predicts future values based on previous data, but regression finds relationship between inputs and outputs so they are in different ways. One of the important concepts in time series models is stationarity. The basic idea of stationarity is that the probability laws that govern the behavior of the process do not change over time (Cryer 1986). Some of the popular time series models are moving average, exponential smoothing, and ARIMA. Classification: classification is a supervised learning process which classifies dataset into predefined groups. Some popular classification algorithms are decision tree, Naive Bayes, hidden Markov model, support vector machine, and k-nearest neighbors (Aung 2013). Deep learning models also can be used for classify data, such NN and CNN. Deep learning models are more accurate, but they need more data and stronger processor (GPU and TPU).

2.1

Data Mining in Demand Response

According to Federal Energy Regulatory Commission, demand response is defined as: “Changes in electric usage by end-use customers from their normal consumption patterns in response to changes in the price of electricity over time, or to incentive payments designed to induce lower electricity use at times of high wholesale market prices or when system reliability is jeopardized.” According to definition, demand response has an important role to play in the electricity market for maintaining the balance between supply and demand by introducing load flexibility instead of only adjusting generation levels, at almost all operational time scales (Balijepalli et al. 2011). Demand response analysis can reduce system peak load in the very short-term load forecasting (around 24 h), short-term load forecasting (around 2 weeks), medium-term load forecasting (around 3 years), and long-term load forecasting (around 30 years) Hong (2010) developed regression and time series based models to predict load and reduce peak load, and also he defined five group of business which need load forecasts. Time has an important role in demand response forecasting so it is obvious that time series algorithms are used in most of related works.

Data Mining Applications in Smart Grid System (SGS)

1561

Demand response methods have two main categories (Babaei et al. 2021): timebased and incentive-based methods. In time-based models, the time of appliances usage changes according to time-based tariffs aiming at minimizing the electricity price of consumer. In incentive-based models, we encourage consumers to manage their consumption. In another article (Nguyen and Yousefi 2011), authors proposed a method for optimally selecting the locations of demand response to achieve a number of objectives like maximizing transfer capacity, reducing energy waste, and minimizing total demand response program capacity. They use optimization methods. In reference Derakhshan et al. (2016), two optimization methods are used for optimizing price-based demand response according to classify demands into three categories: non-sheddable loads, sheddable loads, and shiftable loads. Nonsheddable loads are defined as loads that consumers are not able to turn them off, like refrigerators. Sheddables loads are defined as loads that consumers can turn them off in the hours of day, like computers, and shiftable loads are defined as loads that their consumption during the day is mandatory but no certain time for use of them is specified and (Babaei et al. 2021) use clustering method (k-means) to choose controllable appliances according to their consumptions. Also defined methods called shift method and reduce method to optimize demand response in SGS. Luo et al. (2019) developed a data mining-driven incentive-based demand response. Their model is made of three modules: data management module, data mining module, and decision-making module. The data mining module features two tasks as follows: (1) Consumer clustering: to partition initial bid-offer samples into a certain number of classes with OPTICS algorithm and (2) Consumer classification: to establish a classification model by which new consumers can be assigned to one of the classes identified in clustering, by reference to their bid-offers by k-nearest-neighbor. Big offer consumer defined as a vector which contains curtailment capacity and big price of consumer. In reference Park et al. (2014), authors introduced a framework for baseline load estimation in demand response. The proposed framework is based on unsupervised learning for data mining. After preprocess data with data selection, data cleaning, and data reduction, they used two clustering methods. In data selection, more significant data is selected to process from the initial database. Significant data is made according to the power level of the consumers. In data cleaning, outliers’ data that distorted the information about the customer facilities are found and removed. In data reduction, load conditions based on season and type of weekday are classified and electricity consumption pattern of each day is characterized by a single curve called load vector. In clustering, data is clustered with two models based on their load vectors: self-organizing map which is an AI clustering method and k-means. In another paper (Fan 2011), a distributed framework had been proposed for demand response and user adaptation in smart grid networks. In this framework, individual users adapt to the price signals to maximize their own benefits. They considered a fully distributed system where the only information available to the end users is the current price which is dependent on the overall system load so users reduce peak load by load scheduling.

1562

M. T. Dehghan Nezhad and M. m. Sarbishegi

Accurate renewable generation and load forecasting play a key role in the progress of power grid planning schemes. In Dou et al. (2018), a hybrid shortterm forecasting method based on clustering and variational mode decomposition technique is proposed to deal with the problem of forecasting accuracy. The proposed method consists of data preprocessing, data mining, VMD decomposition, and forecasting model. In data preprocessing, the original data normalized and invalid data substituted through mean value method. In data mining segment, 2 k-means clustering method applied to partition data and optimal clusters are selected by correlation coefficient as the training set. First k-means is used to cluster data according to meteorological factors, and second kmeans is applied to classify data based on historical power. Variational mode decomposition (VMD) is a time series analysis method which decomposes a real valued input signal into a discrete number of sub-signals (Dragomiretskiy and Zosso 2013). After using VMD to detect sub-signals, learning machine employed to predict the renewable energy resources power generation and loads. Similarly, Lorenzo Navarro et al. (2011) proposed an architecture for an estimator of short-term wind farm power. They applied k-means clustering to split the input space into subspaces and then training a multilayer perceptron for each cluster. The linear machine classifies the samples into one of several subsets which has been previously obtained with a clustering analysis. In another paper (Deng and Jirutitijaroen 2010), short-term Singapore electricity demand forecasted with two time series model: the multiplicative decomposition model and the seasonal ARIMA model. Demand response depends on wide range of sectors. For example, in peak load, we can predict consumption of energy with time series tools and turn off some unnecessary equipment of consumers, or we can increase energy generation in SGS.

2.2

Data Mining in Transmission and Distribution

Generated power is fed into step-up transformer and intensified into high voltage, and is transmitted to substations closer to the customers. At each substation, the high voltage electricity is reconstructed to a low state, which is best suited for real time consumption. Distributed control is propriety control, and monitoring systems are used in power transmission and distribution (Shyam et al. 2015). Fault identification and classification are necessary to ensure stable and reliable operation of the power system. The feeders of the grid are critical and have a significant failure rate. In reference Gross et al. (2006), authors developed machine learning system to ranking primary distribution feeders according to their susceptibility to failure. They used SVM and linear regression models for the sake of comparison and to produce in real time a list of the network’s feeders sorted from most to least susceptible to failure. Also, martingale boosting technique is applied to increase accuracy. Boosting is a machine learning technique that converts weak learning models with low individual accuracy into a powerful single model with high predictive

Data Mining Applications in Smart Grid System (SGS)

1563

accuracy (Freund and Schapire 1997). In another paper (Kirubadevi et al. 2017), discrete wavelet transform based signal analysis is applied to identify and classify the type of transmission line faults. Wavelet transform is a function that divides a continuous signal into different components. Wavelet transform is very efficient for the analysis of nonstationary signals with local transients. Zhao et al. (2019) developed a novel fault section location method based on random forest. In the training process, the relevance between the fault attributes and the fault section location is mined by the random forest algorithm. Random forest is a classification or regression supervised learning model that constructs a multitude of decision trees at training time. In the preprocessing stage, the following two techniques were performed: the oversampling and the feature selection. After preprocessing, a random forest model is trained with sample sets for tuning related parameters. Also, the random sampling method applied which could partly solve the problem of over fitting. One of the important issues in SGS is islanding situation. Islanding is the situation in which a distributed system becomes electrically isolated from the other distribution systems of the power system, but continues to be energized by distributed generation that is connected to it (Mahat et al. 2008). Many data mining models have been developed to detect this phenomenon. Samantaray et al. (2010) proposed a fuzzy rule-based approach for islanding detection in distributed generation. In the proposed approach, two major steps are involved: features extraction and classification for islanding detection. In features extraction, 11 features were chosen and defined for any target distributed resource from distributed generation. In the classification step, decision tree is used for initial classification, and after training process, the decision tree is transformed to a fuzzy rule base by developing the fuzzy membership functions. Another paper (Najy et al. 2011) developed islanding detection model using Naive Bayes classifier. K-fold cross-validation is also applied to prevent over-fitting. Naïve Bayes classifier is a statistical classification method based on Bayes’ theorem, assuming that all features are independent in a given class (Rish 2001). Lidula et al. (2009) investigated three classification techniques for fast detection of power islands in a distribution network: decision trees, probabilistic neural networks, and support vector machines. Wavelet transform is also used to decompose currents and voltages signals into several signals in different frequency bands. And finally, k-fold cross-validation is applied. The best classification according to the average overall accuracy was achieved from the decision tree classifier. Power quality is also an important issue in the power system. He and Starzyk (2005) proposed power quality disturbances classification based on the wavelet transform and self-organizing learning array system. They used wavelet transform for feature extraction. A system for the identification of power quality violations is proposed (Elmitwally et al. 2001). Wavelet multiresolution signal analysis is exploited to de-noise and then decompose the monitored signals, and then artificial neural networks are trained to detect power quality violations. Ibrahim and Morcos (2002) provided a survey in artificial intelligence applications like neural networks and mathematical tools like wavelet analysis for power quality applications.

1564

2.3

M. T. Dehghan Nezhad and M. m. Sarbishegi

Data Mining in Security

Security is one of the most important features in SGS, because a lot of nodes connect to each other, and it is possible be a hacker or cracker in the network (Faquir et al. 2021). The kind of security of SGS is divided into user privacy, smart meter security, and protection from cyberattacks (Baumeister 2010). A popular trend of researches in the three kinds of security of SGS are anomaly detection, false detection, denialof-service (Dos)attack protection, and in all of them, data mining has key roles.

2.3.1 Anomaly Detection When a point (global and contextual outliers) or a bath of points (collective outliers) (Aggarwal 2021) behave out of normal, happen anomaly. Sometimes, it has a logical reason, but another time it is a fraud or attack, and should stop it. For example, a sale website must sell 2000–2200$ in every Monday, but on the specific Monday, its amount of sale is 100 or 5000$, and this is determined an anomaly (Coha 2021), although it is possible that this Monday is near to Charismas and be a logical anomaly. Some of the related work about data mining application in anomaly detection in SGS: Lambda architecture based on supervised learning and statistical (Liu and Nielsen 2016) and simple clustering (Quinde et al. 2021) and clustering silhouette thresholding (Rossi et al. 2016) were methods to analyze real-time data and mining algorithms on a data stream. Also, they were used in parallel processing systems and collective and contextual anomaly detection. Anwar and Mahmood (2016) formulated the problem as quadratic assignment problem, and graph comparison-based matching is an approach for anomaly detection in the database and solve quadratic assignment problem, and Singh and Govindarasu (2018) proposed decision tree for discovering malicious tripping attack in remedial action scheme. One of the classic machine learning algorithms named neural network (Ford et al. 2014) and one of the modern machine learning algorithms named recurrent neural network that implements in encoder-decoder framework and reconstructed its model with time series (Fengming et al. 2017) are proposed for finding anomaly and fraud in input data. 2.3.2 False Data Injection When some of the data measurements and smart meters send false data to center control network (in distribution network send to nodes of the network), it is called false data injection (Liu et al. 2011), and it is a type of data integrity attack (Anwar et al. 2022). False data injection has a long history and a lot of traditional solutions proposed against it, but Esmalifalak et al. (2014) claimed traditional models (such as traditional state estimation) cannot detect a new type of attack such as stealth attack that an attacker does not leave any trace. Therefore, after 2014, some of the researchers migrate from traditional models and algorithms to new approaches

Data Mining Applications in Smart Grid System (SGS)

1565

based on data mining and machine learning or at least, combined traditional models with new approaches, to specify false data injection. Some of the related work about data mining application in false data injection (FDI) detection in SGS: Esmalifalak et al. (2014) proposed a distributed super vector machine for false data detection. Liao and Chakrabortty (2016) proposed a model that detects FDI and identities FDI attacker using distributed optimization, and Anwar et al. (2022) proposed measurement-driven blind topology estimation for false data injection attack that both of them were based on alternating direction multiplier method (ADMM). Cui et al. (2020) categorize FDI into three categories: (1) nontechnical loss such as Messinis et al. (2019) and Jokar et al. (2016) used from super vector machine (SVM) algorithm, Jindal et al. (2016) used SVM and decision tree, and also, Costa et al. (2013) used from ANN and Liu and Nielsen (2016) used from PARX that is regression-based method; (2) state estimation such as Yan et al. (2016) used from SVM and KNN, and Jacob Sakhnini (2019) used from SVM and KNN and ANN; and (3) load forecasting such as MLR and NN algorithm (Liang et al. 2019) and Naive Bayes (Cui et al. 2019).

2.3.3 Denial-of-Service (DoS) DoS or distributed DoS (DDoS) is a prevalent attack on the network, and they are engaging servers and computation nodes, then creating problems for legitimate users of the network (Tama and Rhee 2015; Bhuyan et al. 2014). Some of the related work about data mining applications in DoS and DDoS detection in SGS: Tama and Rhee (2015) introduce some data mining researches in DoS/DDoS attack detection (not just in SGS), then categorize them in classification, clustering, association, and hybrid. Afterward, they showed SVM is the most used commonly technique in DoS detection. Furthermore, Zhe et al. (2020) first reduced data dimension with PCA algorithm, second with test naïve base and DT and SVM, chose SVM, because it is best of them in their test. Choi et al. (2012) used from decision tree (DT) for SYN flood attack (a form of DoS), and Boumkheld et al. (2016) developed an intrusion detection system (IDS) and used from Naive Bayes classification for its system (nevertheless, their system was centralized and cause privacy violation), then both of them ran their data mining technique on data set using the Waikato Environment for Knowledge Analysis (WEKA), which proved their approach was good.

2.4

Data Mining in Predictive Maintenance and Fault Management (PM&FM)

PM&FM have a key role in SGS, and the importance of them has increased with the Industry 4 revolution (Mahmoud et al. 2021). PM&FM decreased synchronization issues such as outage faults, equipment breakdown, electrical SGS faults, and arc faults.

1566

M. T. Dehghan Nezhad and M. m. Sarbishegi

2.4.1 Power Outage Power outage is a term for not receiving power by users (Dehbozorgi et al. 2021). By surveying the factors in outage such as vegetation, animal, and equipment, Bashkari et al. (2020) tried to find the dominant factor (or the dominant factors). First, they showed the impact of the features through data visualization, and second, with association rule techniques, they found the factors correlated with the outage. Momentary outages lead to long-term permanent outages, and analyzing momentary outages by using the Apriori algorithm can reduce the probability of permanent outages (Dehbozorgi et al. 2021). Another approach for detecting power outages is using social media data such as Twitter and Facebook. When happened a power outage in a region, the region’s users would be offline and did not any post. (Sun et al. 2016) with developing a probabilistic framework, analyzing big data in Twitter in real-time, and then with a heterogeneous information network, a supervised topic model, and analyzing words of tweets, improved accuracy of outage detection. 2.4.2 Equipment Breakdown Equipment breakdowns is created by network disorder factors such as irregular loads or physical factors such as cracks in insulation or relays (Ullah et al. 2017; Wang et al. 2017; Mahmoud et al. 2021) with improved drosophila optimization algorithm and (Ullah et al. 2017; Mahmoud et al. 2021) by Infrared Thermography with using a multilayered perceptron predicted Equipment breakdown rate and found defect components. 2.4.3 Electrical SGS Faults Electrical SGS faults such as synchronize voltage and power accidents are another part of PM&FM. Stefenon et al. (2020) and Mahmoud et al. (2021) proposed wavelet energy coefficient (WEC) for feature extraction and group method of data handling (GMDH), and Zarei et al. (2019) with SVM and Diao et al. (2009) with DT solved some parts of electrical SGS faults. 2.4.4 Arc Faults Arc faults are one of the most common reasons for fire accidents in a variety of industry. Arc faults usually happen because of damaged wire or poor connection, and discrete wavelet transform (DWT) is an approach for detection of arc faults in SGS (Qi et al. 2017; Mahmoud et al. 2021).

3

Processing Software of Data Mining

The previous sections described the application of data mining and the kind of its in SGS. Now, we will introduce some of the popular tools for data mining and some of the popular tools for data visualization with focus on the SGS.

Data Mining Applications in Smart Grid System (SGS)

3.1

1567

Data Mining Tools

Tools of data mining are as important as algorithms of data mining. Each data mining tool is used in a specific field (or fields) and nobody cannot tell a software is better than other software. Data mining tools divide into open-source and proprietary, and sometimes used in enterprise and sometimes used for personal application (Sargam 2021). Data mining open-source software has no cost and is good for small and medium enterprises and everyone wants to be familiar with state-of-the-art data mining algorithms (Chen et al. 2007). Researchers and data analysts with the understanding approach of his using software, can tune and improve his code based on the structure of software, and it is another advantage of open-source (Altalhi et al. 2017; Chen et al. 2007). In this section, first introduced five popular data mining open-source tools and then are compared with each other in Table 1. Knime: Konstanz Information Miner (Knime) is a graphical user interface tool based on Java language and is developed by a software company specialized in pharmaceutical applications (Santos-Pereira et al. 2021) and has the ability in preprocessing, data cleaning, etc. (Altalhi et al. 2017). WEKA: Waikato Environment for Knowledge Analysis (WEKA) is a visualization and data mining tool, that developed with java language, and it has graphical user interface. It is notably used amongst researchers (Altalhi et al. 2017; Chen et al. 2007), for example, Choi et al. (2012) and Boumkheld et al. (2016) used WEKA to run code of DoS detection algorithms. Rapid miner: Rapid miner (also named YALE) is similar to WEKA and Knime and is based on Java language and graphical user interface. It also has application in industrial and researches, and support from machine learning and text mining algorithms (Altalhi et al. 2017). Orange: Orange, unlike the previous above three software, is developed with python language and support from classic machine learning (Altalhi et al. 2017; Chen et al. 2007). Apache spark: SGS collect huge data from sensors and monitoring in real time, therefore a big data platform and technology infrastructures need to analyze it. Shyam et al. (2015) present Apache spark as a unified cluster computing platform that is suitable for storing and performing big data analytics on smart grid data

Table 1 Comparison between data mining software based on release date of each method and 5v model of Zhang et al. (2017). 5V are volume, variety, velocity, veracity, and value Methods Rapid miner Orange WEKA Knime Apache spark

Classification +++ ++ +++ ++ +++

Clustering +++ ++ ++ ++ +++

Regression +++ +++ +++ +++ +++

Time series ++ + + ++ +++

1568

M. T. Dehghan Nezhad and M. m. Sarbishegi

for applications like automatic demand response and real-time pricing. The main types of processing techniques employed in big data analysis are batch, stream, and iterative processing.

3.2

Data Visualization Tools

Visualization is a technique for providing insight from data by images, diagrams, or animations and usually used for getting a brief awareness from the system in real time (Lundstrom et al. 2015; Wikipedia 2021). The Sect. 2.3 presented some data mining algorithms for control and monitoring network and anomaly detection in networks. Moreover, for managing and monitoring in the network, we can use dashboards in data visualization to observe real-time changes in SGS (dashboard usually employs bar chart, heatmap, etc.), and for anomaly can overview anomaly with using charts such as scatter plot, box plot, cat plot, etc. (Mukherjee et al. 2015; Lundstrom et al. 2015; Bashkari et al. 2020). However, state-of-the-art data mining software introduced in the Sect. 3.1 have a collection of visualization tools, but for especially application visualization, it is better to use from especially data visualization software: Google chart: A free data visualization tool developed by Google; it pulls data from various sources and it can work without any coding. It can embed into the web. Sisense: A proprietary business intelligence-based data visualization system, and it is focused on the companies, its primary application is creating dashboards, and it has fantastic ability to permit developers for creating API. Tableau: A popular data visualization tool, and it is not free (but tableau pubic is free). It is usually used at an enterprise level, but due to its simplicity, the normal user also can use it, and it has an important feature called high security. Microsoft Power BI: A proprietary data visualization platform to interact with excel. It is similar to Tableau, usually used at an enterprise level. It can give output in various formats (Harkiran78 2021; Velarde 2021; Stobierski 2021).

4

Challenges of Using Data Mining in SGS and Future Works

The current system is dependent on the central center in the smart grid system, and it has primary problems such as privacy violation and poor security. In addition, as described in the Sect. 1, scalability was an important issue in SGS, and when SGS is dependent on the central center and integrate data from different sources, this feature is unlikely, because real-time processing cannot be accomplished and also, the probability of error and mistake is high. As a result, it is possible to analyze data that give an inverse outcome and makes a wrong decision, and then data mining algorithms are useless. Another reason for the inefficiency of creating a central SGS is genesis a monopoly in wide of the system and also if wanted create SGS in the level of international, then governess do not coworker with each other and SGS will fail.

Data Mining Applications in Smart Grid System (SGS)

1569

(Saputra et al. 2019; Taïk and Cherkaoui 2020) For solving this problem, combined federated learning with data mining algorithms on SGS, but this problem still stands and do not removed, because, however, federated learning has distributed processing, but at the end of the process, aggregate parameters of the data model in a central agency. Future works about data mining algorithms on SGS should be developed based on distribution from A to Z. Some useful technology for a distributed system is blockchain and the Ethereum platform, and they are among the best platforms for implementing the distributed system.

5

Conclusion

SGS is a significant system, because it influences various industries and its impact is at the macro level. And after which, in the future, millions (or even billions) lives will be woven with SGS. This article investigated data mining applications in all dimensions of SGS, then introduced the best of software for data mining in SGS, and in the end acquaints some challenges of central center in SGS and problems to creation the scalability and security and privacy in SGS.

6

Cross-References

Applications of Data Envelopment Analysis (DEA) for Optimizing Energy

Consumptions Applications of Machine Learning for Renewable Energy: Issues, Challenges, and

Future Directions Big Data Applications for Improving the Reliability of the French Electricity

Distribution Grid Data Analytics Applications in the Energy Systems Concerning Sustainability Data-Driven Techniques for Optimizing the Renewable Energy Systems Opera-

tions

References R. Aggarwal, Types of outliers in data mining – GeeksforGeeks (2021). Retrieved 24 Dec 2021, from https://www.geeksforgeeks.org/types-of-outliers-in-data-mining/ A.H. Altalhi, J.M. Luna, M.A. Vallejo, S. Ventura, Evaluation and comparison of open source software suites for data mining and knowledge discovery. WIREs Data Min. Knowl. Discov. 7(3), e1204 (2017) A. Anwar, A.N. Mahmood, Anomaly detection in electric network database of smart grid: graph matching approach. Electr. Power Syst. Res. 133, 51–62 (2016) A. Anwar, A.N. Mahmood, Z. Tari, A. Kalam, Measurement-driven blind topology estimation for sparse data injection attack in energy system. Electr. Power Syst. Res. 202, 107593 (2022) Z. Aung, Database systems for the smart grid, in Smart Grids, (Springer, London, 2013), pp. 151–168

1570

M. T. Dehghan Nezhad and M. m. Sarbishegi

M. Babaei, A. Abazari, M.M. Soleymani, M. Ghafouri, S.M. Muyeen, M.T. Beheshti, A datamining based optimal demand response program for smart home with energy storages and electric vehicles. J. Energy Storage 36, 102407 (2021) V.M. Balijepalli, V. Pradhan, S.A. Khaparde, R.M. Shereef, in Review of Demand Response Under Smart Grid Paradigm. ISGT2011–India (IEEE, 2011, December), pp. 236–243 M.S. Bashkari, A. Sami, M. Rastegar, Outage cause detection in power distribution systems based on data mining. IEEE Trans. Ind. Inf. 17(1), 640–649 (2020) T. Baumeister, Literature review on smart grid cyber security. Collaborative Software Development Laboratory at the University of Hawaii (2010), p. 650 M.H. Bhuyan, H.J. Kashyap, D.K. Bhattacharyya, J.K. Kalita, Detecting distributed denial of service attacks: methods, tools and future directions. Comput. J. 57(4), 537–556 (2014) N. Boumkheld, M. Ghogho, M. El Koutbi, in Intrusion Detection System for the Detection of Blackhole Attacks in a Smart Grid. 2016 4th International Symposium on Computational and Business Intelligence (ISCBI) (IEEE, 2016, September), pp. 108–111 S. Chakrabarti, M. Ester, U. Fayyad, J. Gehrke, J. Han, S. Morishita, W. Wang, Data mining curriculum: A proposal (Version 1.0). Intensive Working Group of ACM SIGKDD Curriculum Committee, 140, 1–10 (2006) X. Chen, Y. Ye, G. Williams, X. Xu, in A Survey of Open Source Data Mining Systems. PacificAsia Conference on Knowledge Discovery and Data Mining (Springer, Berlin/Heidelberg, May 2007), pp. 3–14 K. Choi, X. Chen, S. Li, M. Kim, K. Chae, J. Na, Intrusion detection of NSM based DoS attacks using data mining in smart grid. Energies 5(10), 4091–4109 (2012) I. Coha, What is anomaly detection? | Anodot (2021). Retrieved 24 Dec 2021, from https:// www.anodot.com/blog/what-is-anomaly-detection/ B.C. Costa, B.L. Alberto, A.M. Portela, W. Maduro, E.O. Eler, Fraud detection in electric power distribution networks using an ANN-based knowledge-discovery process. Int. J. Artif. Intell. Appl. 4(6), 17 (2013) J.D. Cryer, Time Series Analysis, vol 286 (Duxbury Press, Boston, 1986) M. Cui, J. Wang, M. Yue, Machine learning based anomaly detection for load forecasting under cyberattacks. IEEE Trans. Smart Grid 10(5), 5724–5734 (2019) L. Cui, Y. Qu, L. Gao, G. Xie, S. Yu, Detecting false data attacks using machine learning techniques in smart grid: a survey. J. Netw. Comput. Appl. 2020, 102808 (2020) M.R. Dehbozorgi, M. Rastegar, A. Sami, Data mining-based cause identification of momentary outages in power distribution systems. Sustain. Cities Soc. 2021, 103587 (2021) Department of Energy, U.S. (2009) Smart grid system report. Available via Online. http:// www.doe.energy.gov/. Cited 30 Jan 2013 G. Derakhshan, H. A. Shayanfar, A. Kazemi, The optimization of demand response programs in smart grids. Energy Policy, 94, 295–306 (2016) J. Deng, P. Jirutitijaroen, in Short-Term Load Forecasting Using Time Series Analysis: A Case Study for Singapore. Proceedings of the 2010 IEEE Conference on Cybernetics and Intelligent Systems (CIS) (2010), pp. 231–236 R. Diao, K. Sun, V. Vittal, R.J. O’Keefe, M.R. Richardson, N. Bhatt, . . . S.K. Sarawgi, Decision tree-based online voltage security assessment using PMU measurements. IEEE Trans. Power Syst.24(2), 832–839 (2009) C. Dou, Y. Zheng, D. Yue, Z. Zhang, K. Ma, Hybrid model for renewable energy and loads prediction based on data mining and variational mode decomposition. IET Gener. Transm. Distrib. 12(11), 2642–2649 (2018) K. Dragomiretskiy, D. Zosso, Variational mode decomposition. IEEE Trans. Sig. Process. 62(3), 531–544 (2013) A. Elmitwally, S. Farghal, M. Kandil, S. Abdelkader, M. Elkateb, Proposed wavelet-neurofuzzy combined system for power quality violations detection and diagnosis. IEE Proc. Gener. Transm. Distrib. 148(1), 15–20 (2001) M. Esmalifalak, L. Liu, N. Nguyen, R. Zheng, Z. Han, Detecting stealthy false data injection using machine learning in smart grid. IEEE Syst. J. 11(3), 1644–1652 (2014)

Data Mining Applications in Smart Grid System (SGS)

1571

Z. Fan, in Distributed Demand Response and User Adaptation in Smart Grids. Proceedings of the 2011 IFIP/IEEE International Symposium on Integrated Network Management (IM) (2011), pp. 726–729 D. Faquir, N. Chouliaras, V. Sofia, K. Olga, L. Maglaras, Cybersecurity in smart grids, challenges and solutions. AIMS Electron. Electr. Eng. 5(1), 24–37 (2021) Z. Fengming, L. Shufang, G. Zhimin, W. Bo, T. Shiming, P. Mingming, Anomaly detection in smart grid based on encoder-decoder framework with recurrent neural network. J. China Univ. Posts Telecommun. 24(6), 67–73 (2017) V. Ford, A. Siraj, W. Eberle, in Smart Grid Energy Fraud Detection Using Artificial Neural Networks. 2014 IEEE Symposium on Computational Intelligence Applications in Smart Grid (CIASG) (IEEE, December 2014), pp. 1–6 C. S. French, Data Processing and Information Technology. Continuum. (1996) Y. Freund, R. Schapire, A decision-theoretic generalization of on-line learning and an application to boosting. J. Comput. Syst. Sci. 55(1), 119–139 (1997) P. Gross, A. Boulanger, M. Arias, D. L. Waltz, P. M. Long, C. Lawson, A. Kressner, (2006). Predicting electricity distribution feeder failures using machine learning susceptibility analysis. In AAAI (pp. 1705–1711) Harkiran78, 10 Best data visualization tools in 2020 – GeeksforGeeks (2021). Retrieved 28 Dec 2021, from https://www.geeksforgeeks.org/10-best-data-visualization-tools-in-2020/ H. He, J.A. Starzyk, A self-organizing learning array system for power quality classification based on wavelet transform. IEEE Trans. Power Deliv. 21(1), 286–295 (2005) T. Hong, Short term electric load forecasting, Ph.D. thesis, North Carolina State University, 2010 W.A. Ibrahim, M.M. Morcos, Artificial intelligence and advanced mathematical tools for power quality applications: a survey. IEEE Trans. Power Deliv. 17(2), 668–673 (2002) A.D. Jacob Sakhnini, in Smart Grid Cyber Attacks Detection Using Supervised Learning and Heuristic Feature Selection. 2019 7th International Conference on Smart Energy Grid Engineering (SEGE 2019) (2019). A. Jindal, A. Dua, K. Kaur, M. Singh, N. Kumar, S. Mishra, Decision tree and SVM-based data analytics for theft detection in smart grid. IEEE Trans. Ind. Inf. 12(3), 1005–1016 (2016) P. Jokar, N. Arianpoo, V.C. Leung, Electricity theft detection in AMI using consumption patterns. IEEE Trans. Smart Grid 1(7), 216–226 (2016) M.I. Jordan, C.M. Bishop, Neural networks, in Computer Science Handbook, Section VII: Intelligent Systems, ed. by A. B. Tucker, 2nd edn., (Chapman & Hall/CRC Press LLC., Boca Raton, 2004) L.P. Kaelbling, M.L. Littman, A.W. Moore, Reinforcement learning: a survey. J. Artif. Intell. Res. 4, 237–285 (1996) S. Kirubadevi, S. Sutha, (2017). Wavelet based transmission line fault identification and classification. In 2017 International Conference on Computation of Power, Energy Information and Commuincation (ICCPEIC) (pp. 737–741). IEEE M. Lenzerini, (2002). Data integration: A theoretical perspective. In Proceedings of the twenty-first ACM SIGMOD-SIGACT-SIGART symposium on Principles of database systems (pp. 233– 246) Y. Liang, D. He, D. Chen, in Poisoning Attack on Load Forecasting. 2019 IEEE Innovative Smart Grid Technologies-Asia (ISGT Asia) (IEEE, 2019), pp. 1230–1235 M. Liao, A. Chakrabortty, in A Round-Robin ADMM Algorithm for Identifying Data-Manipulators in Power System Estimation. In 2016 American Control Conference (ACC) (IEEE, July 2016), pp. 3539–3544 N. Lidula, N. Perera, A. Rajapakse, in Investigation of a fast islanding detection methodology using transient signals. Proc. IEEE Power Energy Soc. Gen. Meeting (2009), pp. 1–6. X. Liu, P.S. Nielsen, Regression-based online anomaly detection for smart grid data (2016). https:/ /arxiv.org/abs/1606.05781 Y. Liu, P. Ning, M.K. Reiter, False data injection attacks against state estimation in electric power grids. ACM Trans. Inf. Syst. Secur. 14(1), 1–33 (2011)

1572

M. T. Dehghan Nezhad and M. m. Sarbishegi

J.J. Lorenzo Navarro, J.Á. Méndez Rodríguez, M. Castrillón-Santana, J.D. Hernández Sosa, Shortterm wind power forecast based on cluster analysis and artificial neural networks. Lect. Notes Comput. Sci. 20, 520 (2011) B. Lundstrom, P. Gotseff, J. Giraldez, M. Coddington, in A High-Speed, Real-Time Visualization and State Estimation Platform for Monitoring and Control of Electric Distribution Systems: Implementation and Field Results. 2015 IEEE Power & Energy Society General Meeting (IEEE, 2015, July), pp. 1–5 Z. Luo, S. Hong, Y. Ding, A data mining-driven incentive-based demand response scheme for a virtual power plant. Appl. Energy 239, 549–559 (2019) P. Mahat, Z. Chen, B. Bak-Jensen, in Review of Islanding Detection Methods for Distributed Generation. 2008 Third International Conference on Electric Utility Deregulation and Restructuring and Power Technologies (IEEE, April 2008), pp. 2743–2748) M.A. Mahmoud, N.R. Md Nasir, M. Gurunathan, P. Raj, S.A. Mostafa, The current state of the art in research on predictive maintenance in smart grid distribution network: fault’s types, causes, and prediction methods – a systematic review. Energies 14(16), 5078 (2021) G.M. Messinis, A.E. Rigas, N.D. Hatziargyriou, A hybrid method for non-technical loss detection in smart distribution grids. IEEE Trans. Smart Grid 10(6), 6080–6091 (2019) T. Mitchell, Machine Learning (McGraw-Hill, New York, 1997) A. Mukherjee, R. Vallakati, V. Lachenaud, P. Ranganathan, in Using phasor data for visualization and data mining in smart-grid applications. 2015 IEEE First International Conference on DC Microgrids (ICDCM) (IEEE, June 2015), pp. 13–18 W. Najy, H. Zeineldin, A.K. Alaboudy, W.L. Woon, A Bayesian passive islanding detection method for inverter-based distributed generation using ESPRIT. IEEE Trans. Power Deliv. 26, 2687– 2696 (2011) A. Ng, Machine learning yearning (2017), http://www.mlyearning.org/(96), p. 139 T.T. Nguyen, A. Yousefi, in Multi-Objective Demand Response Allocation in Restructured Energy Market. ISGT 2011 (IEEE, January 2011), pp. 1–8 S. Park, S. Ryu, Y. Choi, H. Kim, in A Framework for Baseline Load Estimation in Demand Response: Data Mining Approach. 2014 IEEE International Conference on Smart Grid Communications (SmartGridComm) (IEEE, 2014, November), pp. 638–643 N. Phuangpornpitak, S. Tia, Opportunities and challenges of integrating renewable energy in smart grid system. Energy Procedia 34, 282–290 (2013) P. Qi, S. Jovanovic, J. Lezama, P. Schweitzer, Discrete wavelet transform optimal parameters estimation for arc fault detection in low-voltage residential power networks. Electr. Power Syst. Res. 143, 130–139 (2017) S. Quinde, J. Rengifo, F. Vaca-Urbano, in Non-technical Loss Detection Using Data Mining Algorithms. 2021 IEEE PES Innovative Smart Grid Technologies Conference-Latin America (ISGT Latin America) (IEEE, September 2021), pp. 1–5 I. Rish, in An Empirical Study of the Naive Bayes Classifier. IJCAI 2001 Workshop on Empirical Methods in Artificial Intelligence, vol. 3, no. 22 (August 2001), pp. 41–46. B. Rossi, S. Chren, B. Buhnova, T. Pitner, in Anomaly Detection in Smart Grid Data: An Experience Report. 2016 IEEE International Conference on Systems, Man, and Cybernetics (SMC) (IEEE, 2016, October), pp. 002313–002318) S. Russell, P. Norvig, Artificial Intelligence: A Modern Approach (Prentice Hall, Hoboken, 2002) S.R. Samantaray, K. El-Arroudi, G. Joós, I. Kamwa, A fuzzy rule-based approach for islanding detection in distributed generation. IEEE Trans. Power Deliv. 25, 1427–1433 (2010) J. Santos-Pereira, L. Gruenwald, J. Bernardino, Top data mining tools for the healthcare industry. J. King Saud Univ. Comput. Inf. Sci. 2021, 52 (2021) Y.M. Saputra, D.T. Hoang, D.N. Nguyen, E. Dutkiewicz, M.D. Mueck, S. Srikanteswara, in Energy Demand Prediction with Federated Learning for Electric Vehicle Networks. 2019 IEEE Global Communications Conference (GLOBECOM) (IEEE, December 2019), pp. 1–6 A. Sargam, Top 10 data mining tools (2021). Retrieved 28 Dec 2021, from https:// www.jigsawacademy.com/blogs/data-science/data-mining-tools

Data Mining Applications in Smart Grid System (SGS)

1573

R. Shyam, H.B. Bharathi Ganesh, S. Kumar, P. Poornachandran, K.P. Soman, Apache spark a big data analytics platform for smart grid. Procedia Technol. 21, 171–178 (2015) V.K. Singh, M. Govindarasu. in Decision Tree Based Anomaly Detection for Remedial Action Scheme in Smart Grid Using PMU Data. 2018 IEEE Power & Energy Society General Meeting (PESGM) (IEEE, August 2018), pp. 1–5 S.F. Stefenon, M.H.D.M. Ribeiro, A. Nied, V.C. Mariani, L.D.S. Coelho, D.F.M. da Rocha, R.B. Grebogi, A.E.D.B. Ruano, Wavelet group method of data handling for fault prediction in electrical power insulators. Int. J. Electr. Power Energy Syst. 123, 106269 (2020) T. Stobierski, Top 6 data visualization tools for business professionals (2021). Retrieved 28 Dec 2021, from https://online.hbs.edu/blog/post/data-visualization-tools H. Sun, Z. Wang, J. Wang, Z. Huang, N. Carrington, J. Liao, Data-driven power outage detection by social sensors. IEEE Trans. Smart Grid 7(5), 2516–2524 (2016) A. Taïk, S. Cherkaoui, in Electrical Load Forecasting Using Edge Computing and Federated Learning. ICC 2020–2020 IEEE International Conference on Communications (ICC) (IEEE, June 2020), pp. 1–6 B.A. Tama, K.H. Rhee, Data mining techniques in DoS/DDoS attack detection: a literature review. Int. J. Inf. 18(8), 3739 (2015) I. Ullah, F. Yang, R. Khan, L. Liu, H. Yang, B. Gao, K. Sun, Predictive maintenance of power substation equipment by infrared thermography using a machine-learning approach. Energies 10(12), 1987 (2017) O. Velarde, Top 10 data visualization tools for 2021 (2021). Retrieved 28 Dec 2021, from https:// visme.co/blog/data-visualization-tools/ H. Wang, L. Zhao, J.S. Liu, X. Ji, in Prediction of Electrical Equipment Failure Rate Based on Improved Drosophila Optimization Algorithm. 2017 13th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD) (IEEE, July 2017), pp. 1915–1921 Wikipedia Contributors, Visualization (graphics). Wikipedia (2021, December 14), https:// en.wikipedia.org/wiki/Visualization_(graphics) S. Wu, A review on coarse warranty data and analysis. Reliab. Eng. Syst. Saf. 114, 1–11 (2013) J. Yan, B. Tang, H. He, in Detection of False Data Attacks in Smart Grid with Supervised Learning. 2016 International Joint Conference on Neural Networks (IJCNN) (IEEE, 2016), pp. 1395–1402 M.E. Zarei, M. Gupta, D. Ramirez, F. Martinez-Rodrigo, Switch fault tolerant model-based predictive control (MPC) of a VSC connected to the grid. IEEE J. Emerg. Sel. Top. Power Electron. 5(2), 112 (2019) Y. Zhang, J. Ren, J. Liu, C. Xu, H. Guo, Y. Liu, A survey on emerging computing paradigms for big data. Chin. J. Electron. 26(1), 1–12 (2017) Zhao, J.; Xia, X.; Su, D.; Xu, C.; Wu, F., in Fault Section Location Method Based on Random Forest Algorithm for Distribution Networks with Distribution Generations. Proceedings of the 2019 IEEE Innovative Smart Grid Technologies-Asia (ISGT Asia), Chengdu (21–24 May 2019), pp. 4165–4169 W. Zhe, C. Wei, L. Chunlin, in DoS Attack Detection Model of Smart Grid Based on Machine Learning Method. 2020 IEEE International Conference on Power, Intelligent Computing and Systems (ICPICS) (IEEE, July 2020), pp. 735–738

Part III Intelligent Development of Energy Systems

The Role of Blockchain and Cryptocurrency in Smart Grid: Renewable Energy Trading, System Security and Privacy Preservation Wenlin Han

Contents 1 2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Smart Grid Key Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Renewable Energy Trading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 How to Make the Economics Work? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 How to Deal with Self-Interests and Concerns when Making a Political Decision? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Where to Build and how Big to Build? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 How to Manage Decentralized Generation, Distribution, and Redistribution? . . . 2.6 How to Supply on Demand? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7 How to Cope with the Existing Infrastructure? . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.8 How to Build a Healthy Ecosystem in the Energy Sector? . . . . . . . . . . . . . . . . . . 3 System Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Privacy Preservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Cryptocurrencies and Blockchain Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Blockchain-Based Renewable Energy Trading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Energy Web Token (EWT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Blockchain-Based Smart Grid System Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Blockchain-Based Smart Grid Privacy Preservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1578 1579 1579 1579 1579 1579 1580 1580 1580 1581 1581 1581 1581 1583 1584 1584 1585 1587 1587

Abstract

The emergence of Blockchain technologies may open the world with lots of new possibilities. Will Blockchain innovate the energy sector as well? Renewable energy trading, system security, and privacy preservation are the three key

W. Han () Department of Computer Science, California State University, Fullerton, Fullerton, CA, USA e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_8

1577

1578

W. Han

challenges in Smart Grid. This book chapter will introduce these key challenges and the solutions built on blockchains and cryptocurrencies to address these challenges. After reading this chapter, readers will get to how Blockchain technologies will innovate the energy sector. Readers will also get to learn the latest Blockchain and cryptocurrency projects that aim to address the three key challenges in energy sector. Keywords

Blockchain · Cryptocurrency · Security · Privacy · Smart Grid · Renewable energy

1

Introduction

Renewable energy and renewable energy trading are one of the most appealing features of Smart Grid. With the energy generated from renewable resources, such as sunlight, wind, and tide, and the ability to trade these energies to where they are needed, people are expecting a big and promising vision of Smart Grid. However, renewable energy trading is also one of the most difficulty problems in Smart Grid due to the following challenges. The number one challenge for renewable energy is to provide power on demand since it is not always on. The next challenge is to estimate the scale of the demand and decide how big to build. Also, renewable energy traders have the potential to disconnect from the grid causing unreliability issues. Regarding trading itself, challenges include pricing, secure payment, bidding, transaction disputing, platform, and many more. System security and privacy concern are the other two big obstacles in the development of Smart Grid (Han and Yang n.d.). It is a double-edged sword when a system becomes “smart.” On the one side, the system becomes more efficient. On the other side, it is more vulnerable to attacks. The two-way communication and electricity flows have brought tremendous number of benefits and attractions to Smart Grid. But privacy concern is one of the major challenges that has been raised. The emergence of Blockchain technologies may open the world with lots of new possibilities. A blockchain is a sequential list of transaction blocks which are distributed across many computers in the networks. This list is public so that any attempt to alter the transaction records can be found out. The list is maintained in a decentralized manner so that the third party is no longer needed in any transaction. As the main application of Blockchain, cryptocurrencies have redefined money, payment systems, and monetary policies. This book chapter will introduce the major challenges in Smart Grid with a focus on privacy preservation, renewable energy trading and system security. Towards addressing these challenges, what are the roles of Blockchain technologies? There are lots of non-Blockchain-based solutions to these challenges. What do Blockchain technologies buy us in terms of privacy preservation, renewable energy trading and system security?

The Role of Blockchain and Cryptocurrency in Smart Grid: Renewable Energy. . .

2

Smart Grid Key Challenges

2.1

Renewable Energy Trading

1579

Renewable Energy refer to energy generated by renewable natural resources, such as sunlight, wind, geothermal heat, tides, etc. In contrast to fossil energy, renewable energy has lots of advantages including reduce carbon emission, mitigate energy crisis, prevent global warming, improve public health, and so much more. Power industry is a monopolized business in most parts of the world. There are a few big power companies which control the power generation, distribution and redistribution process in a certain country or area. When it comes to renewable energy, this centralized business model has been facing huge challenges.

2.2

How to Make the Economics Work?

The transition from carbon to renewable energy is a major shift. The investment on new equipment, devices, maintenance, human resources, and other resources is a huge financial cost. Renewable energy sector often seeks financial support from governments, funding agencies and large organizations. However, they are reluctant to put a drop of water into the ocean.

2.3

How to Deal with Self-Interests and Concerns when Making a Political Decision?

The perception and awareness of renewable energy is positive for most people. But when it comes to political decisions in a centralized and monopolized business, politicians also struggle with the polar opposite requirements and pressures of the voting public to make the changes they need.

2.4

Where to Build and how Big to Build?

To build a renewable energy plant, such as a solar farm or a wind turbine farm, land use is one of the major challenges. Other land- use requirements, such as housing, food production, agriculture, etc., have been competing with renewable energy sector for optimal lands. Deciding how big to build can also be challenging. Energy could be wasted if it is too big. We also do not want to undersize since it is hard to start over. The traditional power grids only support one-way electricity flow and one-way communication flow. Smart Grid can support two-way way electricity flow and oneway communication flow. The two-way communication enables grid to user and

1580

W. Han

user to grid communication which is similar to mobile phones or computer networks that people are very familiar with. The two-way electricity flow not only allows the grids to supply end users but also allow the end users to sell electricity back to the grids or to other end users. Renewable energy trading is one of the most exciting features supported by the two-way electricity flow of Smart Grid. People can not only produce electricity to supply their own but also can sell the excess electricity to the grids or to other end users. If people start to install solar panels on their rooftops and/or install windmills in their backyards, it is the first step to decentralize energy. It could not only make households self-sustainable but also monetize energy and bring profits. If lots of people participate in renewable energy trading, the energy costs will go down. The energy sector will no longer feel headache about where and how big build. The governments could be relieved from the pressure of economics and politics. However, lots of challenges have been raised towards energy decentralization in renewable energy trading.

2.5

How to Manage Decentralized Generation, Distribution, and Redistribution?

To enable renewable energy trading, we will need a system to manage energy generation, distribution, and redistribution in a decentralized manner. But the existing energy system is a centralized system which is controlled by several big corporations. As we all know, centralized systems are easier to manage when compared to decentralized systems. Even the existing centralized power systems are becoming increasingly unreliable and fragile. Therefore, we can foresee the challenges involved in decentralized systems.

2.6

How to Supply on Demand?

To supply on demand is a problem that has been studied for a long time. In a centralized power system, a power company is in charge of a certain area. The company builds models to study this problem, predict the demand, and get ready for backup plans. When it comes to a decentralized system, how to supply on demand when we lack of central control and analysis?

2.7

How to Cope with the Existing Infrastructure?

Suppose we have a new system which is able to address the above challenges. What do we do with the existing power infrastructure? Will the stakeholders of the old infrastructure object the new system because of conflict of interests? Can we reserve the existing power systems as much as possible?

The Role of Blockchain and Cryptocurrency in Smart Grid: Renewable Energy. . .

2.8

1581

How to Build a Healthy Ecosystem in the Energy Sector?

The energy sector is complicated which is composed of stake holders playing all different roles, such as financial Institutions, state legislatures, local governments, public utility commissions, energy associations and organization, utilities, contractor, installer, vendor, investor, etc. How can we build a healthy ecosystem to better support renewable energy trading for the long run?

3

System Security

System security is one of the major concerns in Smart Grid. It is a double-edged sword when a system becomes “smart.” On the one side, the system becomes more efficient. On the other side, it is more vulnerable to attacks. The attacks on Smart Gird can be physical or cyber. In the existing power systems, Intrusion Detection Systems (IDS) (Jow et al. 2017) are usually used as security guards to defend the power systems from being hacked. SCADA (Gao et al. 2014) is one of the most popular and widely used IDS in Smart Grid. However, these security systems could still suffer from single point failure and fail to detect new attack types (Han and Yang 2019) (Fig. 1).

4

Privacy Preservation

The two-way communication and electricity flows have brought tremendous number of benefits and attractions to Smart Grid. But privacy concern is one of the major challenges that has been raised. Personal data could be leaked in a variety of processes or components of the power grids, such as payment and billing, data management, communication protocols, battery management for electric cars, charging and discharging for electric cars, etc. (Han and Xiao 2016). Take payment and billing for electric cars as an example. Various existing payment systems are compared and shown in Table 1. However, none of these methods can address all major challenges.

5

Cryptocurrencies and Blockchain Technologies

Bitcoin, the very first cryptocurrency, was first introduced in a white paper entitled “Bitcoin: A Peer-to-Peer Electronic Cash System” in late 2008. The author, Satoshi Nakamoto, is a pseudo name (Nakamoto 2021). The second cryptocurrency is Namecoin which was first launched in April 2011 (Namecoin 2021). In the same year, Litecoin was released. Till now, there are over 5,000 cryptocurrencies on the market.

1582

W. Han

Placement of IDS Topology Training of IDS

Distributed

Update Profiles

Centralized Hierarchical

Update Signatures

Attack Classification Robust Algorithm Reduce FP

Real time and Non real time monitoring

Resource Constraints Low Memory

Reduce FN Improve TPR

Analysis of Stored SG communication data

Limited Computation Power

Real time Packet analysis and filtering

Effective and Efficient IDS for Smart Grids

Reliability Redundancy Self-healing

Constant Removal and Introduction of Devices

Detection Latency Detection Speed

Customer off The Shelf

Rcovery Availability

Network Energy Management Services

Database

Fig. 1 Effective and efficient Intrusion Detection System (IDS) for Smart Grid (Jow et al. 2017) Table 1 Comparison of the Existing Payment Systems (Han and Xiao 2016) Scheme Paper cash Prepaid Cash card Transferrable e-cash Credit card PayPal

Locationprivacy Yes Yes

Prevention of cheating Yes Yes

Lost protection No No

2-way transaction Yes Partial

Stolencar trace No No

Yes

No

No

Yes

No

No Partial

Yes Yes

Yes Yes

Yes Yes

No No

The technologies that used by Bitcoin and other cryptocurrencies have been studied and extended to apply in other fields. People call these technologies as Blockchain technologies and treat cryptocurrency as one of the applications of

The Role of Blockchain and Cryptocurrency in Smart Grid: Renewable Energy. . .

1583

Fig. 2 A simple example of a blockchain

Blockchain technologies. Sometimes, we call Bitcoin, Namecoin, Litecoin, etc. as the first generation Blockchain technologies. Ethereum emerged as the leader of the second generation blockchain technologies in 2014. Ethereum provides its own programming language to write computer programs that can bundle events and actions relevant to an agreement together and automatically execute which we call Smart Contract. Generally speaking, a blockchain is a sequential list of transaction blocks which are distributed across many computers in the networks. This list is public so that any attempt to alter the transaction records can be found out. The list is maintained in a decentralized manner so that the third party is no longer needed in any transaction (Fig. 2).

6

Blockchain-Based Renewable Energy Trading

Blockchain technologies can provide good solutions to address challenges in renewable energy trading that the traditional technologies cannot address. In an example architecture for Blockchain-based renewable energy trading (Alladi et al. 2019), Energy Aggregators (EAGs) form a blockchain network which is responsible for collecting transactions, proposing blocks and maintaining the ledger. Energy nodes, such as household consumers, industrial end users, and electric cars from anywhere can join the network anytime. These energy nodes do not maintain the blockchain network, but they can initiate transactions as sellers or buyers. The first Blockchain project in the energy sector is SolarCoin which was launched in 2014. Since then, quite a few projects released their energy coins. As shown in Table 2 is a list of top cryptocurrency energy coins ranked by their market capitalization.

1584

W. Han

Table 2 Top 10 Cryptocurrency Energy Coins by Marketcap (Cryptoslate Energy Coins Ranking 2021) Name Energy Web Token (EWT) Power Ledger (POWR) Efforce (WOZX) WePower (WPR) SunContract (SNC) Grid+ (GRID) WPP Token (WPP) Renewable Electronic Energy Coin (REEC) Lition (LIT) Restart Energy MWAT (MWAT) SolarCoin (SLR)

7

Marketcap $229.6 M $41.65 M $44.14 M $8.78 M $8.26 M $7.87 M $2.09 M $1.92 M $1.78 M $1.52 M $1.35 M

Blockchain Own blockchain Ethereum Ethereum Ethereum Ethereum Ethereum Ethereum Ethereum Ethereum Ethereum Own blockchain

Year launched 2017 2016 2018 2016 2017 2017 2018 2017 2019 2017 2014

Energy Web Token (EWT)

Energy web token (EWT) was launched in 2017. Currently, the token is ranked as #87 of all cryptocurrencies with a market cap of $229.6 M and is ranked as #1 in energy coins (Energy Web Token 2021; Energy Web Token at CryptoSlate 2021). The blockchain that supports EWT is Energy Web (EW) Chain, which is “a global blockchain infrastructure that we believe to be the only public, Proof-ofAuthority blockchain supporting commercial applications with well-known organizations as validator nodes.” (Energy Web Token 2021). Unlike most of the other Blockchain-based smart energy solutions which are built on Ethereum, Energy Web uses its own Proof-of-Authority blockchain and provides a decentralized operating system, EW-DOS. EW-DOS provides a stack of opensource software and standards to enable stake holders to participate in smart energy markets digitally.

8

Blockchain-Based Smart Grid System Security

Providing security to a distributed and decentralized system where multiple entities don’t trust each other but still need to reach a consensus, Blockchain is the right technology. Blockchain technologies can provide a tamper-resistant ledger for securing IDs and policy information. To get validated by the Blockchain system, any user needs to follow its built-in rules, thus enforces security. Any user can join and leave the network anytime which allows the network to be scaled to millions of users easily. Xage Fabric (2021) is a Blockchain-based security system aims to address critical security challenges in Smart Grid. There are two types of nodes in Xage Fabric:

The Role of Blockchain and Cryptocurrency in Smart Grid: Renewable Energy. . .

1585

Xage Brokers and Xage Nodes. The blockchain ledger stores the topology, policies, IDs, authentication information, and activity logs. The traditional IDS-based security systems suffer from single point failure. The blockchain ledger enables Fabric to tolerant more faults. Since security information are spread across the network, tampering with one or several of the nodes will not comprise the entire system. Instead, such malicious behaviors will be recorded by the ledger and easily found out. The nodes which performed malicious actions will be isolated and healed by Fabric, which the Xage team calls self-healing. Different from cryptocurrency-based projects where the blockchain is an appendonly ledger, the Xage Security Fabric employs hierarchical design with branches. The reason why cryptocurrencies usually do not allow branches is to prevent double spending attacks. But since Xage Fabric is not a currency, double spending attack is not a problem here. The benefits of using hierarchical design include reducing transaction time and improve scalability. Each branch stores a subset of data and maintains its local consensus.

9

Blockchain-Based Smart Grid Privacy Preservation

Satoshi Nakamoto, the mysterious creator of Bitcoin, is thought to have mined 1.1 million Bitcoins which are now worth in excess of $30 billion. But nobody knows who she or he is. The reasons are twofold. First, Satoshi Nakamoto, the name that appeared as the author of Bitcoin whitepaper is not a real name. Second, Bitcoin does not use real IDs in the network. Instead, Bitcoin uses public keys as addresses for holding coins and these public keys do not link to people’s real IDs. Therefore, using cryptocurrencies as payment is a method to protect privacy. In the previous sections, I’ve talked about several cryptocurrencies and Blockchain projects, such as Energy web token and Xage Fabric, and they have their built-in schemes for privacy preservation. But the energy sector needs to migrate to or adopt these schemes which might take a long time. A simple solution to fill the gap is to use privacy-preserving cryptocurrencies in the existing grid systems. As shown in Table 3, top anonymous cryptocurrencies and their privacy-preserving techniques are listed. Dash (DASH) was launched as “the first privacy-centric cryptographic currency.” in January 2014. The anonymization method that Dash used is DarkSend, later renamed as PrivateSend, which is built on coinjoin (CoinJoin 2021), a privacypreserving technique first used in Bitcoin. In coinjoin, multiple payments from several users are combined and merged into one single transaction which makes it difficult to distinguish who paid whom. Monero (XNP) was launched in the same year as Dash. But different from Dash, Monero proposed and employed a privacy-preserving technique that used in cryptocurrencies for the first time, RingCT protocol (Monera RingCT 2021). RingCT protocol belongs to the family of ring signature. Although RingCT protocol

1586

W. Han

Table 3 Top anonymous cryptocurrencies and privacy-preserving techniques Cryptocurrency Monero (XNP)

Year released 2014

Forked from N/A

PIVX (PIVX) Komodo (KMD)

2016 2017

Dash ZCash

Dash (DASH) Zcash (ZEC)

2014 2016

Bitcoin Bitcoin

Verge (XVG) Zcoin (XZC) Navcoin (NAV) Zencash (ZEN) ZClassic (ZEC) Deeponion (ONION) Spectrecoin (XSPEC) Bitcoin Private (BTCP) Quisquis

2014 2016 2014 2017 2016 2017

N/A N/A N/A ZClassic ZCash N/A

2016

Shadow-Cash

2018

Bitcoin and ZClassic N/A

Under development as of March 2020

Privacy-preserving technique Ring confidential transaction (RingCT) Zero knowledge proof, zerocoin Zero knowledge proof, Delayed Proof of Work (dPoW), Atomic swap Coinjoin (PrivateSend) Zero knowledge succinct arguments of knowledge (zkSNARKs) Tor and I2P Zerocoin NavTech Zentalk, Zenpub, Zenhide zkSNARKs X13 Proof of Work, OBFS4 OBFS4 Bridges, improved stealth addresses zkSNARKs Zero-knowledge proof, Updatable Public Keys

was first used in cryptocurrencies by Monero, ring signature has a long history in cryptography, and it is famous for its strong anonymity. Zcash (ZEC) was launched in 2016 and it employs a privacy-preserving technique, named zero knowledge succinct arguments of knowledge (zkSNARKs) (What are zk-SNARKs? 2021). The addresses in Zcash are classified into different types such as shielded addresses (z-addr) and transparent addresses (taddr). The input and output values in each transaction are split and spread into shielded addresses and transparent addresses. Therefore, it is difficult for outside entities to extract private information from the input and output values. Quisquis (Fauzi et al. 2018 ) is a newly proposed anonymous cryptocurrency that has not officially launched yet. Quisquis claims to achieve provably secure notions of anonymity and resolves the limitations outlined above for existing solutions. Quisquis is built on a privacy-preserving technique, zero-knowledge proof, which only needs to store a relatively small amount of data and does not require trusted setup. Different from most of the existing cryptocurrencies that addresses are used repeatedly, each Quisquis address appears on the blockchain at most twice. This feature limits the adversaries’ ability to find patterns through the transactions where the same addresses occur repeatedly.

The Role of Blockchain and Cryptocurrency in Smart Grid: Renewable Energy. . .

1587

Table 4 The security properties and efficiency considerations for Dash, Zcash, Monero, and Quisquis (Fauzi et al. 2018) Anonymity Yes Yes No Yes

Dash Zcash Monero Quisquis

Deniability No No Yes Yes

Theft prevention Yes Yes Yes Yes

UTXO growth Non-consistent Consistent Consistent Non-consistent

As shown in Table 4, the security properties are compared over for Dash, Zcash, Monero, and Quisquis.

10

Conclusion

In this chapter, the three most challenging problems, renewable energy trading, system security, and privacy preservation in Smart Grid were introduced. The nonBlockchain-based solutions can mitigate some issues, but lack of systematic ways to address them. The state-of-the-art Blockchain-based solution was introduced in this chapter. These solutions have the potential to solve renewable energy trading, system security, and privacy preservation problems.

References T. Alladi et al., Blockchain in smart grids: a review on different use cases. Sensors 19(22), 48–62 (2019) CoinJoin. https://coinioin.io/. Accessed Mar 2021 Cryptoslate Energy Coins Ranking. Available at: https://crvptoslate.com/crvptos/energy/. Accessed Jan 2021 Energy Web Token at CoinMarketCap. Available at: https://coinmarketcap.com/currencies/ energy-web-token/ . Accessed Jan 2021 Energy Web Token at CryptoSlate. Available at: https://cryptoslate.com/crvptos/energv/. Accessed Jan 2021 Energy Web Token official website. Available at: https://energvweb.org/technologv/token/. Accessed Jan 2021 P. Fauzi et al., QuisQuis: a new design for anonymous cryptocurrencies. IACR Cryptol EPrint Arch 2018, 990 (2018) J. Gao, J. Liu, B. Rajan, R. Nori, B. Fu, Y. Xiao, W. Liang, C.L.P. Chen, Scada communication and security issues. Secur Commun Netw Secur Comm 7(1), 175–194 (2014) W. Han, Y. Xiao, Privacy preservation for V2G networks in smart grid: A survey. Comput. Commun., 91, 17–92, 28 (2016) W. Han, X. Yang, Edge computing enabled NonTechnical loss fraud detection for big data security analytic in smart grid. J. Ambient. Intell. Humaniz. Comput., 1–12 (2019) W. Han, X. Yang, IP2 DM: Integrated PrivacyPreserving Data Management Architecture for Smart Grid V2G Networks. Wirel Commun Mob Comput J, Wiley 16(17), 2775–3186 (n.d.)

1588

W. Han

J. Jow, Y. Xiao, W. Han, A survey of intrusion detection in smart grid. Int J Sens Netw 23(3), 170–186 (2017) Monera RingCT. https://www.getmonero.org/resources/moneropedia/ringCT.html. Accessed Mar 2021 Nakamoto S. Bitcoin: a peer-to-peer electronic cash system. Available at: https://bitcoin. org/bitcoin.pdf . Accessed Mar 2021 Namecoin. https://www.namecoin.org/. Accessed Mar 2021 What are zk-SNARKs?. https://z.cash/technology/zksnarks/. Accessed Mar 2021 Xage Fabric. https://xage.com/. Accessed Jan 2021

Energy Harvesting for Smart Energy Systems Shirin Momen, Javad Nikoukar, Arsalan Hekmati, Soheil Majidi, Zahra Zand, Mohammad Zand, and Mostafa Eidiani

Contents 1 2 3 4 5 6 7 8 9 10

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Smart Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Main Characteristics of Smart Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conscious and Active Participation of Consumers in Smart Grids . . . . . . . . . . . . . . . . Modification of Production and Amount of Reserves . . . . . . . . . . . . . . . . . . . . . . . . . . . Provide the Required Power Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flexibility in the Face of Disasters and Natural Disasters . . . . . . . . . . . . . . . . . . . . . . . New Products, New Services, and New Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Intelligent Networks for Optimizing Equipment Use and Increasing Performance Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Rearrangement Goals in Distribution Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1591 1591 1593 1594 1594 1594 1595 1595 1595 1595 1596

S. Momen · J. Nikoukar Department of Electrical Engineering, Islamic Azad University, Saveh, Iran e-mail: [email protected]; [email protected] A. Hekmati Revterra Co., Houston, TX, USA e-mail: [email protected] S. Majidi Research and Development Department, BLUE&P Group, Tehran, Iran e-mail: [email protected] Z. Zand Razi University, Kermanshah, Iran M. Zand () CTIF Global Capsule, Department of Business Development and Technology, Denmark and Renewable Energy Lab (REL), Aarhus University, Herning, Denmark M. Eidiani Khorasan Institute of Higher Education, Mashhad, Iran e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_12

1589

1590

S. Momen et al.

12 13 14 15 16 17 18 19

Network Rearrangement in the Presence of Loss Reduction . . . . . . . . . . . . . . . . . . . . . Rearrangement in the Network in Order to Balance the Load of Feeders . . . . . . . . . . . Rearrangement in Order to Recover the Load and Increase Reliability . . . . . . . . . . . . . Rearrangement in Order to Melt the Ice of the Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods of Reducing Network Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Improved Frog Mutation Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analysis of Results and Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Following Scenarios Are Considered in Solving this Problem . . . . . . . . . . . . . . . . 19.1 Scenario 1: Basic Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.2 Scenario 2: Rearrangement Alone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.3 Scenario 3: Placement of Scattered Products in the Network . . . . . . . . . . . . . . . 19.4 Scenario 4: Locating a Capacitive Bank in the Network . . . . . . . . . . . . . . . . . . . 19.5 Scenario 5: Rearrangement with Simultaneous Placement of Distributed Generation Sources and Optimal Capacitance in the Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Offers: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1596 1596 1596 1597 1597 1598 1600 1601 1601 1602 1604 1605

1607 1607 1610 1610

Abstract

With increasing the amount of electricity demand by subscribers, increasing attention to the quality of energy delivered to consumers and welcoming planners and independent users of the power system from the presence of distributed generation resources as a part of the network needs, formulating a technical and economic framework How much better are these resources in the power grid. One of the current research areas on distributed generation resources is to study how these resources are optimally allocated in the power grid. Optimal allocation of distributed generation resources in the network means locating and determining the appropriate size of production of these resources in order to achieve specific goals. These goals include maximizing investor profits, minimizing supply costs, reducing losses, improving voltage profiles, improving reliability indicators, and reducing emissions from electricity generation. The existence of distributed generation (DG) is no longer valid. One of the important issues and challenges in the field of distribution networks is the discussion of power losses and in fact its optimization, which needs to be analyzed for safe operation. In this chapter of the book Optimal Allocation of Distributed Generation Units in a Nonlinear Optimization Problem with Constraints to Minimize Loss Using the Improved Shuffled Frog Leaping Algorithm (ISFLA) Analysis is located. This study uses the ISFLA optimization algorithm to minimize losses in a standard 33-bus radial distribution system. Voltage followed that this is very important in distribution systems. Keywords

Particle Swarm Optimization (PSO) · Scattered Production Resources · Distribution Production Unit · Restructuring · Frog Jump Algorithm · Sequential Quadratic Programming · Versatile Energy Resource Allocation

Energy Harvesting for Smart Energy Systems

1

1591

Introduction

Today, increasing population growth and the process of industrialization and the growing need for energy on the one hand and the risk of depletion of fossil fuel resources on the other hand as well as human need to achieve a sustainable source to meet their current and future needs, cause more attention to renewable resources. Like solar panels and wind turbines, it has been cleaned up as energy. Free energy from this type of resource is one of the reasons for their acceptance. The advantages of renewable energy include having a long life, reducing energy purchase costs associated with rising fuel prices and improving system reliability, reducing losses, delaying investment, improving voltage profiles, and more. Renewable energy sources in power systems are widely used to generate electricity. Due to the mentioned advantages, these renewable sources have problems in energy production due to their random nature in the distribution system, so that it is difficult to predict the output power, and this causes a sharp fluctuation in the output power, which causes many problems. For this reason, the use of energy storage systems in different parts of the power system is necessary to balance production and consumption. New technological advances in the power systems, especially the connection of renewable energy sources and the move toward free trade in electricity markets, have opened up new opportunities and horizons for electric energy storage methods. On the other hand, the importance and various applications of electrical energy storage technologies have doubled in the last decade due to advances in related sciences. Storage devices are one of the emerging equipment in distribution networks whose optimal location depends on the consumption curve, the presence or absence of traditional and renewable distributed generation sources, as well as design goals. Intrusion of energy storage systems as a solution to solve the system stability problems. Among energy storage technologies, batteries have a better ability to disperse renewable energy, especially photovoltaic system and wind energy, due to their high energy density.

2

Literature Review

There is a great deal of research on the location of energy sources today, not only for distributed energy sources and renewable energy sources, but also heating. In spite of this, there are two main approaches: placement requirements and power line layout (storage and power requirements). The goal of both is to minimize power losses and maintain power quality. A conventional approach to microgrid configuration design in terms of cost optimization has been proposed to meet reliability constraints (Ghardashi et al. 2022; Zeynal et al. 2010). A dynamic programming approach underlies this approach and involves determining the optimal layout of power lines between keyword sources: Radial Distribution System - Scattered Production Resources - Distribution Production Unit - Restructuring, Frog Jump Algorithm.

1592

S. Momen et al.

Small and load points may provide their locations and line rights for connections (Kasaei et al. 2017; Nikoukar 2018; Heydari et al. 2020; Ahmadi-Nezamabad et al. 2019). In fact, an algorithm for microgrid programming has been proposed as an alternative to optimizing the extensible electrical power systems associated with transmission and generation (Zand et al. 2020). An annual reliability problem and a planning problem may be created from the optimization problem. A software called Versatile Energy Resource Allocation (VERA) is used to minimize the costs of total system planning. Based on local weather conditions, demand coverage is also forecasted. A Sequential Quadratic Programming (SQP) approach is used to solve the problem’s nonlinear aspects (Ghasemi et al. 2020). As discussed in (Nasri et al. n.d.), they focused on selecting the best configuration for an urban heating zone network. By using refrigeration simulation, an economic objective function is optimized for the network users. An urban microgrid operating in a competent manner can use a similar approach, even though this is not a microgrid programming issue. The improved discrete PSO method has been used to plan for increasing the efficiency of the distribution system over a period of 20 years. The goal is to minimize the total lifetime cost of the system, which includes: investment, reliability costs and line losses needs in transformers, lines, capacitors and distributed generation (Rohani et al. 2020). The scattered source output power, feeder current and bus voltage are affected in the optimization process as sequential constraints. The method enables the transfer of an existing radial distribution network to a microgrid that operates autonomously. The method has been used to develop positioning strategies and sizing for structural modification and scatter generators for autonomous microgrids (Hayati and Karimi 2020). The corresponding sizes of renewable resources and optimal locations for the autonomous performance of the system have been obtained using genetic algorithms and particle swarm optimization (Tightiz et al. 2020). Consideration of equal constraints, balances and production loads is used to formulate an optimization problem for costs and system losses. Several multi-objective optimization algorithms are analyzed for microgrid placement problems. The two-step multi-objective optimization process for microgrid programming in two primary distribution systems is implemented using MATLAB software. In the first step, the microgrid area in an initial distribution system is identified using the loss sensitivity factor. To determine the number and locations of scattered generators in a microgrid, a Pareto-based NSGA-II is presented in the second step. In addition to actual power losses, annual investment costs, and load voltage deviation, multi-objective functions are available. Using fuzzy decision making, optimal solutions have been found (Sanjeevikumar et al. 2021; Nasab et al. 2021) to solve the sizing and placement issues in distribution networks. Among the cost of power losses, the costs of upgrading the grid, the cost of power quality, the cost of unsupplied energy, and the cost of energy required, the best alternative should be found (Zeynal et al. 2014). Using a genetic algorithm, which is applied by the ε-constrained method to obtain a non-quadratic agreement solution (Morales et al. 2009). As explained, innovative methods are also used to solve localization problems. Bus placement

Energy Harvesting for Smart Energy Systems

1593

was also performed using loss sensitivity analysis (Ma et al. 2015). In order to maximize profit-to-cost ratio, MATLAB software has been used to implement the PSO algorithm. It is possible to minimize the cost of generating electricity not only by using distributed energy sources based on simultaneous generation of electricity and heat, but also by using them in the microgrid system based on bus location, capacity size, and type (Goel et al. 2006). Genetic algorithm is used to discover the optimal location of distributed generation in a MV distribution network based on real-size scenarios with several hundred nodes (Parvania and Fotuhi-Firuzabad 2010). A three-step process based on a genetic algorithm is presented, which is used to obtain the best solution for locating and determining the size of distributed generation in a medium voltage distribution network (Ioakimidis et al. 2014). It also examines a method for locating a distributed energy source within an optimal microgrid design framework that minimizes connection costs, sizing, and locating a distributed generation source with global and local reliability criteria (Nguyen et al. 2012). The problem of placement takes into account factors such as development costs and maintenance revenue using the simultaneous generation of electricity and heat, and formulates it as a refrigeration simulation optimization problem. An economic model and optimal location for industrial photovoltaic microgrids are presented in (Moreno et al. 2012; Bitar et al. 2012). The economic indicators analyzed include energy costs, reduction of environmental pollution and payback period. PSO is used to solve the optimization problem (Baringo and Conejo 2013). The need to integrate microgrid load distribution and network reconfiguration is considered, which leads to a non-convex nonlinear problem (Zugno et al. 2013). Four methods of evolutionary computational optimization such as particle swarm optimization (PSO), artificial neural network, genetic algorithm, and artificial vaccine neural network are compared in this paper (de la Nieta et al. 2014).

3

Smart Grids

The smart distribution network quickly identifies and disables devices that are likely to cause errors in the distribution network, as well as quickly detects leakage currents and quickly announces locations that require the presence of force to repair the slow network. The use of advanced measurement software quickly identifies subscribers who are out of service. Providing such information to incident personnel in a place of silence is very costly and greatly increases performance. Smart grids are attributed to the evolution and upgrading of existing networks and include automation, advanced monitoring, control of power generation, distribution and transmission. A smart grid can be defined from three points of view: A Smart Grid consumer can thus manage their consumption intelligently to pay less during peak energy prices.

1594

S. Momen et al.

Using this network helps address harmful climate change and reduce excessive carbon emissions, according to environmentalists. For couriers in the electricity industry, it is courier and smart decision making and providing accurate information about the state of the network. The world’s first smart grid was introduced in March 2008, and the city of Balder, Colorado, became the first city with a smart grid to design. Intelligent technology has the ability to make fundamental changes in the distribution, transmission, generation, and use of electricity along with environmental and economic benefits that ultimately lead to meeting the needs of customers and the availability of sustainable and reliable electricity. In addition, the system can prevent unwanted blackouts by making decisions based on the information collected. Increasing use of control technologies and digital information will increase the efficiency, security, and, reliability of the power grid, as well as the integration of distributed generation, demand response, and energy efficiency.

4

The Main Characteristics of Smart Grids

The main characteristics of smart grids are actually expressing the characteristics of these networks based on their capabilities. Smart grids were defined in order to eliminate the mentioned disadvantages of the existing grids and have the following specifications.

5

Conscious and Active Participation of Consumers in Smart Grids

Active consumer participation in electricity markets has tangible benefits for the grid and electricity companies. Smart grids provide the consumer with the necessary information about the consumption pattern and cost of electricity consumption, and this allows subscribers to operate in new electricity markets. Proper and accurate information to consumers allows them to change the amount of consumption based on the balance between the requested power and local production sources and the existing electricity network. The ability to reduce or change peak load times makes it possible for power generators to reduce investment and operating costs while combining environmental benefits by minimizing the operating time of efficient power plants and reducing line losses. It is less productive.

6

Modification of Production and Amount of Reserves

An intelligent power grid has the ability to utilize large and centralized power plants as well as distributed energy generation sources at the consumer site. Although large power plants, including advanced nuclear power plants, still play a key role in the smart grid, a large number of small distributed generation sources such as fuel cells,

Energy Harvesting for Smart Energy Systems

1595

advanced batteries, photovoltaic cells, plug-in hybrid vehicles, and wind can be used in this network (Heydarian-Forushani et al. 2014). Scattered generation sources can be easily connected to the power grid, and the ability to easily use different types of sources together is achieved as Plug and Play.

7

Provide the Required Power Quality

Monitoring, detecting and reacting to low power quality will lead to a significant reduction in subscriber losses compared to the present. Advanced control methods for monitoring primary sources provide rapid detection and solutions to factors that have reduced power quality, such as lightning, severe fluctuations, line errors, and ISFLA sources. Using smart grids, different levels of power quality can be achieved at different prices.

8

Flexibility in the Face of Disasters and Natural Disasters

Smart grids are able to deal with unexpected events and can disconnect the problematic part from the grid so that the rest of the grid can return to normal operation. This automatic detection and performance reduces customer downtime and provides better management of existing power delivery infrastructure by power companies.

9

New Products, New Services, and New Markets

Smart grids enable communication between buyers and sellers, from consumers to regional power companies (RTOs). This capability creates new markets that range from the level of energy management at the consumer location to the offer of energy sales at that level. With the increase of transmission routes and installation of energy production sources near the consumer, the share of subscribers’ participation in the market increases.

10

Intelligent Networks for Optimizing Equipment Use and Increasing Performance Efficiency

Smart grids use the latest technologies to be more efficient than equipment. The effectiveness of maintenance is optimized by considering the conditions of maintenance in such a way that it expresses the exact time required for maintenance of equipment. Line congestion and losses can be reduced by adjusting system control devices. In this type of control devices, it is possible to deliver energy to the final consumers at the lowest cost, which can ultimately increase the efficiency of operation.

1596

11

S. Momen et al.

Rearrangement Goals in Distribution Networks

Rearrangement in distribution networks is a boundless nonlinear optimization problem and is used to apply significant goals in the network. In this section, the objectives of the rearrangement discussion will be briefly mentioned.

12

Network Rearrangement in the Presence of Loss Reduction

Reduction of losses is one of the most basic goals of rearrangement and in order to implement it in the distribution network, knowing the diversity of energy consumers, changes the flow paths in the distribution network and thus reduces the losses of the distribution network. This is done by predicting the number of keys in different parts of the network. So that in normal operation of the system, some of these switches are in the open position, which are called maneuver switches, and on the other hand, to power all loads, a number of switches are in the closed position, which are called sectional switches. Therefore, the arrangement of feeders and distribution network is done by changing the position of maneuver and sectionalizer keys (Chua-Liang and Kirschen 2009).

13

Rearrangement in the Network in Order to Balance the Load of Feeders

Here, the network load is transferred from high load network feeders to low load feeders. This will not only reduce active power losses, but also improve system operating conditions and distribute the load evenly over the system components, eliminating or minimizing the risk of feeder or transformer overload. Load transfer is also possible through switching in the network.

14

Rearrangement in Order to Recover the Load and Increase Reliability

1-1-1- Load recovery is another important goal of rearrangement of distribution networks. This operation is performed in the event of a network fault in order to re-establish fast power supply to consumers. Using the network rearrangement, the faulty line can be separated from the circuit by the sectionalizer function and the isolated loads can be fed from another path. In this way, by reducing the exit time of network subscribers due to errors, the quality of the distribution network also improves in terms of reliability.

Energy Harvesting for Smart Energy Systems

15

1597

Rearrangement in Order to Melt the Ice of the Lines

1-1-2- One of the recent explorations on the issue of rearrangement has been devoted to the freezing of lines that are located in the winter, especially in the highlands or near the poles. Sometimes such a diameter of ice is placed on these lines that the mechanical load imposed on the line reaches more than the threshold of rupture of the lines and causes line breakage or mechanical failure of the rig and endangers the reliability of the network. In this regard, the simplest and least expensive method is to use the network rearrangement in order to pass a special current through the network lines, which causes the line to heat up and consequently the lines do not freeze. Such a flow is called a freezing critical flow, which is calculated exactly for each line (Alaee et al. 2016; Ding et al. 2020; Shafie-khah et al. 2011; Karki et al. 2006; Vandezande et al. 2010; Guimãraes et al. 2016; Eidiani et al. 2009; Mnatsakanyan and Kennedy 2015; Elmakias 2008; Eusuff and Lansey 2003).

16

Methods of Reducing Network Losses

1-1-3- There are several ways to reduce losses in distribution networks. In a recent study in Canada, the profit-to-cost ratio of each of the available methods was obtained. The results of this research are presented in Table 1. According to Table 1, it can be seen that the two methods of distribution transformers’ load management and rearrangement give the highest ratio of profit to cost. Load management of transformer is achieved by implementing distribution station automation. On the other hand, by implementing the automation of distribution stations, it is possible to apply network rearrangement. Rearrangement of distribution networks at two levels of medium pressure and low pressure. It is clear that the rearrangement of distribution networks at low pressure levels can also be considered as part of the transformer load management.

Table 1 Canadian research, profit to cost ratio in each method of loss reduction (Alaee et al. 2016) Method of reduce losses Capacitor Replacing conductors Voltage level change Load management of distribution transformers Rearrangement

Profit to cost ratio 2–8 6–7 3–5.1 1–15 13

1598

17

S. Momen et al.

Improved Frog Mutation Algorithm

Algorithms that are trans-innovative, trans-heuristic, or trans-evolutionary are random algorithms used to find optimal solutions. There are two types of optimization methods and algorithms: approximate and exact algorithms. In the case of hard optimization problems, accurate algorithms do not find the optimal solution accurately, and their execution time increases exponentially with the size of the problem. With approximate algorithms, difficult optimization problems can be solved fairly accurately (near-optimal). There are three types of approximate algorithms: hyperheuristic, meta-heuristic, and heuristic algorithms. Early convergence to local optimal points and entrapment in local optimal points are the two main problems of innovative algorithms. These problems have been solved using innovative algorithms. As one type of approximation optimization algorithm, meta-heuristic algorithms provide solutions to exiting local optimal points, and they are applicable to many different problems. This type of algorithm has been developed in numerous classes over the past few decades. These answers are virtual frogs divided into different categories. It is possible for the characteristics of a frog group to change based on the characteristics of other frog groups. By exchanging information, the frogs in each group improve their position in relation to food. The information obtained from the groups is compared after each local search. The SFLA steps are generally as follows: ⎧ ⎫ [LDR (r, d, t) (r, d, t)] + [pv .Nbat .price] + N P Cls + Pinv−load ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ z ⎪ ⎪ i d r ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎬ ⎨ upstream − πT OU × Loadt + πsell × Pt − + CostCH P − Costloss O.F. = max t t i ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ − Cost − Cost − Cost − Cost − Cost ⎪ Boiler FC WT PV P oll ⎪ ⎭ ⎩ t

k

t

j

t

r

t

g

Step I: To determine the fit of the members of the initial population, we randomly generate the number Np . Step II: The fitness of the population members is arranged in ascending order. Step III: There are m groups of frogs, with n frogs in each group, i.e., Np = n × m. According to this division, the first frog in the population belongs to the first group, the second to the second group, and the m frog belongs to the m group. Afterward, the frog number m + 1 will be placed in the first group, and so forth until n frogs are placed in each group. Step IV: Here, the local search includes the following steps:

Energy Harvesting for Smart Energy Systems

1599

ml represents the number of categories. In the first step, ml is equal to 0 and is compared to the sum of the categories (m). Local searches are also counted by the variable yl . Compare yl = 0 with the maximum local search steps in this step. This will result in: m1 = ml + 1 y1 = yl + 1 In each category, Xb and Xw are the frogs with the best and worst fitness, respectively. Xg is also the position of the most fittest frog among all the frogs. According to Fig. 1, the position of the frogs improves using the following equation. Di = rand × (Xb − Xw ) new = X old + D Xw i w − Dmax < Di < Dmax The frogs’ maximum positions are determined by Dmax , where rand is a random number between 0 and 1. When a better frog (better response) is obtained during the stage, it will be replaced with the previous frog. Otherwise, in Eq. 4, Xg will be substituted for Xb , and the above steps will be repeated until a better answer is obtained, replacing the previous answer. Otherwise, the previous frog will be replaced by a random frog. If ml < m, it should be ml = ml + 1 and if yl < ymax it should be replaced by yl = yl + 1. Otherwise we will move on to step 2. Step V: A repeat of step III is necessary if the convergence conditions are not met. It is stopped if the algorithm does not yield a suitable answer, and the output is determined by the best answer. It should be noted that when frogs with the worst fit strengthen their position relative to the group or the best frog, the classical SFLA algorithm is placed along the line Xb and Xw . As a result, the algorithm produced incorrect results. In order to improve this algorithm, a method is introduced. This method aims to prevent

Fig. 1 Improving the situation of frogs

1600

S. Momen et al.

the algorithm from converging incorrectly by expanding the direction and length of each frog’s jump. The change position vector of the frogs is selected as relation (2) if the first four frogs in the arranged frogs are randomly selected in such a way as to establish relation (1). Xg1 = Xg2 = Xg3 = Xg4

Xchange = Xg1 + r1 Xg2 − Xg3

(1)

+ r2 Xg − Xg4

new Xw,j

=

Xchange Xg

if r3 < r4 or j = rp otherwise

(2)

Assume that rp represents the random number between 1 and the number of categories in the category and j represents the number of iterations of the category. All variables r1 , r2 , r3 , and r4 have a random number between zero and one. In the event that this improvement results in a better answer, then this answer replaces the previous one. Otherwise, a random answer will be generated and replace the previous one.

18

Analysis of Results and Simulation

In this section, we present the results obtained in the simulation. The intended function is the energy losses in the network. A total of five scenarios will be considered. In the studied scenarios, we will use the issues of rearrangement, capacitance and distributed generation sources alone or in combination, and their impact on losses and grid voltage profiles will be evaluated. The IEEE 33-bus distribution network has been used to implement the problem. The load model used is the RTS model. The following figure shows the network under study. A total of 10 branches can be disconnected and connected, and with the opening and closing of each, we will have a new arrangement. In each case, 5 keys will be closed and 5 keys will be opened, and in this way we will achieve different network configurations. The important point is that each time these 10 switches are turned on and off, the radial grid arrangement must remain. These 10 keys are 33, 34, 35, 36, 37, 7, 11, 14, 17 and 28 (Fig. 2). In fact, in our programming, we have modeled the load using the IEEE RTS standard, in which each chapter is shown with a sample day.

Energy Harvesting for Smart Energy Systems

1601

Fig. 2 IEEE 33 bus network under study

19

The Following Scenarios Are Considered in Solving this Problem

19.1

Scenario 1: Basic Scenario

In this scenario, we examine the primary network with the same initial arrangement. In this case the losses are equal

1602

S. Momen et al.

Fig. 3 Voltage profile diagram in scenario 1

Table 2 The results of scenario 2 Basic makeup Rearrangement

Open keys 33, 34, 35, 36, 37 7, 34, 35, 36, 37

Closed keys 7, 11, 14, 17, 28 34, 11, 14, 17, 28

Losses (MWh) 595.0345 406.177

Ploss = 595.0345MWh Obtained. Since here we are talking about different times during a year and at these different times the network load is different, so naturally the voltage profile will also be different, but here the voltage profile during the maximum network load is discussed. We put the diagram of the voltage profile in this case is as follows (Fig. 3):

19.2

Scenario 2: Rearrangement Alone

In this scenario, we intend to rearrange the network. For this case, as mentioned, 5 keys must always be opened and 5 keys must be closed. The answer to the optimization problem is as follows (Table 2): Also, the network voltage profile diagram in this scenario is shown in the figure below along with scenario one for better comparison. As can be seen, in this case we see a significant improvement in the voltage profile (Fig. 4).

Energy Harvesting for Smart Energy Systems

1603

Fig. 4 Voltage profile diagram in scenario 2

Fig. 5 ISFLA algorithm process for scenario 2

The optimal process of the problem using the ISFLA algorithm is as follows. It should be noted that here the ISFLA algorithm with a population of 20 and a number of 15 generations is used (Fig. 5).

1604

19.3

S. Momen et al.

Scenario 3: Placement of Scattered Products in the Network

In this scenario, we will locate distributed generation resources in the network. The distributed generation sources used are renewable. In this case, we will use two types of solar source and wind source. The purpose of locating the three sources is to discuss the number, the optimal location and also to determine the optimal capacity for them. The intended objective function, like the previous scenarios, is a function of annual losses. To solve the optimization problem, a ISFLA algorithm with 15 generations and a population of 20 per generation is used. The table below shows the results obtained (Fig. 6 and Table 3). By solving the optimization problem, it is determined that the optimal state is obtained if a wind turbine with a capacity of 2.366 MW is added to bus 12 and a solar source with a capacity of 1.986 MW is added to bus 6. The amount of losses in this case is equal to 449.139MWh and the voltage profile diagram in this case compared to the baseline scenario is as follows. The blue chart is the chart in the mode after optimal DG placement. As can be seen, the voltage profile has been greatly improved after DG placement. The process of optimizing the problem in this case is as follows (Fig. 7):

Fig. 6 Voltage profile diagram in scenario 3 compared to baseline scenario

Table 3 The results of scenario 3

Location DG Bass 12 of the wind type Solar 6 bass

DG capacity (KW) 2366 1986

Energy Harvesting for Smart Energy Systems

1605

Fig. 7 How to optimize the problem in scenario 3

Table 4 Scenario Results 4

19.4

Capacitor location Bass 30 Bass 12

Capacitor Capacity (Kvar) 363 303

Scenario 4: Locating a Capacitive Bank in the Network

In this scenario, we will place the capacitor in the network. The purpose of locating the three sources is to discuss the number, the optimal location and also to determine the optimal capacity for them. The intended objective function, like the previous scenarios, is a function of annual losses. To solve the optimization problem, a ISFLA algorithm with 15 generations and a population of 20 per generation is used. The table below shows the results obtained (Table 4). The amount of losses in this case is equal to 499.027MWh and the voltage profile diagram in this case compared to the baseline scenario is as follows. The blue diagram is the diagram in the state after optimal capacitor placement. As can be seen, the voltage profile has improved after the capacitor is installed. Of course, this is not as much as the use of scattered production and rearrangement (Figs. 8 and 9). The process of the ISFLA algorithm in this scenario is as follows:

1606

Fig. 8 Voltage profile diagram in scenario 4 compared to baseline scenario

Fig. 9 ISFLA algorithm trend in scenario 4

S. Momen et al.

Energy Harvesting for Smart Energy Systems Table 5 Values obtained in scenario 5

19.5

Parameter Wind power plant bass The size of the wind farm Solar power plant bass The size of the solar power plant Capacitor charge 1 Capacitor charge 1 Capacitor load 2 Capacitor size 2 Keys attached in makeup remodel Cut keys in makeup remodeling Annual energy losses

1607 value 33 907 Kw 6 1206 Kw 6 431 Kvar 13 332 Kvar 34, 7, 11, 17, 28 14, 33, 35, 36, 37 289.199

Scenario 5: Rearrangement with Simultaneous Placement of Distributed Generation Sources and Optimal Capacitance in the Network

In this scenario, we intend to analyze all three cases of rearrangement, DG placement, and optimal capacitance simultaneously in a single optimization problem. The intended objective function, like the previous scenarios, is a function of annual energy losses. For this scenario, we will use a ISFLA algorithm with 15 generations and a population of 20 per generation. The results obtained from this scenario are given in the table below (Table 5). The diagram of the voltage profile in this case compared to the baseline scenario is as follows (Figs. 10, 11, and 12). How to optimize the problem is as follows We will now compare the different scenarios. In terms of loss reduction, as stated in the results, in the case of simultaneous use of three methods of loading, capacitance and distributed generation sources, the loss is much greater than other scenarios. In terms of voltage profiles, we better examine all five scenarios in a diagram for better comparison. This is shown in the figure below (Table 6). We will now compare the different scenarios. In terms of loss reduction, as stated in the results, in the case of simultaneous use of three methods of loading, capacitance and distributed generation sources, the loss is much greater than other scenarios. In terms of voltage profiles, we better examine all five scenarios in a diagram for better comparison. This is shown in the figure below.

20

Conclusion

In this chapter, in smart grids, the losses and voltages of an IEEE 33-bus system are investigated using network rearrangement and capacitor placement, as well as the use of scattered output. To reshape the grid, the keys are connected and then the

1608

Fig. 10 Voltage profile diagram in scenario 5 compared to baseline scenario

Fig. 11 How to optimize the problem in scenario 5

S. Momen et al.

Energy Harvesting for Smart Energy Systems

1609

Fig. 12 How to optimize the problem in scenario 5

Table 6 Results Scenario number 1 2 3 4 5

Network losses (objective function) 595.035 MWh 406.177 MWh 449.139 MWh 499.027 MWh 289.199 MWh

Voltage deviation (at worst bus) 7% 4.6% 3% 5.9% 3.1%

Blue number 18 33 33 18 18

Blue voltage 93% 95.4% 97% 94.1% 96.9%

keys are cut to make the grid radial. The effect of these factors on the reduction of losses has been shown and all factors cause this reduction of losses. Comparing the different scenarios, it can be said that in all scenarios, the voltage profile has improved compared to scenario one. But in this case, the use of distributed generation resources from all scenarios has performed better. After this is a scenario that involves the simultaneous use of all methods. In this scenario, the simultaneous use of all methods, the losses have been significantly reduced. In this respect, we have the best answer among all scenarios. But in a general summary it can be said. Because loss in the distribution network is of great importance and at the same time the simultaneous use of all methods in terms of improving the voltage profile is second in all scenarios and has acceptable conditions in this regard. Therefore, it seems that the optimal choice is the simultaneous use of all methods of rearrangement, capacitance and scattered production.

1610

21

S. Momen et al.

Offers:

• Distribution model reliability modeling using Monte Carlo simulation • Modeling the batteries of electric vehicles and examining the parameters of power quality and system losses

References H. Ahmadi-Nezamabad et al., Multi-objective optimization based robust scheduling of electric vehicles aggregator. Sustain. Cities Soc. 47, 101494 (2019) S. Alaee, R. Hooshmand, R. Hemmati, Stochastic transmission expansion planning incorporating reliability solved using SFLA meta-heuristic optimization technique. CSEE J. Power Energy Syst. 2(2), 79–86 (2016) L. Baringo, A.J. Conejo, Strategic offering for a wind power producer. IEEE Trans. Power Syst. 28(4), 4645–4654 (2013) E. Bitar, R. Rajagopal, P. Khargonekar, K. Poolla, P. Varaiya, Bringing wind energy to market. IEEE Trans. Power Syst. 27(3), 1225–1235 (2012) S. Chua-Liang, D. Kirschen, Quantifying the effect of demand response on electricity markets. IEEE Trans. Power Syst. 24(3), 1199–1207 (2009) A.A.S. de la Nieta, J. Contreras, J.I. Munoz, M. O’Malley, Modeling the impact of a wind power producer as a price-maker. IEEE Trans. Power Syst. 29(6), 2723–2732 (2014) W. Ding, Y. Sun, L. Ren, H. Ju, Z. Feng, M. Li, Multiple lesions detection of fundus images based on convolution neural network algorithm with improved SFLA. IEEE Access 8, 97618–97631 (2020) M. Eidiani, Y. Ashkhane, M. Khederzadeh, Reactive power compensation in order to improve static voltage stability in a network with wind generation, International Conference on Computer and Electrical Engineering, ICCEE 2009, 2009, 1, pp. 47–50, Dubai, UAE, December 28–30, 2009 D. Elmakias, New Computational Methods in Power System Reliability, vol 111 (Springer Science & Business Media, 2008) M. Eusuff, K. Lansey, Optimization of water distribution network design using the shuffled frog leaping algorithm. J. Water Resour. Plan. Manag. 129(3), 210–225 (2003) G. Ghardashi, M. Gandomkar, S. Majidi, M. Eidiani, S. Dadfar, Accuracy and speed improvement of microgrid islanding detection based on PV using frequency-reactive power feedback method. Proceedings of the 16th International Conference on Protection and Automation of Power Systems, IPAPS 2022, 2022 M. Ghasemi et al., An efficient modified HPSO-TVAC-based dynamic economic dispatch of generating units. Electric Power Components Syst. (2020). https://doi.org/10.1080/15325 008.2020.1731876 L. Goel, Q. Wu, P. Wang, Reliability enhancement of a deregulated power system considering demand response. IEEE Power & Energy Society General Meeting, 2006 J. A. Guimãraes, L. M. V. G. Pinto and N. Maculan, “What will be the proxy value for a Brazilian utility company triggering its demand side management in the light of price elasticity of demand?,” in IEEE Lat. Am. Trans., 14, 8, pp. 3746–3754, Aug. 2016 H.M. Hasanien, Shuffled frog leaping algorithm for photovoltaic model identification. IEEE Trans. Sustain. Energy 6(2), 509–515 (2015) M. Hayati and G. Karimi, Short-channel effects improvement of carbon nanotube field effect transistors, 2020 28th Iranian Conference on Electrical Engineering (ICEE), Tabriz, Iran, 2020, pp. 1–6, https://doi.org/10.1109/ICEE50131.2020.9260850 R. Heydari, J. Nikoukar, and M. Gandomkar, Optimal operation of virtual power plant with considering the demand response and electric vehicles. Journal of Electrical Engineering & Technology, 2021. M. Zand, M. A. Nasab, A. Hatami, M. Kargar and H. R. Chamorro, “Using

Energy Harvesting for Smart Energy Systems

1611

Adaptive Fuzzy Logic for Intelligent Energy Management in Hybrid Vehicles,” 2020 28th ICEE, pp. 1–7, 10.1109/ICEE50131.2020.9260941.IEEE Index E. Heydarian-Forushani, M. Moghaddam, M. Sheikh-El-Eslami, M. Shafie-khah, J. Catalao, Riskconstrained offering strategy of wind power producers considering intraday demand response exchange. IEEE Trans. Sustain. Energy 5(4), 1036–1047 (2014) C.S. Ioakimidis, L.J. Oliveira, K.N. Genikomsakis, Wind power forecasting in a residential location as part of the energy box management decision tool. IEEE Trans. Ind. Inf. 10(4), 2103–2111 (2014) R. Karki, P. Hu, R. Billinton, A simplified wind power generation model for reliability evaluation. IEEE Trans. Energy Convers. 21(2), 533–540 (2006) M.J. Kasaei, M. Gandomkar, J. Nikoukar, Optimal management of renewable energy sources by virtual power plant. Renew. Energy 114, 1180–1188 (2017) K. Ma, G. Hu, C.J. Spanos, A cooperative demand response scheme using punishment mechanism and application to industrial refrigerated warehouses. IEEE Trans. Indust. Inform. 11(6), 1520– 1531 (2015) A. Mnatsakanyan, S.W. Kennedy, A novel demand response model with an application for a virtual power plant. IEEE Trans. Smart Grid 6(1), 230–237 (2015) J.M. Morales, A.J. Conejo, J. Pérez-Ruiz, Economic valuation of reserves in power systems with high penetration of wind power. IEEE Trans. Power Syst. 24(2) (2009) M. Moreno, M. Bueno, J. Usaola, Evaluating risk-constrained bidding strategies in adjustment spot markets for wind power producers. Int. J. Electr. Power Energy Syst. 43, 703–711 (2012) A. Nasab et al., Optimal planning of electrical appliance of residential units in a smart home network using cloud services. Smart Cities 4, 1173–1195 (2021). https://doi.org/10.3390/ smartcities4030063 Nasri, Shohreh, et al, Maximum Power Point Tracking of Photovoltaic Renewable Energy System Using a New Method Based on Turbulent Flow of Water-based Optimization (TFWO) Under Partial Shading Conditions. 978-981-336-456-1 D.T. Nguyen, M. Negnevitsky, M.D. Groot, Walrasian market clearing for demand response exchange. IEEE Trans. Power Syst. 27(1), 535–544 (2012) J. Nikoukar, Unit commitment considering the emergency demand response programs and interruptible/curtailable loads. Turkish J. Electric. Eng. Comp. Sci. 26(2), 1069–1080 (2018) M. Parvania, M. Fotuhi-Firuzabad, Demand response scheduling by stochastic SCUC. IEEE Trans. Smart Grid 1(1) (2010) A. Rohani et al., Three-phase amplitude adaptive notch filter control design of DSTATCOM under unbalanced/distorted utility voltage conditions. J. Intell. Fuzzy Syst. (2020). https://doi.org/10.3233/JIFS-201667 P. Sanjeevikumar, et al, Spider community optimization algorithm to determine UPFC optimal size and location for improve dynamic stability, 2021 IEEE 12th Energy Conversion Congress & Exposition – Asia (ECCE-Asia), 2021, pp. 2318–2323, https://doi.org/10.1109/ ECCE-Asia49820.2021.9479149 M. Shafie-khah, P. Moghaddam, M.K. Sheikh-El-Eslami, Unified solution of a non-convex SCUC problem using combination of modified branch-and-bound method with quadratic programming. Energy Convers. Manag. 52, 3425–3432 (2011) L. Tightiz et al., An intelligent system based on optimized ANFIS and association rules for power transformer fault diagnosis. ISA Trans. 103, 63–74., ISSN 0019-0578 (2020). https://doi.org/10.1016/j.isatra.2020.03.022 L. Vandezande, L. Meeus, R. Belmans, M. Saguan, J.M. Glachant, Well-functioning balancing markets: A prerequisitefor wind power integration. Energy Policy 38, 3146–3154 (2010) M. Zand, M.A. Nasab, P. Sanjeevikumar, P.K. Maroti, J.B. Holm-Nielsen, Energy management strategy for solid-state transformer-based solar charging station for electric vehicles in smart grids. IET Renewable Power Generat. (2020). https://doi.org/10.1049/iet-rpg.2020.0399. IET Digital Library, https://digital-library.theiet.org/content/journals/10.1049/iet-rpg.2020.0399 H. Zeynal, A.K. Zadeh, K.M. Nor, M. Eidiani, Locational marginal price (LMP) assessment using hybrid active and reactive cost minimization. Int. Rev. Electric. Eng. 5(5), 2413–2418 (2010)

1612

S. Momen et al.

H. Zeynal, Y. Jiazhen, B. Azzopardi, M. Eidiani, Flexible economic load dispatch integrating electric vehicles, Proceedings of the 2014 IEEE 8th International Power Engineering and Optimization Conference, PEOCO2014, Lankawi, Malaysia, pp. 520–525, 24–25 March 2014 M. Zugno, J.M. Morales, P. Pinson, H. Madsen, Pool strategy of a price-maker wind power producer. IEEE Trans. Power Syst. 28(3), 3440–3450 (2013)

Dynamic Bayesian Network Based Approach for Modeling and Assessing Resilience of Smart Grid System Niamat Ullah Ibne Hossain and Chiranjibi Shah

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Resilience of Smart Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 The Perspective of Resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Resilience Capacities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Dynamic Bayesian Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Approach Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1 Sensitivity Analysis of the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1614 1616 1616 1618 1620 1620 1624 1628 1628 1629 1631

Abstract

Smart grid is an emerging technology in the field of power generation and linked to different critical infrastructures, including telecommunication, transportation, water supply, and fuel distribution. There exist many uncertainties in the complex smart grid systems because of rapidly increasing new technologies. Disruption in the smart grid may lead to uncertainties because of many reasons, including natural disasters, such as snowstorm, lightning, hurricane, as well as human

N. U. I. Hossain () Department of Engineering Management College of Engineering and Computer Science, Arkansas State University, State University, USA e-mail: [email protected] C. Shah Department of Electrical and Computer Engineering, Mississippi State University, Starkville, MS, USA e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_16

1613

1614

N. U. I. Hossain and C. Shah

errors to machinery failure. After the occurrence of failure, it is prudent to discover approaches for recovering from the damage instead of just focusing on preventing the failure before it occurs. The static Bayesian Network (BN) approach concentrates only on cases that are in equilibrium; however, the system may fluctuate with time because of disturbing events. Dynamic Bayesian network (DBN) based approach can be appropriate to consider temporal dimension for resilience assessment of such time changing circumstances. Most of the methods prevalent in the extant literature only design and assess resilience based on performance loss due to disruption but do not the evaluate the associated probabilities while the state of the system will go through different resilience capacities such as absorption, adaptation, restoration, learning to achieve its optimal functionality over time. This chapter adopts probabilistic assessment method leveraging dynamic Bayesian network based approach to assess the resilience of the smart grid system. It incorporates temporal analysis in absorption, adaptation, restoration, and learning for analysis of system functionality considering scenarios both during the disruption and after the disruption. Keywords

Smart grid · Power system · Resilience · Dynamic Bayesian network · Disruption

1

Introduction

The electric grid network provides power generation to several critical infrastructures, including fuel distribution, transportation, telecommunication, and water supply (Hossain et al. 2019). Modern complex systems such as the electric grid system integrate with other heterogeneous systems to achieve new goals that are beyond the capability of the individual systems. For example, the smart grid systems contain different heterogeneous distributed systems, including wireless mesh networks, metering systems, SCADA systems, remote terminal devices, and microprocessors (Bojkovic and Bakmaz 2012; Nazir et al. 2015). The primary reasons for smart grid failures are varying from a human error to technical failure to malicious cyberattack. Because the risks associated with the smart grid cannot be readily predicted or even anticipated, different government organizations and national security allies are working intensively with other stakeholders to increase the resilience of the smart grid system. On the other hand, experts are now concerned with developing an intrusion fee resilient smart grid system that could endure and recover from any disturbance. During the disruptions within the smart grid, the system changes from one state to another rapidly over time. Therefore, the resilience assessment approach is useful to capture the uncertainties involved. Traditional risk assessment tools might not be adequate to ensure safety in situations where uncertainties are unpredictable such as the case in the smart grid system (Park et al. 2013). There is a growing trend in the literature of developing methods that are pertained to the management of disruption by restoring the system’s functionality

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . .

1615

and performance after the disruption. During the pre-disruption and post-disruption stages, the resilience assessment approach is suitable for better management of any emergent threat or events that might interrupt the smart grid system. In this spirit, there is a need for metrics and models for resilience (Vamanu et al. 2016). Uday and Marais (2015) conducted a survey of resilience metrics and recognized that time interval of failure, time duration of recovery, ratio of performance recovery to performance loss, pace of recovery, performance before and after the disruption, recovery strategies are imperative factors to assess resilience. Research on the smart grid resilience has lately been furthered owing to the significant increase in the amount of natural disasters and cyber-attacks. Liang et al. (2013) used an approach of protection framework and reliability for the smart grid that offers feasible recommendations to lessen the cyber risk of the smart grid. In this work, information systems of smart grid network are analyzed that describes various security threats faced by the network. Nazir et al. (2015) used macro and micromanagement techniques for the resilience of a smart grid to obtain information exchange and better governance in the smart grid system that consists of components interconnected through an advanced communication network. It discusses benefit of including communications and computing techniques for incorporating infrastructures of electricity in the smart grid systems. Radoglou-Grammatikis et al. (2018) used appropriate protocols to deal with various cyber-attacks in the smart grid, including critical states and suitable specifications, by introducing the utility of the firewall system. This research work reported an analysis of various scientific methods related to firewalls in the smart grid system. In another research, Parra et al. (2019) used a software-defined network for dealing with security challenges in information and communication for the smart grid. In this approach, network capabilities of nodes are utilized to control communication among the controlling centers. In the same year, Xia et al. (2019) adopted an adaptive algorithm based scanning approach to detect the malicious attack in the nearby vicinity of the smart grid system within the quickest detection time. In this approach, malicious users are identified from average numbers of users within minimum time steps. Shapsough et al. (2015) developed a five-layer security conceptual model based on the internet of things (IoT) platform and offered some innovative strategies to confirm the smooth operations of smart grids. A Survey of different model architectures proposed in the smart grid is described in this work. A somewhat different approach has been adopted by Wadhawan et al. (2018), who applied a multi-echelon probabilistic framework for a cyber-physical system that can calculate the probability of cyber risks on the smart grid. It estimates the likelihood along with the impact of the attack on smart grid systems and also includes the dynamic nature of the Bayesian network for assessing vulnerabilities. The researchers also listed the countermeasures to develop a cyberintrusion-free smart grid system. Along the same line, Hossain et al. (2020) applied a static Bayesian network to evaluate the overall cyber resilience of the smart grid system. In this method, authors identified the main factors liable for cyber-attack and calculated their corresponding probabilities through the Bayesian theorems to assess the overall cyber resilience of the smart grid system.

1616

N. U. I. Hossain and C. Shah

As it could be seen, a plethora of deployment technologies evolved, and countermeasures have been listed to subside the impact of cyber-attack on the smart grid system. Even if all the proposed approaches share the same purpose, they differ in policies and supported mechanisms. However, the extant literature lacks an effort that considers temporal dimension in probabilistic assessment, the probability of system functionality changes during and after disruption in the smart grid system. To fill this void, this chapter adopted the dynamic Bayesian network (DBN) approach to assess the overall cyber resilience of a smart grid system over adjacent time steps. The variables in the underlying DBN are interconnected to themselves or other variables in different time slices. As a result, DBN captures the temporal dependence or time dimension in reasoning. Thus, the prediction of cyber resilience of the smart grid system over time can be quantified as well. The following is a summary of the main contributions of the proposed chapter: • Identify the potential factors related to resilience capacities that could enhance the overall resilience of the smart grid system. • Develop a dynamic Bayesian network to diagnose the overall resilience of the smart grid system via an illustration of a numerical example. • Demonstrate the efficacy of the proposed model for the faster recovery of the smart grid system from disruption.

2

Resilience of Smart Grid

2.1

The Perspective of Resilience

Resilience is characterized as the capability of a system to respond and to withstand a disruption (Haimes 2009; Vugrin et al. 2010). Most research defines resilience based on the effect of unexpected disturbances on the overall performance of the system by means of functionality or performance loss (Henry and RamirezMarquez 2012; Youn et al. 2011). For an effective resilience methodology, a “low functionality” state and high functionality” state can be defined when a system falls into malfunction and recovers to its original functional state. Bruneau et al. (2003) characterized four measurements for resilience in the notable resilience triangle model: (i) robustness (strength of the system), (ii) rapidity (speed at which a framework can re-visit its unique state or, a satisfactory degree of usefulness after an interruption), (iii) resourcefulness (degree of ability to apply data, mechanical, physical and human resources to react to a problematic occasion), and (iv) redundancy (the degree to which a framework limits the probability and disturbance effects). Bruneau et al. (2003) research also provided a deterministic static measurement of resilience loss of a network to a seismic tremor as indicated by Eq. (1). The time at which the interruption happens is t0 , and the time at which the network re-visits

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . .

1617

its normal pre-disturbance state is t1 . The nature of the network framework at time t, which could speak to various types of execution measures, is expressed as function of Q(t). RL =

t1

[100 − Q(t)] dt

(1)

t0

Henry and Ramirez-Marquez (2012) created a time-reliant resilience metric that measures resilience as a proportion of recuperation time to misfortune time. Given that the presentation of the framework at a point in time is estimated by the performance function ϕ(t), three frameworks that are significant in measuring flexibility are presented in Fig. 1: (i) the steady unique state, which expresses the ordinary practicality of a framework before an interruption, begins at time t0 and ends at time te , (ii) the disrupted state, which is caused by a problematic occasion (ej ) at time te and whose impacts are felt until time td , is depicted from time td to ts , and (iii) the stable recovered state, which refers to the new consistent execution level following completion of the recuperation activity started at time ts is depicted from tf onwards. Significant components of versatility that are shown in Fig. 1 include the capacity to keep up regular activity preceding a disturbance, the capacity to fight off adverse effects after occasion ej , and the capacity to recuperate in an ideal way from ej . Equation (2) notes that the resilient behavior adopted by Whitson and Ramirez-Marquez (2009), is generally considered reliable. ϕ t|ej − ϕ td |ej j t|e = ϕ (t0 ) − ϕ td |ej

(2)

As described above, the numerator of this metric indicates recuperation up to time t, whereas the denominator represents complete misfortune because of disturbance ej .

Fig. 1 Framework execution and state progress to describe resilience. (Adapted from Henry and Ramirez-Marquez (2012))

1618

N. U. I. Hossain and C. Shah

The authors additionally determined the total expense of the recuperated framework followed by disruption as the sum of the execution cost for resilience action and the misfortune cost of non-operability due to the interruption. A few ensuing advancements with regard to resilience measurement and arrangement depend on the framework state of change referred to in Fig. 1 and the measurement in Eq. (2) by Henry and Ramirez-Marquez.

2.2

Resilience Capacities

The execution of resilience in a system necessitates the incorporation of both analytic and holistic processes. Specifically, the utilization of architecting with correlated heuristics is essential. Inputs are the expected resilience level and the attributes of interruptions/perturbations while outputs are the system’s attribute, particularly the structural aspects and the nature of the components (e.g., hardware, software, or humans). The underlying characteristics of resilience are the capacity of the system to absorb, adapt, restore, and learn from unexpected disruptions (Francis and Bekera 2014). These characteristics are translated into recursive actions through a cyclic order to maintain the desired state of the system. Absorption is the inherent ability of a system to resist and survive from disruption; adaptation refers to the capability of the system to adapt itself to the exposure of the shock, while restoration means the intentional efforts to stop or repair the damages caused by disturbances and restore it to its normal state (Hossain et al. 2019; Tong et al. 2020). Learning is the cumulative knowledge obtained from the past incidents or observations in a sense that how different strategies and actions have been implemented or adopted for a similar event in the past to subside, halt or completely recover the damage due to the internal or external perturbations (Park et al. 2013). An illustration of the above discussion is presented in Fig. 2, where it is observable that during the unexpected disturbance, the state of a system transforms from P1 to P2 with functionality dips down sharply from F1 to F2 during the time steps t to t + 1 (Hosseini et al. 2016; Haimes 2009; Tong et al. 2020). As soon as the overall functionality gets its lowest, the system starts to heal itself by regaining its functionality and subsequently runs to state P2 from P3 through adaptation (from time step t + 1 to t + 2). At this stage, the overall system functionality can be denoted by F3 . Finally, during the restoration process (from time step t + 2 to t + 3), the system reaches its desired functional level (F4 ) at state P4 . Learning ability of the system via past unexpected events also catalysts the functionality rate and determines how fast the system could reach its desired level.

2.2.1 Absorptive Capacity Absorptive capacity is a more internal measure of a system. It is the capacity for a system to self-regulate the overall impact of a disruption to the system in order to minimize the shock of said disruption (Hossain et al. 2019; Béné et al. 2008; Lengnick-Hall and Beck 2009). The absorptive capacity is considered as an endogenous component of the system. Absorptive capacity is described as the

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . .

1619

F4 F1 P1 Functionality

Disruption (P1-P2)

P3 F3

Restoration + Learning (P3-P4)

P2 F2

Adaptation (P2-P3)

t

t+1

t+2

Time

t+3

Fig. 2 Functionality curve under a disruption (Hosseini and Barker 2016)

customary capacity to manage emergencies with specific abilities emphasized in system design. It is also known as the line of defense that depends on assets and choices that are ordinarily accessible during non-disaster times. These capacities permit the system to naturally overcome the effects of system perturbations.

2.2.2 Adaptive Capacity Adaptive capacity is the ability of a system to be self-reliant by utilizing nonstandard practices to defeat the impact of the interruption (Hossain et al. 2019; Lengnick-Hall and Beck 2009; Béné et al. 2008). Adaptive capacity is the second line of protection because of the expanded degree of exertion it takes to minimize disturbances to the system. It is the behavioral characteristic of organizing or reconfiguring additional assets to mitigate the impacts of disruptions. 2.2.3 Restorative Capacity Restorative capacity is the capacity of a framework to be fixed effectively and productively. These practices profoundly change framework structures and objectives (Hossain et al. 2019; Lengnick-Hall and Beck 2009; Béné et al. 2008). The restorative capacity is the last safeguard as it requires the highest degree of exertion and it is not normally utilized unless the absorptive and adaptive limits are inadequate to handle the perturbation of the system.

1620

3

N. U. I. Hossain and C. Shah

Dynamic Bayesian Network

Although the underlying logical reasoning for both BN and DBN are the same, to compensate the limitations of the static Bayesian network, in a dynamic Bayesian network (DBN), the temporal dimensions are added as an extension of the static Bayesian network (BN). The temporal relationship among the variables is timestamped, and the state of the variable at time t relies only on its immediately preceding variables (Ayele et al. 2016). For instance, if a DBN structure consists of series of random variables A = A1 , A2 , A3 . . . An , the temporal dependencies among the variables over a discretized timeline can be computed using the following equation: P (At At−1 ) =

n i=1

P Ai, |P a Ai,t

(3)

where Ai,t denotes the ith variable at time t, pa(Ai,t ) refers to the parent nodes of Ai,t at the corresponding time-slice. The above equation can be further streamlined as follows: n t P A = P A1 , A2 , A3 , . . . , An = . . . pa A0i P Ati |At−1 , pa A i i i=1

(4) where pa A0i . . . pa Ati represents parent node at time step 0, . . . , t−2 , t−1 , t, and At−1 signifies node’s previous step at time step t−1 . i

4

Approach Details

The rate at which transitions occur among the states of P1 , P2 , P3 , and P4 is predicated based on the four capacities of the system. These four capacities are absorption, adaptation, restoration, and learning. The proposed methodology of this chapter is outlined in Fig. 3 (extended from Tong et al., 2020). In order to grasp a clearer understanding of the state transitions, Fig. 4 demonstrates how the Markov model can be transformed into dynamic Bayesian Network. Figure 4a illustrates the associated transition probabilities γ 1 , β 1 , γ 2 , β 2 for the absorption, restoration, adaptation, and learning capacities of the system, while Fig. 4b demonstrates DBN network to explain how the functionality state of the system would change from P1 to P4 based on the influence exerted by each resilience capacity node. The proposed model takes each transition probability of every state and converts it into a conditional probability of “state of functionality” node for each resilience capacity. An illustration of this concept would be converting transition probability γ 1 contained in the Markov chain into a conditional probability of the state of functionality changes from state P1 to state P2 during the absorption state in the dynamic Bayesian network (Zinetullina et al. 2020). These changes of functionality

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . . Fig. 3 Flowchart of the proposed model

1621

1622

N. U. I. Hossain and C. Shah

Fig. 4 Transformation of Markov chain to DBN considering the change of functionality of the system (Tong et al. 2020)

under different states are illustrated in Fig. 4. The system starts a new cycle once it reaches stable state P4 , and the transition probability from P4 to P1 is γ 2 . In other words, γ 2 is the constant failure rate under normal operating conditions, which can be further calculated is following equation (Tong et al. 2020; Zinetullina et al. 2020):

γ2 =

1 MT BF

(5)

where MTBF is the mean time between failures. Also, repair rate can be streamlined as

β2 =

1 MT T R

(6)

where MTTR is the mean time to repair. Figures 5 and 6 visualizes the state changes from the previous time step of t−1 to time t for the different discrete-time slices and demonstrates the dynamic relationships between them. The resilience attributes, along with disruption at the current state, and the functionality of the system at the previous discrete time step, will combinedly contribute to the output for the state change of system functionality. However, when the system faces external perturbation that hinders the system’s performance, the system’s resilience will be affected, and the overall state of functionality (P1 –P4 ) can be summed at the various discrete time steps in order to quantify the resilience for that specific time step (Tong et al. 2020).

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . . Fig. 5 Quantification of system changes of system functionality using DBN model (Zinetullina et al. 2020)

1623

State of Functionality Disruption

Absorption

Adaptation

Restoration

Learning

Fig. 6 Discrete-time slice for dynamic representation of DBN

Disruption

t

Learning

Restoration

Adaption

Absorption

State of functionality

t–1

State of functionality Absorption Adaption Restoration Learning

1624

5

N. U. I. Hossain and C. Shah

Case Study

The conventional power grids focus on fundamental operations like distribution, generation, and control of the quality of electric power. For more efficient transmission of electricity, the United States has taken different developmental initiatives for smart grid from the last decade. A policy at the federal level was postulated that the Energy Independence and Security Act of 2007 set funding of $100 million/year from 2008 for advancing and making efficient smart grid capabilities (ref). The architecture of such a smart grid is illustrated in Fig. 7. Interested readers are directed to Majeed Butt et al. (2021) to have a detailed understanding of the working principles for smart grid. In order to quantify the resilience of the smart grid system network and demonstrate the multi-echelon relationship of the contributing factors belong to different resilience capacities, an underlying Bayesian model is developed as depicted in Fig. 8. For the real-case scenarios, the conditional probability can be estimated in three ways depending upon the historical data of MTBF and MTTR: (i) MTBF for the smart grid during usual circumstances and following disturbances, (ii) the responding time of the smart grid for restoring its functionality by itself, and (iii) the MTTR due to maintenance from external sources (Adedigba et al. 2016a; Tong et al. 2020; Weber and Jouffe 2006). MTBF(mean time between failures) for the smart grid during usual condition can be obtained through the historical data from the respective authority, and the failure rate Y2 could be could as Y2 = MT1BF for usual conditions). The MTBF for the smart grid following the disruption, which refers the period between the occurrence of damage and the failure of the smart grid, is associated with Y1 , (where Y1 = MT1BF following damaging) (Zinetullina et al. 2020; Tong et al. 2020). If the smart grid network experiences any disruption, it might start repairing itself prior to the additional maintenance. This phenomenon can be denoted as adaptation, and the period during the adaptation process is

Fig. 7 Smart grid architecture (Majeed Butt et al. 2021)

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . .

State of functionality

Disruption

Absorption

1625

Restoration Restorative self-healing

Learning

AMI

Adaptation Restoration of control

Conditioning Monitoring Visualization Technology

Grid Partitioning

Restoration of service

Service restoration FLISR

SCADA WAMS

Large Battery Pack

AI in Cyber Defense

PMU WADC

Delay Adaptive Control

CACC

AICA

Intelligent power flow

Fig. 8 System resilience of DBN for smart grid

denoted by responding time (RT). For instance, if the smart system undergoes any disturbance due to the operational error, the response time during the adaptation 1 (Tong et al. 2020). Similarly, is estimated with period can be expressed as β1 = RT the capability of the maintenance from external sources, MTTR is the time between the end of the response of a system and the activity period of the external resources, which can be calculated as follows: β2 = MT1T R (Zinetullina et al. 2020). Weights of the proposed approach were the adjusted classical data for different networks or systems depending upon expert knowledge or specialists. Absorption, Adaptation, restoration, are critical factors that contribute towards the resilience of a smart grid system that is all affected by learning, factors from external sources, and even the system by itself. There are different contributing factors such as conditioning monitoring, Advance metering infrastructure (AMI), and visualization technology that influence the absorption capacity of the smart grid. Adaptation capability can be affected by elements such as large battery packs, AI in cyber defense, grid partitioning, and delay adaptive control. In an analogs fashion, restorative ability can be affected by restorative self-healing, restoration of control, and restoration of service. To delve further, conditioning monitoring can be further affected by wide area measurement system (WAMS), phase measurement unit (PMU), wide area damping controller (WADC), Supervisory control and data acquisition system (SCADA), and AI in cyber defense can be influenced by Autonomous intelligent cyber defense agent (AICA), Cooperative adaptive cruise

1626

N. U. I. Hossain and C. Shah

control (CACC), and intelligent power flow. In a similar vein, Fault detection, location, isolation, and service restoration (FLISR) and service restoration can be utilized to predict the restoration of service (Hossain et al. 2020). Figure 8 conveys the detailed architecture for the resilience depending upon DBN topology for smart grid system, which is related to the smart grid system with analytical differences. These different factors are changes in the parent nodes and child nodes contributing towards the resilience of the system. Fuzzy Analytical Hierarchical process (FAHp) is applied to obtain conditional probabilities that are related to weights of each parent node of the DBN smart grid topology. Table 1 illustrates a scale of conversion in a triangular fuzzy manner (Bozbura et al. 2007) based on the expert analysis through questionnaires-based survey. Later, the output of the pairwise comparison is summarized in Table 2 in detail, each answer leading to a different triangular fuzzy scale value. The average of answers recorded from the questionnaire among the different numbers of experts can be taken into consideration for the importance of parameters. For instance, one expert may assume AMI as strongly more important than visualization technology resulting in a triangular fuzzy scale of (3/2,2,5/2). In a similar manner, fuzzy scale of different factors associated in Fig. 8 are estimated as in Table 2 in order to obtain weights that can be obtained using Chang’s extent analysis (Chang 1996) as summarized in Table 3. The causal influence of the parent node then can be used to estimate the impact of the parent node on the child node (Adedigba et al. 2016b). Considering the high state of learning and occurrence of disruption at the 0th time step, the resilience of the system can be obtained for different time steps that depend upon the previous time step. All the contributing factors are considered to have a state of high and low, and the situation of every contributing factor is kept high with the state of learning also being high. Conditional probabilities of node learning upon absorption, adaptation, and restoration are considered as 0.35. Figure 9 shows that the time to recover 89%

Table 1 Conversion scale for triangular fuzzy (Bozbura et al. 2007) Scale of linguistic Equal Equally important Slightly more important Significantly more important Very significantly more important Absolutely more important

Triangular fuzzy scale (1,1,1) (0.5,1,3/2) (1,3/2,2) (3/2,2,5/2) (2,5/2,3) (5/2,3,7/2)

Reciprocal triangular fuzzy scale (1,1,1) (2/3,1,2) (1/2,2/3,1) (2/5,1/2,2/3) (1/3,2/5,1/2) (2/7,1/3,2/5)

Table 2 Parameters with pairwise comparison Absorption AMI Visualization technology Conditioning monitoring

AMI (1,1,1) (0.5,2/3,1) (2/5,1/2,2/3)

Visualization technology (1,3/2,2) (1,1,1) (0.5,2/3,1)

Conditioning monitoring (1.5,2,2.5) (1,3/2,2) (1,1,1)

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . .

1627

Table 3 Weights of contributing factors in overall resilience Child nodes Absorption

Parent nodes AMI Visualization technology Conditioning monitoring Adaptation Grid partitioning Large battery packs Delay adaptive control AI in cyber defense Restoration Restoration of control Restorative self-healing Restoration of service Conditioning monitoring SCADA PMU WAMS WADC CACC AI in cyber defense Intelligent power flow AICA Restoration of service Service restoration FLISR

Fig. 9 Dynamic resilience of smart grid system

Conditional probabilities/weights 0.5552 0.3489 0.0957 0.3379 0.2859 0.2251 0.1509 0.4497 0.3432 0.2069 0.3763 0.3491 0.2540 0.0203 0.5552 0.3489 0.0957 0.6944 0.3055

1 0.9

Resilience

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0

5

10

15 Time

20

25

30

of lost resilience of 0.89 × 0.70 + 0.30 = 0.9230 is 6 time steps (11 − 5 = 6 time steps). It can be observed that absorptive, adaptive, and restorative abilities of systems are used as 0.89, 0.78, and 0.82, respectively for estimating the resilience of the system. The final resilience value is calculated as 0.93, which is less than 1 (0.97) because lower absorption and restoration ability are assumed in the case as depicted in Fig. 9.

1628

N. U. I. Hossain and C. Shah

6

Discussion

6.1

Sensitivity Analysis of the Model

Figure 8 shows the factors that contribute for the resilience of system and Table 4 lists the number of such contributing factors. Figure 10 presents the analysis of the sensitivity by showing plot of sensitivity with different resilience values in terms of conditional probabilities for varying γ1 β 1 , γ2 , andβ 2 , It can be observed that the variation in self-repair rate (β 1 ) and failure rate over usual condition (γ2 ) affect more towards the change on the estimated value of resilience in comparison to the variation on the rate of repairing by efforts from external sources (β 2 ) and failing rate over disruptions (γ1 ). Moreover, in Fig. 10a, change of contributing factors adaptation and restoration contribute more towards change in resilience than the change in restoration for recovery. Thus, results on both Figs. 10 and 11a are consistent. Change of γ2 and β 1 also influence the time to 89% recovery more than γ2 and β 2 . As the change of γ2 increases, it consumes much recovery time for attaining normal state and less time is required if the change of β 2 , γ1 and β 1 becomes larger. Moreover, Fig. 11c, it shows final stabilized resilience line with red line (base line) of 0.93 when all the nodes in Table 4 are in the high state. The blue line displays the resilience of the system for the cases if any of the nodes fails to perform (low), and the remaining nodes are operating in good condition (high). It is illustrated that the resilience is lowest if node 1 (AMI) fails followed by nodes 13 (restoration of control), 12 (AICA), 6 (WADC), 7 (Grid partitioning), and 11 (intelligent power flow) with remaining nodes having smaller effect towards resilience. In Fig. 11b, it

Table 4 Contributing factors along with node number

Number of nodes 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Node name AMI Visualization technology SCADA PMU WAMS WADC Grid partitioning Large battery pack Delay adaptive control CACC Intelligent power flow AICA Restoration of control Service restoration FLISR Restorative self-healing

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . .

1629

Fig. 10 Sensitivity analysis for conditional probabilities (a) varying γ 1 (b) varying β 1 (c) varying γ 2 (d) varying β 2

shows time variance to recover 89% of lost resilience with change in contributing factors. It is utilized for making the decision about allocating the resources for assuring about the smart grid system if it is operating properly prior to, during, and following disruptions. For designing a robust smart grid system, many sources need to be allocated to improve the learning ability of the system. Additionally, Fig. 11d shows that the system having lower redundancy value or lower robustness value will require lower time for achieving 89% t improvement from the damaged resiliency. So, it is apparent from the discussion that designing a system with higher resilience, proper level of robustness, and redundancy needs to be applied as too higher or too lower level can result in the reduction of the resilience for the overall system.

7

Conclusion

When resilience is measured through inherent properties, it can be considered as a constant value. However, with the influence of the external factors, resiliency becomes dynamic and uncertain. A dynamic and probabilistic method for assessing the resiliency of a system is conducted to measure the probability of sustainability or restorability into a normal state for a system. It incorporates both probabilistic and temporal approaches for resiliency assessment in the smart grid. When disruptions occur, the model developed can assess probabilities and estimates dynamic

1630

N. U. I. Hossain and C. Shah

Fig. 11 (a) Change of resilience for varying contributing factors, (b) change of time for varying contributing factors, (c) sensitivity analysis of contributing factors, (d) time to recover 89% recovery with failed nodes, (e) system resilience for observation of P4 at 30 min

quantification of resilience such that it can define the speed of recovery to a normal state after disruptions. Additionally, it can provide resilience assessment in dynamic manner, which varies upon different time steps. To reduce the limitations, following recommendations can be incorporated in future study: considering other variation pattern instead of exponential rate of failure among different time steps may enhance the resilience performance of system. In addition, time step to recover

Dynamic Bayesian Network Based Approach for Modeling and Assessing. . .

1631

from disrupted node may change depending upon the resilience performance of the system, and DBN-based technique for assessing the resilience can be applied on any other systems to get better resilience performance.

References S.A. Adedigba, F. Khan, M. Yang, Dynamic safety analysis of process systems using nonlinear and non-sequential accident model. Chem. Eng. Res. Des. 111, 169–183 (2016a) S.A. Adedigba, F. Khan, M. Yang, Process accident model considering dependency among contributory factors. Process Saf. Environ. Prot. 102, 633–647 (2016b) Y.Z. Ayele, J. Barabady, E.L. Droguett, Dynamic Bayesian network-based risk assessment for arctic offshore drilling waste handling practices. J. Offshore Mech. Arct. Eng. 138(5), 62 (2016) C. Béné, A. Newsham, M. Davies, Review article: resilience, poverty and development. Annu. Conf. Hum. Dev. Capab. Assoc. New Delhi 623, 1–30 (2008) Z. Bojkovic, B. Bakmaz, in Smart grid communications architecture: a survey and challenges. Proceedings of the 11th International Conference on Applied Computer and Applied Computational Science (ACACOS) (2012), pp. 83–89 F.T. Bozbura, A. Beskese, C. Kahraman, Prioritization of human capital measurement indicators using fuzzy AHP. Expert Syst. Appl. 32, 1100–1112 (2007) M. Bruneau, S.E. Chang, R.T. Eguchi, G.C. Lee, T.D. O’Rourke, A.M. Reinhorn, M. Shinozuka, K. Tierney, W.A. Wallace, D. von Winterfeldt, A framework to quantitatively assess and enhance the seismic resilience of Communities. Earthquake Spectra, 19(4), 733–752 (2003). https:// doi.org/10.1193/1.1623497 D.-Y. Chang, Applications of the extent analysis method on fuzzy AHP. Eur. J. Oper. Res. 95, 649–655 (1996) R. Francis, B. Bekera, A metric and frameworks for resilience analysis of engineered and infrastructure systems. Reliab. Eng. Syst. Saf. 121, 90–103 (2014) Y.Y. Haimes, On the definition of resilience in systems. Risk Anal. 29(4), 498–501 (2009) D. Henry, J.E. Ramirez-Marquez, Generic metrics and quantitative approaches for system resilience as a function of time. Reliab. Eng. Syst. Saf. 99, 114–122 (2012) N.U.I. Hossain, R. Jaradat, S. Hosseini, M. Marufuzzaman, R.K. Buchanan, A framework for modeling and assessing system resilience using a Bayesian network: a case study of an interdependent electrical infrastructure system. Int. J. Crit. Infrastruct. Prot. 25, 62–83 (2019) S. Hosseini, K. Barker, Modeling infrastructure resilience using Bayesian networks: a case study of inland waterway ports. Comput. Ind. Eng. 93, 252–266 (2016). https://doi.org/10.1016/j.cie.2016.01.007 S.Hosseini, K. Barker, J.E. Ramirez-Marquez, A review of definitions and measures of system resilience. Reliab. Eng. Syst. Saf. 145, 47–61 (2016) N.U. Ibne Hossain, M. Nagahi, R. Jaradat, C. Shah, R. Buchanan, M. Hamilton, Modeling and assessing cyber resilience of smart grid using Bayesian network-based approach: a system of systems problem. J. Comput. Des. Eng. 7(3), 352–366 (2020). https://doi.org/10.1093/jcde/qwaa029 C.A. Lengnick-Hall, T.E. Beck, Resilience Capacity and Strategic Agility: Prerequisites for Thriving in a Dynamic Environment (College of Business, San Antonio, 2009), pp. 39–69 X. Liang, K. Gao, X. Zheng, T. Zhao, in A study on cyber security of smart grid on public networks. 2013 IEEE Green Technologies Conference (GreenTech) (IEEE, 2013), pp. 301–308 O. Majeed Butt, M. Zulqarnain, T. Majeed Butt, Recent advancement in smart grid technology: future prospects in the electrical power network. Ain Shams Eng. J. 12(1), 687–695 (2021). https://doi.org/10.1016/j.asej.2020.05.004 S. Montani, L. Portinale, A. Bobbio, D. Codetta-Raiteri, Radyban: a tool for reliability analysis of dynamic fault trees through conversion into dynamic Bayesian networks. Reliab. Eng. Syst. Saf. 93, 922–932 (2008)

1632

N. U. I. Hossain and C. Shah

S. Nazir, H. Hamdoun, J. Alzubi, Cyber attack challenges and resilience for smart grids. Eur. J. Sci. Res. 134(1), 111–120 (2015) T. Nielsen, Bayesian Networks and Decision Graphs (Springer, Berlin, 2007) J. Park, T.P. Seager, P.S.C. Rao, M. Convertino, I. Linkov, Integrating risk and resilience approaches to catastrophe management in engineering systems. Risk Anal. 33, 356–367 (2013) G.D. Parra, P. Rad, K.R. Choo, Implementation of deep packet inspection in smart grids and industrial Internet of Things: challenges and opportunities. J. Netw. Comput. Appl. 135, 32– 46 (2019). https://doi.org/10.1016/jJnca.2019.02.022 P. Radoglou-Grammatikis, P. Sarigiannidis, T. Liatifis, T. Apostolakos, S. Oikonomou, in An overview of the firewall systems in the smart grid paradigm. 2018 Global Information Infrastructure and Networking Symposium (GIIS) (2018). https://doi.org/10.1109/giis.2018.8635747 S. Shapsough, F. Qatan, R. Aburukba, F. Aloul, A.R. Al Ali, in Smart grid cybersecurity: challenges and solutions. 2015 International Conference on Smart Grid and Clean Energy Technologies (ICSGCE) (IEEE, 2015), pp. 170–175 Q. Tong, M. Yang, A. Zinetullina, A dynamic Bayesian network-based approach to resilience assessment of engineered systems. J. Loss Prev. Process Ind. 65, 104152 (2020). https://doi.org/10.1016/j.jlp.2020.104152 P. Uday, K. Marais, Designing resilient systems-of-systems: a survey of metrics, methods, and challenges. Syst. Eng. 18(5), 491–510 (2015) B.I. Vamanu, A.V. Gheorghe, P.F. Katina, Critical Infrastructures: risk and Vulnerability Assessment in Transportation of Dangerous Goods (Springer, Cham, 2016) E.D. Vugrin, D.E. Warren, M.A. Ehlen, R.C. Camphouse, A framework for assessing the resilience of infrastructure and economic systems, in Sustainable and Resilient Critical Infrastructure Systems, (Springer, Berlin/Heidelberg, 2010), pp. 77–116 Y. Wadhawan, A. AlMajali, C. Neuman, A comprehensive analysis of smart grid systems against cyber-physical attacks. Electronics 7(10), 249 (2018) P. Weber, L. Jouffe, Complex system reliability modelling with Dynamic Object Oriented Bayesian Networks (DOOBN). Reliab. Eng. Syst. Saf. 91, 149–162 (2006) J.C. Whitson, J.E. Ramirez-Marquez Resiliency as a component importance measure in network reliability. Reliability Engineering & System Safety. 94(10), 1685–1693 (2009). https://doi.org/ 10.1016/j.ress.2009.05.001 X. Xia, Y. Xiao, W. Liang, ABSI: an adaptive binary splitting algorithm for malicious meter inspection in smart grid. IEEE Trans. Inf. Forensics Secur. 14(2), 445–458 (2019). https://doi.org/10.1109/tifs.2018.2854703 M. Yang, F.I. Khan, R. Sadiq, Prioritization of environmental issues in offshore oil and gas operations: a hybrid approach using fuzzy inference system and fuzzy analytic hierarchy process. Process Saf. Environ. Prot. 89, 22–34 (2011) B.D. Youn, C. Hu, P. Wang, Resilience-driven system design of complex engineered systems. J. Mech. Des. 133, 101011 (2011) A. Zinetullina, M. Yang, N. Khakzad, B. Golman, Dynamic resilience assessment for process units operating in Arctic environments. Saf. Extreme Environ. 2(1), 113–125 (2020)

Application of Machine Learning in Occupant and Indoor Environment Behavior Modeling: Sensors, Methods, and Algorithms Farzad Dadras Javan, Hamed Khatam Bolouri Sangjoeei, Behzad Najafi, Alireza Haghighat Mamaghani, and Fabio Rinaldi

Contents 1 2 3 4

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Literature Review on Data-Driven Occupancy Estimation . . . . . . . . . . . . . . . . . . . . . . . . Literature Review on Machine Learning-Based Window State Estimation . . . . . . . . . . . Literature Review on Machine Learning-Based Window Opening/Closing Action Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Literature Review on Machine Learning-Based Indoor Temperature Prediction . . . . . . 6 Literature Review on Machine Learning-Based Occupancy Status Prediction and Implementation of Occupant-Centered HVAC Control . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1634 1635 1637 1641 1645 1652 1654

Abstract

The present chapter is focused on providing a comprehensive perspective of the applications of sensor-driven machine learning-based methodologies for occupant and indoor environment behavior modeling. In the first part of the chapter, various methodologies employed for non-intrusive occupancy status estimation, including the utilized sensors, feature generation methods, and detection algorithms, are reviewed. The second part is instead dedicated to comparing different methods that have been proposed in the literature for estimating the status of windows. Next, a thorough review on data-driven approaches utilized for simulating and predicting the thermal behavior of indoor environments is

F. Dadras Javan · H. Khatam Bolouri Sangjoeei · B. Najafi () · F. Rinaldi Dipartimento di Energia, Politecnico di Milano, Milan, Italy e-mail: [email protected]; [email protected]; [email protected]; [email protected] A. Haghighat Mamaghani Department of Chemical Engineering, University of Waterloo, Waterloo, ON, Canada e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_18

1633

1634

F. Dadras Javan et al.

provided. Finally, the results of studies dedicated to machine learning-based occupancy prediction and implementing occupant-centered HVAC control are reviewed. For each case, the most promising set of sensors and algorithms, utilizing which has been proved in the previous studies to result in achieving a promising performance, have been provided. In addition, the methodologies that can be employed in order to simplify the corresponding pipelines, enhance the achieved accuracy, and facilitate the physical interpretation of the obtained results have also been discussed. Keywords

Machine learning · Smart building · Window status estimation · Indoor temperature prediction · Occupancy estimation · Internet of things

1

Introduction

Consumption of the buildings accounts for 30% of the global final energy use and is responsible for 55% of the global electricity consumption (IEA 2019). Since 2000, the building-related energy use has steadily been increasing at the rate of around 1.1% annually, which is attributed to the rapidly growing use of conditioned spaces (around 65% of increment in the conditioned area since 2000) (IEA 2019). Twenty-eight percentage of the total energy-related CO2 emissions is attributed to the building sector; 66% of the latter share is related to the emissions from electricity generation utilized in buildings. Since 2000, the emission intensity per square meter is increased more than 25%, and the elevated pace of increase in electrical consumption of buildings is putting pressure on the electrical grids in many countries. It is estimated that in case the current trend of increment in the building sector’s energy usage persists, the corresponding energy consumption will be increased by 88% till 2050 (Commerell et al. 2018). Therefore, any attempt to decrease the consumption of buildings can have a notable impact on the global energy demand and can thus be considered a notable step toward decreasing the global emissions in view of the current climate change concerns. In the recent years, owing to the rapid progress in the Internet of Things (IoT) technology and the significant reduction in the corresponding cost, smart building solutions (employing sensor-driven methodologies) have received increasing attention. The smart building framework includes a wide range of solutions, among which smart HVAC systems and smart lighting are the most commonly utilized approaches for reducing the energy consumption. Occupancy detection or estimation is a substantial step in implementing smart lighting systems and smart HVAC solutions (Narayanan et al. 2018). It has recently been demonstrated that implementing occupancy-based HVAC management system can result in an energy saving in the range of 30% to 42% (Kleiminger et al. 2013). Different approaches (employing various sensors and detection methodologies) are utilized for estimating the occupancy status. Among these alternatives, the most

Application of Machine Learning in Occupant and Indoor Environment. . .

1635

promising solutions are the ones which facilitate obtaining an elevated detection accuracy with the lowest possible cost of required sensors and the minimum operating cost (owing to the necessity of utilizing cloud resources). Thus, the key objective while designing occupancy detection systems is achieving an elevated accuracy while minimizing the required cost (thus reducing the number of utilized measurements) and the corresponding calculation cost. In addition, successful implementation of smart HVAC systems requires an accurate estimation of the status of the windows. Opening and closing the windows substantially modify the thermal behavior of the indoor environment, which should clearly be taken into account while optimizing the operation of the HVAC systems. Another principal step in implementing smart HVAC system (including smart thermostats) is developing accurate models of the thermal behavior of indoor environment (with and without the effect of HVAC system) (Bonneau et al. 2017). The latter models permit optimizing the operation of the HVAC system (specifically the corresponding schedule) aiming at reducing the consumption while guaranteeing the occupants’ thermal comfort. While developing data-driven modeling for indoor temperature prediction, different parameters related to ambient conditions, thermal behavior of the building, and the occupants’ activity should be considered. Accordingly, in the present chapter, a comprehensive literature review on the previous studies focused on utilizing machine learning (ML)-based estimation of occupancy status, window status, and indoor thermal behavior is performed. The results of the studies focused on predicting the occupancy status and implementing the occupant-centered HVAC control have also been discussed. For each case, the most promising set of sensors and algorithms, utilizing which has been proved in the previous studies to result in achieving a promising performance, have been provided. In addition, the methodologies that can be employed in order to simplify the corresponding pipelines, enhance the achieved accuracy, and facilitate the physical interpretation of the obtained results have also been discussed.

2

Literature Review on Data-Driven Occupancy Estimation

Several studies have been conducted on data-driven occupancy estimation employing different types of sensors and methodologies. Rya and Moon (2016) proposed a data-driven model using the data obtained from indoor environmental sensors, cameras, and motion sensors. The obtained results of their study showed that the proposed data-driven models using the decision tree and hidden Markov model (HMM) algorithms are well-suited for occupancy detection and can be utilized for HVAC systems’ control. Application of particle filters and time series neural networks for occupancy models was investigated by Golestan et al. (2018). This study was carried out on two datasets: one containing measurements of occupancyindicative sensors and the other one including the measurements of HVAC sensors. The corresponding results demonstrated that time series neural networks provide the possibility of estimating the number of occupants in each room with an elevated accuracy. In another investigation, conducted by Chi¸tu et al. (2019), data-driven

1636

F. Dadras Javan et al.

techniques based on random forest (RF) and KNN algorithms were developed to estimate the occupancy employing CO2 measurements and airflow. The obtained results demonstrated an occupancy estimation error in the range of 5% to 16% of the corresponding maximum value. Candanedo Ibarra and Feldheim (2015) carried out an investigation on different statistical classification models for occupancy estimation utilizing measurements of light, temperature, humidity, and CO2 . Their obtained results showed that the most promising models with the highest accuracy (ranging from 95% to 99%) are those that utilize linear discriminant analysis (LDA), classification and regression trees (CART), and random forest (RF) models. The authors also demonstrated that the estimation accuracy is notably influenced by the proper selection of features and the employed classification model. Causone et al. (2019) conducted a study dedicated to energetic modeling of residential building, in which a novel datadriven procedure was proposed generating occupant-related input schedules from the electricity recordings of smart meters. The authors demonstrated an increment in the reliability of the building’s energy simulation and the fact that occupantrelated load profiles may account for about 8% of the building’s energy need for space heating. A novel multisensor system (ThermoSense) was developed by Beltran et al. (2013), in which a low-cost sensor array along with a passive infrared sensor was utilized for estimating occupancy. In this system, the ventilation unit was controlled based on the near real-time estimations of occupancy and temperature while employing k-nearest neighbor, linear regression, and artificial neural networkbased models. In another investigation, performed by Ebadat et al. (2013), the relation between the number of occupants in a room and CO2 concentration, room temperature, and ventilation actuation signals was used for developing a dynamic model. Their obtained results showed that the proposed scheme estimated the correct occupancy level 88% of the time and offered a better accuracy than the equivalent neural network (NN) or support vector machine (SVM) estimation strategies. D’Oca and Hong (2015) proposed a data mining framework to determine the occupancy schedule for 16 offices employing a decision tree model with pruned rules. The obtained occupancy rules and schedules could be implemented on building energy modeling programs to identify the impact of the occupant on the energy performance at office buildings. The performance of different ML models for occupancy prediction in terms of accuracy and computational efficiency was investigated by Huchuk et al. (2019). Their obtained results demonstrated that among the simple heuristic and historical average baselines, traditional machine learning classifiers, and sequential models, the random forest algorithm has the most promising performance. Kim et al. (2019b) proposed and implemented a time series occupancy estimation methodology employing the data obtained from various sensors including the indoor temperature and luminance, CO2 density, electricity consumption of lighting, and electric appliances. Their results demonstrated that employing the occupancy estimation framework and implementing occupant-centered HVAC management, the energy consumption performance improves by 17–33%. Furthermore, the

Application of Machine Learning in Occupant and Indoor Environment. . .

1637

authors demonstrated that the selection process of the input variables has a crucial impact on the machine learning-based models’ performance. Yang et al. (2014) evaluated binary and multi-class occupancy estimation by utilizing six different ML algorithms. This study investigated different combinations of sensors to train the model. Results demonstrate that C02, door status, and light are the variables that contribute the most to the results, and accuracy up to 98.2% can be achieved. In this work, the optimized model was then utilized in an occupancy-based demand-response HVAC control procedure, and it was shown that this methodology can result in a saving of to 20% in the gas consumption and 18% in the electricity demand. Hailemariam et al. (2011) carried out an investigation on utilizing different combinations of sensors for a real-time occupancy estimation using decision tree classifier. Their results demonstrated that the occupancy status could be detected with an accuracy of 98.4% by using only infrared motion sensor, while adding other features (CO2, sound, etc.) decreases the accuracy and increases the chance of overfitting. Razavi et al. (2019) implemented an approach for estimating the home occupancy status utilizing the households’ energy consumption data and employing a genetic programming-based feature engineering approach. This study reported that the model performance of occupancy detection increases significantly when the feature engineering method is utilized. Chen et al. (2017) proposed a methodology to automatically learn and extract high-level features from the dataset consisting of environmental sensor’s data, by utilizing convolutional deep bidirectional long short-term memory (CDBLSTM) approach instead of choosing features manually to estimate the occupancy status. A comparison between CDBLSTM result and previous works without feature selection demonstrates a super performance for the proposed approach. Table 1 reports the input parameters and the algorithms that have been employed in the previous studies focused on machine learning-based occupancy estimation and the corresponding achieved accuracy. Regarding the input parameters, it can be concluded that employing only the data obtained from the indoor sensors such as temperature, humidity, CO2 , motion (PIR), and illuminance (or lights) can be sufficient to obtain an elevated accuracy (that has been reported to be up to 99%, depending on the characteristics of the investigated case study). It can also be observed that neural networks (with various configurations), treebased models (e.g., decision trees or random forests), and K-nearest neighbor (KNN) algorithms are among the commonly utilized algorithms, employing which a promising performance has been reported.

3

Literature Review on Machine Learning-Based Window State Estimation

Several research works have also been carried out in the area of window state estimation employing various methodologies. Wei et al. (2019) investigated the utilization of three different algorithms (logistic regression, Markov chain techniques, and

Random forest, KNN Linear discriminant analysis (LDA), classification and regression trees (CART), and random forest (RF) KNN

(1) KNN, (2) linear regression (LR) and artificial neural networks, (3) ANN Support vector machine, neural network

Chi¸tu et al. (2019) Candanedo Ibarra and Feldheim (2015)

Beltran et al. (2013)

Ebadat et al. (2013)

Causone et al. (2019)

Algorithms (1) Decision tree algorithm (2) Hidden Markov model Neural network

Studies Rya and Moon (2016) Golestan et al. (2018)

(1) RMSE of 0.346 (2) RMSE 0.409, (3) RMSE 0.385 Accuracy 82.6% 81.1%

86%

RMSE 0.97% 95% to 99%

Accuracy (1) RMSE 0.2202 (2) 93.2% RMSE 0.3%

Indoor: CO2 , temperature, venting system actuation levels

Indoor: electric energy metering data Outdoor: dry-bulb temperature, relative humidity, wind velocity and direction, precipitation, atmospheric pressure, and global solar radiation Indoor: thermal-based (temperature) sensing and PIR sensors

Input parameters Indoor: energy consumption of the lighting system and appliances, CO2 Outdoor: CO2 Indoor: volatile organic compound concentration, No. of BLE beacons in the range of the receiver, CO2 , VAV Damper position Time-related: calendar with scheduled events, flag indicating a weekday or a weekend Indoor: CO2 , ventilation airflow Indoor: light, temperature, humidity, CO2

Table 1 Summary of the methodologies utilized in the previous studies focused on machine learning-based occupancy status estimation

1638 F. Dadras Javan et al.

Chen et al. (2017)

Hailemariam et al. (2011) Razavi et al. (2019)

Markov-based feedback recurrent neural network Deep learning (convolutional deep bidirectional long short-term memory)

SVM, kNN, ANN, naive Bayesian, tree augmented naive Bayes network, decision tree Decision tree

Yang et al. (2014)

Kim et al. (2019b)

Decision tree and cluster analysis Logistic regression, Markov model, random forest, HMM, recurrent neural network Classification and regression tree (CART)

D’Oca and Hong (2015) Huchuk et al. (2019)

87.78%–95.42%

80.9%–93.9%

81%–98%

69.2%–92.6%

Indoor: CO2 , temperature, relative humidity, pressure

Indoor: CO2 , computer electrical current, illuminance, PIR (motion), and sound pressure level Indoor: energy consumption data

Indoor: Tdry−bulb , φindoor , CO2 , illuminance, electricity consumption data for LED lights, desktop (PC), and electric heat pump (EHP) Indoor: CO2 , Tindoor , humidity, light, sound, motion, passive infrared

Indoor: thermostat data

75–80% (median)

95%

Indoor: occupancy data

90.53%

Application of Machine Learning in Occupant and Indoor Environment. . . 1639

1640

F. Dadras Javan et al.

artificial neural network (ANN)) for modeling the occupant’s interaction with the windows to estimate the window state. The authors’ obtained results demonstrated that the ANN model could estimate the window state more precisely in comparison to other methods. Furthermore, the indoor temperature, outdoor temperature, wind speed, and sunshine hours were demonstrated in this study to be the most influential factors that should be considered to implement an accurate model. Pan et al. (2019) proposed a novel Gaussian distribution-based model to estimate the window state by utilizing indoor temperature, outdoor temperature, and their combination. The results of this study demonstrated that utilizing this novel model results in an increment of 9.5% in the achieved accuracy compared to the one that can be obtained employing the logistic regression algorithm. Shi et al. (2018) investigated how climate and air quality parameters can affect the window opening/closing state. Window opening behavior was estimated for different climatic conditions (cold, mild, and hot season) employing logistic regression algorithm. The obtained results demonstrated that those effects of each parameter could vary as a function of the climatic conditions. In addition, the indoor temperature or relative humidity was shown to be the dominant factor for window opening. Another study was conducted by Markovic et al. (2019) aiming at identifying the influence of using short-term lagged indoor environment’s data in a deep neural network-based window state predictor. The obtained results demonstrated that the optimal performance can be reached by using 60 time steps (each time step corresponding to 1 min) of the lagged data and adding more data does not improve the achieved estimation accuracy. It was also shown that, as can be expected, the prediction accuracy decreases by increasing the forecasting horizon. Another investigation (Markovic et al. 2017) was dedicated to assessing the performance of two different classification algorithms, support vector machine and random forest, which combined with dynamic Bayesian network were analyzed for occupant’s interaction with window detecting model. In this research, in which the receiver operating characteristic (ROC) was considered as the evaluation metric (in order to take the imbalance in the available data into account), the random forest algorithm was demonstrated to result in the highest accuracy. Yao and Zhao (2017b) implemented a window status estimation pipeline utilizing the environmental parameters and time-related features while employing the logistic regression algorithm. The results of this study demonstrated that the outdoor air temperature is the most crucial parameter that contributes to the accuracy of the model. Other important parameters were instead identified to be the indoor CO2 concentration, temperature, and humidity along with the outdoor temperature, relative humidity, wind speed, and direction. In another study Li et al. (2015), a multifactor analysis on the window state variation was performed, which showed that the outdoor air temperature is the most important factor that influences the model’s performance. Furthermore, a Monte Carlo simulation that was carried out in this study indicated that an increment in the outdoor air temperature increases the probability of window opening during the transition season. Stazi et al. (2017) investigated the influence of environmental factors and daily routine on the interaction between the occupant and window in

Application of Machine Learning in Occupant and Indoor Environment. . .

1641

an Italian classroom. The results obtained utilizing linear and logistic regression models demonstrated that the indoor and the outdoor temperatures are the two main factors that have an impact on the performance of the model. Furthermore, it was shown that the daily routine should be considered as an input feature in the model as it can help in increasing the achieved accuracy (e.g., the window opening frequency is higher during break time in the classroom). Table 2 summarizes the details of the methodologies that have been employed in the studies focused on machine learning-based estimation of the window state and the corresponding achieved accuracy. It can be observed that the logistic regression is the most commonly utilized algorithm in these studies, while support vector machines (SVM), neural networks, Markov processes, and random forests have also been utilized. It can also be inferred that both indoor and outdoor parameters play a significant role in the corresponding implemented estimation pipeline and indoor/outdoor temperature has specifically been utilized in most of these studies. It can finally be concluded that employing these methodologies with an elevated accuracy of up to 98% (depending on the considered case) can be achieved.

4

Literature Review on Machine Learning-Based Window Opening/Closing Action Estimation

Considering the same objective as that of Sect. 3, an alternative methodology that has been utilized in the literature is the estimation of the window opening and closing action instead of estimating the window state. The employed methodologies in the studies, in which this approach has been utilized, are discussed in the present section. Barthelmes et al. (2017) investigated the potential of employing Bayesian network model for identifying the key features for window opening/closing behavior of occupants. This study proposed a feature selection method to choose the most influential drivers of window control behavior. Furthermore, the authors’ results proved that choosing the window opening action (instead of the window state) as the model’s target, results in an improvement in the performance of the developed model. Singh et al. (2017) investigated the window opening and closing action employing the variations of indoor temperature in naturally ventilated office concerning adaptive thermal comfort. The obtained results proposed the characteristic of opening event of windows at different temperatures employing linear and nonlinear regression analyses. The relation between the probabilities of window opening action with different outdoor variables was investigated for eight residential houses by Shi and Zhao (2016). The result demonstrated that the outdoor air temperature is the most critical driver of the window opening action among other features (relative humidity, outdoor wind speed, and air pollution) and the dependence of the window opening probability on the outdoor variable varies based on the location and season. Naspi et al. (2018) developed a model to predict the human interaction with the building and specifically the interaction with the window in an office in the Mediterranean climate during the summer. Results showed that the window opening

(1) Logistic regression, (2) Gauss distribution model Logistic regression

ANN, (1) Bayesian networks + SVM, (2) Bayesian networks + random forest Logistic regression

Pan et al. (2019)

Markovic et al. (2019)

Logistic regression

Logistic regression

(1) SVM and (2) random forest

Yao and Zhao (2017b) Li et al. (2015)

Stazi et al. (2017)

Markovic et al. (Li et al. 2015)

Shi et al. (2018)

Algorithms (1) Logistic regression, (2) Markov processes, (3) ANN

Studies Wei et al. (2019)

Neglekerke R2 0.010–0.192 (1) 83% and (2) 89%

71.4% (closed) and 72.9% (opened) 70.15%

88% (Avg) (1) 83% and (2) 88%

Accuracy (1) 52–53.3%, (2) 57–57.9%, (3) 73.2–79.5% (1) 61.3–64.6%, (2) 58.3–74.1% 76.7–98.0%

Indoor: CO2 , φindoor , temperature Outdoor: temperature, wind velocity, wind direction, relative humidity Indoor: CO2 , φindoor , temperature Outdoor: temperature, wind velocity, relative humidity Indoor: temperature, occupancy state Outdoor: temperature, wind velocity, wind direction, relative humidity, rainfall CO2 , Indoor: temperature, relative humidity, occupancy state Outdoor: temperature, precipitation, wind velocity, wind direction, position sun protection

Indoor: CO2 , φindoor , temperature, relative humidity Outdoor: wind speed, wind direction, rainfall, PM2.5 concentration, temperature, relative humidity Indoor:Tindoor , humidity, CO2

Indoor: temperature Outdoor: temperature

Input parameters Indoor: temperature Outdoor: temperature, wind speed, and sunshine hours, PM2.5 concentration, humidity

Table 2 Summary of the input parameters and algorithms that have been employed in the studies dedicated to machine learning-based window status estimation along with the corresponding achieved accuracy

1642 F. Dadras Javan et al.

Application of Machine Learning in Occupant and Indoor Environment. . .

1643

and closing action is influenced by the environmental and factors, time-related features and the geographical area. Coefficient of interactivity, which measures the frequency of interactions considering the corresponding occupancy level, was defined as an index in this study to determine the level of human interaction with windows. Yun and Steemers (2008) studied the window opening control by monitoring indoor variables in an office with and without the nightly ventilation during the summer. This investigation demonstrated that the indoor temperature and the previous window state are statistically significant features for predictive modeling of the window opening behavior. An automatic ventilation control algorithm to open/close windows in order to achieve the optimal indoor air quality was developed by Kim et al. (2019a). In this work, the logistic regression algorithm was utilized in the control strategy, and the indoor air quality factors were utilized as input features. Through performing a comparison with indoor environment standards, it was demonstrated that utilizing this window opening/closing control algorithm a compliance value of 99% can be achieved. A stochastic window status profile generator for desired climatic region (WinProGen) was introduced by Calì et al. (2018). In this study, tree models were developed to generate the profile of window status employing Markov chain technique. The first model considered the time of the day and the daily average ambient temperature as features. The second model instead considered time of the day, the daily average ambient temperature, and the day of the week (working day or weekend day). Finally, the last model took into account the time of the day, the daily average ambient temperature of the current day, and the daily average ambient temperature of the previous day for generating the window state pattern. The obtained results demonstrated that employing the last model results in the highest performance that can be achieved. Yao and Zhao (2017a) performed a study focused on investigating the relation between the environmental factors and the window opening/closing action and proposed a stochastic model of occupant’s window opening behavior utilizing multivariant linear logistic regression model. In this model, the indoor and outdoor parameters (temperature and relative humidity of indoor and outdoor air, indoor CO2 concentration, outdoor PM2.5 concentration, and outdoor wind speed) were employed as input features. The obtained results demonstrated that the outdoor temperature is the most and the indoor temperature the least important factors. Table 3 summarizes the employed machine learning algorithms and input parameters that have been utilized in the previous studies for window opening/closing event estimation. It can be concluded that the logistic regression is the most commonly utilized algorithm in these studies. Many of these studies also showed that the outdoor temperature is the most important driver of the window opening or closing action.

Nagelkerke’s R2 0.337 Sensitivity 76.30 specificity 86.80 Youden’s index 0.631, AUC 0.865

Logistic regression

Logistic regression

Markov chain

Logistic regression

Logistic regression

Naspi et al. (2018)

Yun and Steemers (2008)

Calì et al. (2018)

Yao and Zhao (2017a) Kim et al. (2019a)

AUC 0.54–0.93 McFadden’s R2 0.004–0.55 Neglekerke R2 0.01–0.69 Loglikelihood 87.404–3.074 G-statistic 2.438–26.341 70.15%

R2 0.54 and 0.58 76.7–98.0%

Linear and nonlinear regression Logistic regression

Singh et al. (2017) Shi and Zhao (2016)

Accuracy 93%

Algorithms Bayesian network

Studies Barthelmes et al. (2017)

Indoor: temperature and lagged values, time related, day of the week, time of the day Indoor: temperature, relative humidity, CO2 concentration Outdoor: air temperature, relative humidity, wind speed PM2.5 concentration Indoor: air temperature, humidity, Co2 concentration, PM10 concentration, PM2.5 concentration, TVOC concentration,

Indoor: temperature, previous state if window; time related, time of the day

Input parameters Indoor: dry-bulb temperature, relative humidity, illuminance, CO2 concentration Outdoor: air temperature, relative humidity, wind speed, global solar radiation, and day of the week and time of the day Indoor: temperature Indoor: CO2 , φindoor , temperature, relative humidity Outdoor: wind speed, wind direction, rainfall, PM2.5 concentration, temperature, relative humidity Indoor: temperature, Co2 Outdoor: temperature

Table 3 Summary of the methodologies employed in the studies focused on machine learning-based estimation of the window opening/closing action

1644 F. Dadras Javan et al.

Application of Machine Learning in Occupant and Indoor Environment. . .

5

1645

Literature Review on Machine Learning-Based Indoor Temperature Prediction

Considering the determinant role of the indoor temperature on the occupant comfort and the building energy consumption, accurate prediction of this parameter is of notable importance for implementing the smart building systems and real-time optimization of the operation of HVAC units (Holz et al. 1997). Accordingly, several studies have been dedicated to proposing and implementing methodologies for machine learning-based prediction of indoor temperature. These investigations have utilized various data collection methods, input features, and algorithms. Therefore, aiming at identifying the methods and algorithms utilizing which results in the highest possible accuracy, the details of these methodologies have been discussed in the present section, and the corresponding achieved accuracy has been reported. Gouda et al. (2002) deployed a multilayer feed-forward neural network (FF ANN) with Levenberg-Marquardt backpropagation training algorithm to predict a solar house’s indoor temperature. The objective of this study was facilitating the early shut down of the heating system for energy-saving purposes. Singular value decomposition (SVD) technique was employed to decrease the number of effective inputs. The authors included outdoor temperature, solar irradiance (G), heating valve position, and indoor building temperature as input parameters with the sampling rate (SR) of 15 min. The predictions were reasonable for 2 h ahead prediction; though, it was shown that the error increases notably for longer prediction horizons. A significant contributor to the error was found to be the cloudy winter sky affecting the irradiance. Aiming at predicting the indoor temperature, Ruano et al. (2006) compared the multimode physical model’s performance with the radial basis function neural network (RBF ANN)Radial basis function neural network( RBF NN) model using the data collected from a residential building. The indoor temperature, solar radiation, outdoor temperature, and relative humidity (φ) were used as the model inputs. The authors demonstrated that the neural network model outperforms the physical model (root mean square error (RMSE) of 0.0493 against 0.1777 achieved by the physical model). Correspondingly, Özbalta et al. (2012) explored the performance of ANN models in predicting the daily mean indoor temperature and relative humidity in an educational building while employing the day (of the year), the outdoor temperature, the outdoor relative humidity, and the wind speed(Vwind ) as input parameters. The obtained results were then compared with that of the multiple regression models. It was shown that the performance of the ANN model exceeded that of the linear regression (LR) models and the resulting (R 2 ) scores were reported to be 0.94 and 0.96 for indoor temperature and relative humidity prediction, respectively. In a study dedicated to the prediction of the indoor temperature of a building based on the outdoor conditions, Ashtiani et al. (2014) compared the performance of the regression and the ANN-based algorithms. In this investigation, the outdoor dry-bulb temperature, the direct normal solar radiation, the wind velocity, and the outdoor relative humidity data were employed as input features. The RMSE between

1646

F. Dadras Javan et al.

the predicted and actual value of indoor dry-bulb temperatures was employed to assess the model’s performances. The obtained results indicated that employing the ANNs instead of linear regression models results in an improvement in the achieved accuracy, although it also leads to an increment in the pipeline’s complexity. In an investigation carried out by Mba et al. (2016), which was focused on predicting the indoor temperature and relative humidity in a building located at hot-humid climatic region, a grid search was performed aiming at determining the optimal number of input variables, neurons, and activation functions in the utilized neural network. In this study, the last 12 hourly recorded temperature values (indoor and outdoor) and the relative humidity were used as the input features to predict the indoor parameters 24 h in advance. The optimal ANN model’s predicted values for temperature and humidity were shown to have a very high correlation with the experimental data (the coefficient of correlation being (R) 0.9850 and 0.9853 for temperature and humidity, respectively). In the research conducted by Mechaqrane and Zouak (2004), the performances of the ANN and linear regression models for predicting the indoor temperature were compared, and it was demonstrated that, owing to their capability of capturing the building’s nonlinear behavior, the ANNs outperform the linear models with a large margin. Soleimani-Mohseni et al. (2006) implemented different methodologies (utilizing the experimental data obtained from two different buildings) aimed at estimating the indoor operative temperature while utilizing the estimated easy-to-measure variables, i.e., the indoor and outdoor temperatures, the electrical power use in the room, the wall temperatures, the ventilation flow rates, and the time of the day. The results of this study in terms of mean absolute estimation error (MAEE) demonstrated that the ANN models achieve moderately better results with respect to the linear autoregressive model with an exogenous input (ARX) models. Similarly, Thomas and Soleimani-Mohseni (2007) drew a distinction between the two buildings with regard to the best combination of input parameters. Conclusions of the latter work indicated that indoor temperature was strongly dependent upon the nonlinear combination of solar radiation and time of the day, which further confirmed the artificial neural network’s advantage over the linear models due to its inherent nonlinear nature. Lu and Viljanen (2009) also assessed the capability of neural networks to perform prediction on the indoor temperature and relative humidity in a test house. In this work, a neural network nonlinear autoregressive model with external input (NNARX) was accompanied by a genetic algorithm (GA), seeking the neural network’s optimal structure. The features employed for training were time (time interval of 15 min), temperature (outdoor and indoor), and relative humidity (outdoor and indoor), excluding any of the HVAC system’s inputs. The temperature forecast was fairly precise for both models. Correlation coefficients between predicted values and measured data were determined to be 0.998 and 0.997 for NNARX and genetic models, respectively. The performances of indoor temperature prediction pipelines were observed to be better than those developed for forecasting the relative humidity which is attributed to the nonlinear diffusion equation governing the relative humidity (φ). Furthermore, it was observed that the

Application of Machine Learning in Occupant and Indoor Environment. . .

1647

genetic algorithm can adequately be utilized as a search method for selecting the structure of the developed model. In an investigation conducted by Mustafaraj et al. (2010), on a case study of an open office, the collected data corresponding to the indoor and outdoor temperature, relative humidity, supply air temperature/flow rate, chilled water temperature, and hot water temperature for three consecutive seasons were utilized as input features. Furthermore, the information on the time of the day, the time of the year, and the type of the day (working or non-working day) were also included as input parameters. The authors utilized this data to train a neural network nonlinear autoregressive model with external inputs (NNARX), a neural network nonlinear autoregressive moving average model with external inputs (NNARMAX), and a neural network nonlinear output error (NNOE) model aiming at estimating the indoor temperature and the relative humidity of an office building. In their implemented methodology, once the algorithm’s training had come to an end, the fully connected neural network would go through a pruning procedure (called optimal brain surgeon (OBS)) to avoid over-fitting. Results were reported for 6, 12, and 24 steps ahead predictions (corresponding to 30 min, 1, and 2 h ahead forecasting) for different seasons. The authors demonstrated that NNARX had moderately higher performance than the other algorithm for indoor temperature prediction (resulting in higher goodness of fit, coefficient of determination, mean absolute error, and mean squared error). In another study, Mustafaraj et al. (2011) demonstrated that having included the concentration of CO2(Cco2 ) as an input parameter to the model, an increase in the temperature prediction accuracy is observed that can be attributed to the correlation between the occupancy status and the carbon dioxide concentration level. Marvuglia et al. (2014) employed an autoregressive neural network nonlinear autoregressive model with external inputs (NNARX) to forecast the indoor temperature, which was then utilized to feed a fuzzy logic controller. The controller’s objective was to adjust the inlet air speed and switch the HVAC system on and off. The novelty of this study was the optimal adoption of the NNARX model parameters. The NNARX model included outdoor parameters (i.e., outdoor temperature, air relative humidity, and wind speed) and indoor temperature data for winter and summer scenarios. In a similar context, aiming at developing a controller for guaranteeing the thermal comfort in a residential building, Moon and Kim (2010) presented a thermal control logic consisting of four components: temperature and humidity control with and without ANN and PMV control with and without ANN. In this work, the measured parameters comprised outdoor and indoor temperature and humidity, indoor MRT, and indoor air velocity. Three identical ANN models were developed, predicting the variations of temperature, humidity, and PMV. It was concluded that the predictive control logic employing ANN models essentially decreases the magnitudes of overshoots and undershoots and a reduction in energy consumption, but not in all cases. In a similar attempt, Moon et al. (2013) applied the ANN-based control logic in a double-skin envelope building in the winter. The controller was designed to command the heating system’s operation and opening of the double-skin building’s envelopes through predicting the indoor temperature values. The ANN model was fed with the indoor and outdoor air temperature, cavity

1648

F. Dadras Javan et al.

temperature, solar radiation, and inner and outer surface’s opening condition data, which was aggregated from an actual building located in Ansan, Korea. Results exhibited promising accuracy, though the test was confined to 2 days. Moon (2015) also applied the same conventional- and ANN-based control methods for a doubleskin facade building and demonstrated that using this type of control logic, the indoor temperature stably can be maintained within the comfort range. Aiming at implementing nonlinear autoregressive neural network models for indoor temperature prediction in a secondary school, Aliberti et al. (2019) presented a novel methodology based on (IoT) technology. The building under investigation was deploying 13 IoT devices with a sampling rate of 15 min, which was not ample enough to train the model. Hence, the weather data collected over 6 years was employed alongside the simulations of the thermal behavior of the building performed with EnergyPlus. This generated a dataset almost ten times larger than the actual data, which was then used to train the model. Aiming to extend the prediction horizon and the accuracy of predictions, the ANN model was initially tested on an artificial dataset obtained from simulations; subsequently, the model was tested on a smaller set of real data obtained from the IoT devices. The trained models provided accurate predictions up to 3 h ahead for the individual rooms and 4 h ahead for the entire building. RSMD values of 0.43 and 1.65 were obtained for 15 and 270 min ahead prediction pipelines, respectively. Attoue et al. (2018) proposed a method for indoor temperature forecasting in a smart building while introducing an innovative input feature selection. The developed ANN-based model was deployed on the data extracted from an old building followed by an innovative relevance study to select the input parameters among a more extensive set. The temperature prediction was accomplished in two steps: facade temperature prediction was first forecasted utilizing the indoor and outdoor parameters. The corresponding predicted value was then utilized to predict the temperature at the room’s center. The first step’s model was fed with outdoor temperature, outdoor humidity, solar radiation, outdoor temperature history, time, and facade temperature history, and the obtained predictions demonstrated an elevated accuracy (R = 0.9967 and MSE = 0.0277). Subsequently, IBM SPSS statistical software was utilized to inspect the parameters in terms of their importance by performing a sensitivity analysis for each input parameter. The outdoor temperature and historical facade temperature were recognized to have the highest priority, followed by historical outdoor temperature. Considering only the first two factors as input parameters, the authors could obtain promising results for predictions up to 2 h ahead. Concerning the room center prediction employing the facade temperature, an acceptable performance was observed utilizing an ANN model. In a novel approach, Yu et al. (2018) performed indoor temperature predictions while feeding the models with smart thermostat and weather forecast data (including detailed data of the HVAC system and outdoor weather data). Thermostat data from 16 Canadian and US houses were utilized to develop two generalized regression neural network (GRNN) algorithms and resilient backpropagation neural network (ANN) algorithms. Results were reported in terms of MSE and temperature deviation, which affirmed the GRNN model’s superiority (MSE of 0.03 to 0.82).

Application of Machine Learning in Occupant and Indoor Environment. . .

1649

An issue that is commonly faced while utilizing sensors to record indoor temperature is the data collection failures, leading to gaps in data that may not be filled simply by linear interpolations. Machine learning algorithms were applied to forecast the missing temperature data in a study conducted by Candanedo et al. (2018), with an emphasis on the optimal size of the dataset to train the model. Learning curves were utilized to find the optimal size of the required dataset, and multiple linear regression and random forest algorithms were deployed for the sake of comparing the models’ performance. Electric energy use was reported in 10-min intervals using M-BUS energy counters, while interior temperature and humidity were reported through a wireless sensor network. The weather data at hourly intervals, including the outdoor temperature, pressure, relative humidity, wind speed, and visibility, was reported from the airport’s weather station in the vicinity. Plots of the root mean squared error (RMSE) vs. sample size (with no missing data) illustrated that the regression models’ effectiveness does not brush up significantly after certain sample sizes have been reached. Furthermore, the coefficient of determination, the mean absolute error (MAE), and the mean absolute percentage error (MAPE) were deployed as accuracy metrics. It was shown that in terms of calculation cost, the regression models outperform random forest remarkably as the training time for the LM is a linear function of the sample size while it is a quadratic function for RF. Furthermore, a bigger sample size is required for training the RF models. Nonetheless, the RF model dominated the LM in terms of precision and was proved to be a promising tool for indoor temperature prediction. In pursuit of finding the most promising algorithm to predict the indoor temperature in a smart building, Alawadi et al. (2020) performed a large set of experiments to compare the accuracy of 36 different machine learning algorithms belonging to 20 different families, employing the R-score and root mean squared Error(RMSE) as the metrics. Measurements from the sensors and weather data collected from the weather station constitute the dataset that was received every 10 min. Each algorithm’s experiment was repeated ten times, deploying numerous randomly generated seeds to avoid unintended biases. It was demonstrated that the extra tree algorithm provides the best results among all algorithms (R-score of 97% and RMSE of 0,058%). Other promising models were identified to be the random forest, Cubist, Gradient boosting with regression base trees (BstTree), average neural network committee (avNNet), and kernel ELM (elm-kernel). Table 4 provides a summary of the input features and algorithm that were utilized in the abovementioned studies focused on indoor temperature prediction and reports the corresponding achieved accuracy. The extensive use of the neural network model in building indoor temperature prediction has been observed throughout recent years. Even so, more broad research is also needed directing toward other models with higher precision, e.g., random forest and extra tree algorithms. It is also worth mentioning that implementing suitable feature selection algorithms helps in reducing the complexity of the models, decreasing the number of required measurements while even increasing the achieved accuracy. Furthermore, utilizing limited number of influential features facilitates the physical interpretation of the

Algorithm FF ANN

Lu and Viljanen (2009) Mustafaraj et al. (2010)

Mba et al. (2016) Mechaqrane and Zouak (2004) Soleimani-Mohseni et al. (2006)

NNARX, NNARMAX NNOE

GA, NNARX

ARX, FF ANN

ANN ARX, NNARX

Ashtiani et al. (2014) FF ANN, LR

RBF ANN, multimode physical model Özbalta et al. (2012) LR, FF ANN

Ruano et al. (2006)

Ref Gouda et al. (2002) RMSE = 0.0493 (ANN), 0.1777 (physical) (LR): RMSE = 0.99, 0.56 (FF ANN) RMSE (ANN) = 1.76 ◦ C RMSE (LR) = 2.10 ◦ C R = 0.9850 SSE = 2.06 (NNARX), 0.906 (pruned NNARX), 15.04 (ARX) building (1) MAE (test) = 0.177 (ANN), 0.194 (ARX) building, (2) MAE (test) = 0.1 (ANN),0.1 (ARX) [two steps ahead] MAE = 0.2568 (NNARX) 0.4262 (GA) [six steps ahead] MAE = 0.1039 (NNARX), 0.1378 (NNARMAX), 0.1778 (NNOE)

Accuracy metrics (best case) ±0.5K (1 h)

Table 4 Publications of data-driven indoor temperature prediction models

Indoor: Tindoor ,φindoor , TsupplyAir , φsupplyAir m·supplyAir , TchilledWater , ThotWater Outdoor: Toutdoor ,φoutdoor date, type of the day (SR: 5 min)

Indoor: Tindoor ,φindoor Outdoor: Toutdoor , φoutdoor (SR: 15 min)

Indoor: Tindoor ,φindoor Outdoor: Toutdoor ,φoutdoor (SR: hourly) Indoor: auxiliary heating power, Tindoor Outdoor: Toutdoor ,Girr (SR: hourly) Indoor: Tindoor ,EProom , Twall , m·ventilation Outdoor: Toutdoor , T ime,Girr (SR: 10 min)

Outdoor: Toutdoor , Girr , Vwind , φoutdoor , T ime (SR: hourly)

Input parameters Indoor: heating valve position, Tindoor Outdoor: Toutdoor , solar irradiance (SR: 15 min) Indoor: Tindoor Outdoor: Girr , Toutdoor ,φoutdoor (SR: 5 min (averaged over 1 min)) Indoor: Tindoor Outdoor: Date,Toutdoor , φoutdoor , Vwind (SR: daily)

1650 F. Dadras Javan et al.

NNARX

NAR

ANN

GRNN, BP ANN

LR,RF

Marvuglia et al. (2014) Aliberti et al. (2019)

Attoue et al. (2018)

Yu et al. (2018)

Candanedo et al. (2018) Alawadi et al. (2020)

36 different algorithms

ARX, NNARX

Mustafaraj et al. (2011)

[six steps ahead] MAE = 0.0803 (NNARX) 0.092 (ARX) MAE = 0.250, RMSE = 0.466 [270 min ahead] MAD = 1.01, RMSD = 1.65 [30 min ahead] MSE = 0.0701 MSE = 5.5 (ANN), 0.98 (GRNN) RMSE = 0.438 (LR), 0.35 (RF) [1 h] RMSE = 0.04041 (extraTrees) Indoor: Tindoor Outdoor: Toutdoor , φoutdoor ,Girr , Thistory ,Tfacade (SR: 5 min) Indoor: smart thermostat, HVAC data, Tindoor , φindoor Outdoor: weather data Indoor: Electric energy use, Tindoor , φindoor , Outdoor: Pressure Toutdoor , H eatingP ower, φoutdoor , Vwind , visibility, Tduepoint (SR: 10 min) Indoor: Tunderfloorheating , underfloor heating status, air condition status, TAirConditioning , φAirConditioning , Tindoor , Thistory Outdoor: φourdoor , Toutdoor ,Girr (SR: 10 min)

Indoor: Tindoor ,φindoor Outdoor: Girr ,Toutdoor , φoutdoor (SR: 15 min)

Indoor: Tindoor ,Φindoor , TsupplyAir , φsupplyAir , m·supplyAir , TchilledWater ThotWater Outdoor: Toutdoor Φoutdoor ,date, type of the day, Cco2 (SR: 5 min) Indoor: Tindoor Outdoor: Toutdoor φindoor ,Vwind (SR: hourly)

Application of Machine Learning in Occupant and Indoor Environment. . . 1651

1652

F. Dadras Javan et al.

obtained results. Finally, employing automatic tuning tools that have been developed in the recent years for optimizing the hyper-parameters of machine learning models helps identifying the optimal pipeline and guaranteeing that the highest possible accuracy has been achieved.

6

Literature Review on Machine Learning-Based Occupancy Status Prediction and Implementation of Occupant-Centered HVAC Control

Employing various methodologies for occupancy estimation (that was discussed in details in the second section), several studies have been dedicated to the prediction of the occupancy status and implementing the occupant-centered/occupancy-driven HVAC control (Park et al. 2019). Erickson et al. (2011) implemented an occupancydriven HVAC management system for an office building, in which the occupancy profiles were detected using cameras installed in the offices (detected through changes in the background) and were then predicted using Markov chain model. It was demonstrated that utilizing this methodology results in an annual energy saving of 42%. In another study, Dobbs and Hencey (2014) utilized the occupancy profiles of an office building that were detected using PIR sensor and developed a Markov model for predicting the occupancy status. The authors then employed software simulation to determine the possibility of implementing an occupancybased model predictive HVAC control strategy and demonstrated that it can result in an annual energy saving of 37%. Dong and Lam (2011) utilized a set of sensors (CO2, motion, and light, acoustic, temperature, and humidity) installed in an office building and employed Gaussian mixture models (GMMs) and hidden Markov model (HMM) prediction algorithms and showed that the overall consumption of the building is reduced by 18.5%. Dong and Andrews (2009) fed the data obtained from the same sensors installed in an office building to semi-Markov model to estimate and predict the occupancy status and (through simulation) demonstrated that that occupancy-driven dynamic management of set points can reduce the overall energy consumption of the building by 30.3%. Erickson and Cerpa (2010) employed a set of standard occupancy profiles and demonstrated that through utilizing Markov chain models for occupancy prediction (trained by the latter profiles) and implementing the occupancy-driven control, an overall energy saving of 20% for an office building can be achieved. Wang et al. (2017) utilized iBeacon-enabled indoor positioning system (IPS) for detecting the occupancy in an office building in Hong Kong and, employing simulations, showed that using occupancy-driven HVAC management system results in energy saving of 20%. Wang et al. (2019) employed a Wi-Fi probe-based system installed in an office building in Hong Kong to detect the occupancy status utilizing the strength of occupants’ devices/tags. The occupancy was next predicted using an ensemble of tree-based machine learning algorithms, and it was demonstrated that implementing an occupancy-driven HVAC control results in an overall energy saving of 26.4%.

Application of Machine Learning in Occupant and Indoor Environment. . .

1653

Peng et al. (2018) implemented KNN machine learning algorithm to learn the occupancy patterns (detected using motion sensors) in an office building in Singapore and demonstrated that employing occupancy-driven HVAC control can lead to energy saving of 7–52% for different rooms (11 rooms were controlled) and an average of 21%. The review conducted by the authors had also demonstrated an overall expected saving of 8% to 28%. Finally, in a recent study performed on office buildings in 40 different American cities (located in diverse climatic zones), Pang et al. (2020) simulated the impact of employing occupant-centric building HVAC controls. The considered scenarios in this study included different setback values (4 ◦ C and 8 ◦ C) in unoccupied intervals while employing smart recovery (guaranteeing that the temperature is at comfort level when the occupant comes back to the space). Considering comfort set point temperatures of 21 ◦ C for heating and 24 ◦ C for cooling, for the scenario with setback of 4 ◦ C, the heating and cooling set points in unoccupied intervals were set to be 17 ◦ C and 28 ◦ C, respectively. Similarly, for the scenario with the setback of 8 ◦ C, during the unoccupied intervals, heating set points of 13 ◦ C for the heating and 32 ◦ C for the cooling were imposed. The results of simulations showed that, taking into account all of the considered 40 American cities, an average saving ratio of about 10% with 4 ◦ C setback and around 14% with 8 ◦ C setback can be achieved. However, it was shown that the achievable energy-saving margin is notably higher in warm/hot climatic zones, as employing the 4 ◦ C setback strategy in these area results in an energy saving of over 20%, and for four cities located in this climatic zone, utilizing 8 ◦ C setback strategy can even result in an energy saving ratio of over 30%. In the area of residential buildings, Lu et al. (2010) utilized the motion and door sensor data obtained from eight residential building to estimate the occupancy status and (through simulation for different American cities) showed that (on average) it leads to decrease in consumption by 28%. Scott et al. (2011) installed a motion sensor-based occupancy detection system in five residential buildings (in Seattle and Cambridge) and implemented a heating control system based on predicted occupancy. The authors showed that employing the latter system decreased the overall heating consumption of the UK buildings between 8% and 18%. A field study conducted by Pritoni et al. (2016) demonstrated that employing occupancyresponsive thermostats in a university’s residential hall can reduce the heating and cooling by 0–10% during the periods of regular use and decrease the cooling consumption by 20–30% during months with very low occupancy (e.g., summer). Dong and Lam (2014) utilized the data obtained from sensors CO2, motion, and light, acoustic, temperature, and humidity installed in a solar house to estimate and predict the occupancy status (employing GMMs and HMM) and demonstrated that model predictive control of the HVAC system, employing the predicted occupancy status and weather forecast, results in an overall yearly energy saving of 30.1% for the heating seasons and 17.8% for the cooling season. Through analyzing these studies, it can be concluded that hidden Markov model followed by Gaussian mixture model and KNN method is among the most commonly utilized techniques for occupancy prediction aiming at implementing

1654

F. Dadras Javan et al.

occupant-centered HVAC control, which have resulted in achieving a promising accuracy. Furthermore, it can be observed that a saving margin of 10% to 30% (depending on the building’s occupancy profiles) can be achieved through implementing the occupant-centered HVAC control systems.

References S. Alawadi, D. Mera, M. Fernández-Delgado, F. Alkhabbas, C.M. Olsson, P. Davidsson, A comparison of machine learning algorithms for forecasting indoor temperature in smart buildings. Energy Syst. 15, 1–17 (2020) A. Aliberti, L. Bottaccioli, E. Macii, S. Di Cataldo, A. Acquaviva, E. Patti, A non-linear autoregressive model for indoor air-temperature predictions in smart buildings. Electronics 8(9), 979 (2019) A. Ashtiani, P.A. Mirzaei, F. Haghighat, Indoor thermal condition in urban heat island: comparison of the artificial neural network and regression methods prediction. Energy Build. 76, 597–604 (2014) N. Attoue, I. Shahrour, R. Younes, Smart building: use of the artificial neural network approach for indoor temperature forecasting. Energies 11(2), 395 (2018) V.M. Barthelmes, Y. Heo, V. Fabi, S.P. Corgnati, Exploration of the Bayesian network framework for modelling window control behaviour. Build. Environ. 126, 318–330 (2017) A. Beltran, V.L. Erickson, A.E. Cerpa, Thermosense: occupancy thermal based sensing for HVAC control. In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy- Efficient Buildings. (2013), pp. 1–8 V. Bonneau, T. Ramahandry, IDATE, B. Pedersen, L. Dakkak-Arnoux, L. Probst, Smart building: energy efficiency application. Digital Transformation Monitor (2017) D. Calì, M.T. Wesseling, D. Müller, WinProGen: a Markov-chain-based stochastic window status profile generator for the simulation of realistic energy performance in buildings. Build. Environ. 136, 240–258 (2018) L.M. Candanedo, V. Feldheim, D. Deramaix, Reconstruction of the indoor temperature dataset of a house using data driven models for performance evaluation. Build. Environ. 138, 250–261 (2018) L. Candanedo Ibarra, V. Feldheim, Accurate occupancy detection of an office room from light, temperature, humidity and CO2 measurements using statistical learning models. Energy Build. 112, 28–39 (2015) F. Causone, S. Carlucci, M. Ferrando, A. Marchenko, S. Erba, A data-driven procedure to model occupancy and occupant-related electric load profiles in residential buildings for energy simulation. Energy Build. 202, 109342 (2019) Z. Chen, R. Zhao, Q. Zhu, M.K. Masood, Y.C. Soh, K. Mao, Building occupancy estimation with environmental sensors via CDBLSTM IEEE Transactions on Industrial Electronics. 64(12), 9549–9559 (2017) C. Chi¸tu, G. Stamatescu, A. Cerpa, Building occupancy estimation using supervised learning techniques. In 2019 23rd International Conference on System Theory, Control and Computing (ICSTCC). (IEEE, 2019), pp. 167–172 W. Commerell, G. Mengedoht, M. Narayanan, Importance of buildings and their influence in control system: a simulation case study with different building standards from Germany. Int. J. Energy Environ. Eng. 9, 413–433 (2018) J.R. Dobbs, B.M. Hencey, Predictive hvac control using a Markov occupancy model, in 2014 American Control Conference (IEEE, 2014), pp. 1057–1062 S. D’Oca, T. Hong, Occupancy schedules learning process through a data mining framework. Energy Build. 88, 395–408 (2015)

Application of Machine Learning in Occupant and Indoor Environment. . .

1655

B. Dong, B. Andrews, Sensor-based occupancy behavioral pattern recognition for energy and comfort management in intelligent buildings, in Proceedings of Building Simulation (2009), pp. 1444–1451 B. Dong, K.P. Lam, Building energy and comfort management through occupant behaviour pattern detection based on a large-scale environmental sensor network. J. Build. Perform. Simul. 4(4), 359–369 (2011) B. Dong, K.P. Lam, A real-time model predictive control for building heating and cooling systems based on the occupancy behavior pattern detection and local weather forecasting, in Building Simulation, 7(1), 89–106, (Springer, Berlin Heidelberg, 2014) A. Ebadat, G. Bottegal, D. Varagnolo, B. Wahlberg, K.H. Johansson, Estimation of building occupancy levels through environmental signals deconvolution. In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings. (2013), pp. 1–8 V.L. Erickson, A.E. Cerpa, (2010) Occupancy based demand response hvac control strategy, in Proceedings of the 2nd ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Building, pp. 7–12 V.L. Erickson, M.Á. Carreira-Perpiñán, A.E. Cerpa, Observe: occupancy-based system for efficient reduction of HVAC energy, in Proceedings of the 10th ACM/IEEE International Conference on Information Processing in Sensor Networks (IEEE, 2011), pp. 258–269 S. Golestan, S. Kazemian, O. Ardakanian, Data-driven models for building occupancy estimation. In Proceedings of the Ninth International Conference on Future Energy Systems (2018), pp. 277–281 M. Gouda, S. Danaher, C. Underwood, Application of an artificial neural network for modelling the thermal dynamics of a building’s space and its heating system. Math. Comput. Model. Dyn. Syst. 8(3), 333–344 (2002) E. Hailemariam, R. Goldstein, R. Attar, A. Khan, Real-time occupancy detection using decision trees with multiple sensor types, in Proceedings of the 2011 Symposium on Simulation for Architecture and Urban Design, SimAUD’11 (Society for Computer Simulation International, 2011), pp. 141–148 R. Holz, A. Hourigan, R. Sloop, P. Monkman, M. Krarti, Effects of standard energy conserving measures on thermal comfort. Build. Environ. 32(1), 31–43 (1997) B. Huchuk, S. Sanner, W. O’Brien, Comparison of machine learning models for occupancy prediction in residential buildings using connected thermostat data. Build. Environ. 160, 106177 (2019) IEA, The critical role of buildings (2019). https://www.iea.org/reports/the-critical-role-ofbuildings H. Kim, T. Hong, J. Kim, Automatic ventilation control algorithm considering the indoor environmental quality factors and occupant ventilation behavior using a logistic regression model. Build. Environ. 153, 46–59 (2019a) S. Kim, Y. Song, Y. Sung, D. Seo, Development of a consecutive occupancy estimation framework for improving the energy demand prediction performance of building energy modeling tools. Energies 12(3), 433 (2019b) W. Kleiminger, C. Beckel, T. Staake, S. Santini, Occupancy detection from electricity consumption data. In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings (2013), pp. 1–8 N. Li, J. Li, R. Fan, H. Jia, Probability of occupant operation of windows during transition seasons in office buildings. Renew. Energy 73, 84–91 (2015) T. Lu, M. Viljanen, Prediction of indoor temperature and relative humidity using neural network models: model comparison. Neural Comput. Appl. 18(4), 345–357 (2009) J. Lu, T. Sookoor, V. Srinivasan, G. Gao, B. Holben, J. Stankovic, E. Field, K. Whitehouse, The smart thermostat: using occupancy sensors to save energy in homes, in Proceedings of the 8th ACM Conference on Embedded Networked Sensor Systems (2010), pp. 211–224 R. Markovic, S. Wolf, J. Cao, E. Spinnräker, D. Wölki, J. Frisch, C. van Treeck, Comparison of different classification algorithms for the detection of user’s interaction with windows in office buildings. Energy Proc. 122, 337–342 (2017)

1656

F. Dadras Javan et al.

R. Markovic, J. Frisch, C. van Treeck, Learning short-term past as predictor of window openingrelated human behavior in commercial buildings. Energy Build. 185, 1–11 (2019) A. Marvuglia, A. Messineo, G. Nicolosi, Coupling a neural network temperature predictor and a fuzzy logic controller to perform thermal comfort regulation in an office building. Build. Environ. 72, 287–299 (2014) L. Mba, P. Meukam, A. Kemajou, Application of artificial neural network for predicting hourly indoor air temperature and relative humidity in modern building in humid region. Energy Build. 121, 32–42 (2016) A. Mechaqrane, M. Zouak, A comparison of linear and neural network arx models applied to a prediction of the indoor temperature of a building. Neural Comput. Appl. 13(1), 32–37 (2004) J.W. Moon, Integrated control of the cooling system and surface openings using the artificial neural networks. Appl. Therm. Eng. 78, 150–161 (2015) J.W. Moon, J.-J. Kim, Ann-based thermal control models for residential buildings. Build. Environ. 45(7), 1612–1625 (2010) J.W. Moon, S.-H. Yoon, S. Kim, Development of an artificial neural network model based thermal control logic for double skin envelopes in winter. Build. Environ. 61, 149–159 (2013) G. Mustafaraj, J. Chen, G. Lowry, Thermal behaviour prediction utilizing artificial neural networks for an open office. Appl. Math. Model. 34(11), 3216–3230 (2010) G. Mustafaraj, G. Lowry, J. Chen, Prediction of room temperature and relative humidity by autoregressive linear and nonlinear neural network models for an open office. Energy Build. 43(6), 1452–1460 (2011) M. Narayanan, G. Mengedoht, W. Commerell, Importance of buildings and their influence in control system: a simulation case study with different building standards from Germany. Energy Environ. Eng. (IJEEE) 9, 413–433 (2018) F. Naspi, M. Arnesano, L. Zampetti, F. Stazi, G.M. Revel, M. D’Orazio, Experimental study on occupants’ interaction with windows and lights in Mediterranean offices during the non-heating season. Build. Environ. 127, 221–238 (2018) T.G. Özbalta, A. Sezer, Y. Yıldız, Models for prediction of daily mean indoor temperature and relative humidity: education building in Izmir, Turkey. Indoor Built Environ. 21(6), 772–781 (2012) S. Pan, Y. Han, S. Wei, Y. Wei, L. Xia, L. Xie, X. Kong, W. Yu, A model based on gauss distribution for predicting window behavior in building. 149, 210–219 (2019) Z. Pang, Y. Chen, J. Zhang, Z. O’Neill, H. Cheng, B. Dong, Nationwide hvac energy-saving potential quantification for office buildings with occupant-centric controls in various climates. Appl. Energy 279, 115727 (2020) J.Y. Park, M.M. Ouf, B. Gunay, Y. Peng, W. O’Brien, M.B. Kjærgaard, Z. Nagy, A critical review of field implementations of occupant-centric building controls. Build. Environ. 165, 106351 (2019) Y. Peng, A. Rysanek, Z. Nagy, A. Schlüter, Using machine learning techniques for occupancyprediction-based cooling control in office buildings. Appl. Energy 211, 1343–1358 (2018) M. Pritoni, J.M. Woolley, M.P. Modera, Do occupancy-responsive learning thermostats save energy? A field study in university residence halls. Energy Build. 127, 469–478 (2016) R. Razavi, A. Gharipour, M. Fleury, I.J. Akpan, Occupancy detection of residential buildings using smart meter data: a large-scale study. Energy Build. 183, 195–208 (2019) A.E. Ruano, E.M. Crispim, E.Z. Conceiçao, M.M.J. Lúcio, Prediction of building’s temperature using neural networks models. Energy Build. 38(6), 682–694 (2006) S.H. Rya, H.J. Moon, Development of an occupancy prediction model using indoor environmental data based on machine learning techniques. Build. Environ. 107, 1–9 (2016) J. Scott, A. Bernheim Brush, J. Krumm, B. Meyers, M. Hazas, S. Hodges, N. Villar, Preheat: controlling home heating using occupancy prediction, in Proceedings of the 13th International Conference on Ubiquitous Computing (2011), pp. 281–290 S. Shi, B. Zhao, Occupants’ interactions with windows in 8 residential apartments in Beijing and Nanjing, China. Build. Simul. Int. J. 9(2), 221–231 (2016)

Application of Machine Learning in Occupant and Indoor Environment. . .

1657

Z. Shi, H. Qian, X. Zheng, Z. Lv, Y. Li, L. Liu, P.V. Nielsen, Seasonal variation of window opening behaviors in two naturally ventilated hospital wards. Build. Environ. 130, 85–93 (2018) M.K. Singh, R. Ooka, H.B. Rijal, M. Takasu, Adaptive thermal comfort in the offices of North-East India in autumn season. Build. Environ. 124, 14–30 (2017) M. Soleimani-Mohseni, B. Thomas, P. Fahlen, Estimation of operative temperature in buildings using artificial neural networks. Energy Build. 38(6), 635–640 (2006) F. Stazi, F. Naspi, M. D’Orazio, Modelling window status in school classrooms. Results from a case study in Italy. Build. Environ. 111, 24–32 (2017) B. Thomas, M. Soleimani-Mohseni, Artificial neural network models for indoor temperature prediction: investigations in two buildings. Neural Comput. Appl. 16(1), 81–89 (2007) W. Wang, J. Chen, G. Huang, Y. Lu, Energy efficient hvac control for an IPS-enabled large space in commercial buildings through dynamic spatial occupancy distribution. Appl. Energy 207, 305–323 (2017) W. Wang, T. Hong, N. Li, R.Q. Wang, J. Chen, Linking energy-cyber-physical systems with occupancy prediction and interpretation through wifi probe-based ensemble classification. Appl. Energy 236, 55–69 (2019) Y. Wei, H. Yu, S. Pan, L. Xia, J. Xie, X. Wang, J. Wu, W. Zhang, Q. Li, Comparison of different window behavior modeling approaches during transition season in Beijing, China. Build. Environ. 157, 1–15 (2019) Z. Yang, N. Li, B. Becerik-Gerber, M. Orosz, A systematic approach to occupancy modeling in ambient sensor-rich buildings. Simulation. 90(8), 960–977 (2014) M. Yao, B. Zhao, Factors affecting occupants’ interactions with windows in residential buildings in Beijing, China. Proc. Eng. 205, 3428–3434 (2017a) M. Yao, B. Zhao, Window opening behavior of occupants in residential buildings in Beijing. Build. Environ. 124, 441–449 (2017b) D. Yu, A. Abhari, A.S. Fung, K. Raahemifar, F. Mohammadi, Predicting indoor temperature from smart thermostat and weather forecast data, in Proceedings of the Communications and Networking Symposium (2018), pp. 1–12 G.Y. Yun, K. Steemers, Time-dependent occupant behaviour models of window control in summer. Build. Environ. 43(9), 1471–1482 (2008)

Software Engineering Smart Energy Systems Marco Aiello, Laura Fiorini, and Ilche Georgievski

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Software Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Service-Oriented Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 AI Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Demand-Side Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Software Engineering Future-Proof Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Smart Energy Systems’ General Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 The Blueprint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Case Study: Smart Energy Offices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Case Study: Following Smart Grid Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Optimal Operation Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Cost-Efficiency and Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1660 1661 1661 1662 1663 1664 1665 1665 1670 1672 1673 1675 1676 1681 1681 1683 1684 1685 1685

M. Aiello () · I. Georgievski Service Computing Department, IAAS, University of Stuttgart, Stuttgart, Germany e-mail: [email protected]; [email protected]; [email protected] L. Fiorini Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence, University of Groningen, Groningen, The Netherlands e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_21

1659

1660

M. Aiello et al.

Abstract

Digitalization is having a major impact on the energy sector at all levels, supporting its transition to more sustainable and smart infrastructure. Such transformation, which enables highly improved monitoring, forecasting, and automation, comes though at the cost of increased complexity. Smart Energy Systems, therefore, pose new engineering challenges that require novel tools. In the present chapter, we focus particularly on the software-engineering challenges and propose a blueprint for the design and implementation of smart energy systems. We discuss specific qualities for such systems in the form of functional, non-functional, and user requirements. We instantiate the proposed blueprint in two case studies, one about smart energy offices and one regarding demand-side management in a residential setting. Keywords

Artificial intelligence planning · Demand-side management · Software engineering · Service-oriented computing · Requirement engineering · Smart offices · Smart homes

1

Introduction

The success of reliable, cost-effective, sustainable, and intelligent access to energy strongly depends on how energy systems are designed and implemented, and what methods are used to operate them. The advent of the digital era offers unprecedented opportunities on how to realize such systems by empowering them with real-time monitoring and automation. Digitalization is particularly relevant in current times of deep transformation of the energy sector. The ever-increasing use of Distributed Energy Resources (DERs), such as solar panels, the abandoning of the centralized system for generating and distributing power, and the greater autonomy of loads are increasing the uncertainty and complexity of energy systems. The shift to electricitybased mobility is an additional source of uncertainty and potential system stress. Digitalization is a fundamental tool to support the current trend and make our energy systems smarter. Data can be cheaply collected and processed at all ends, enabling, in turn, accurate monitoring and forecasting at all levels. Models and patterns emerging from the data are crucial for the consequent possibility of automating and optimizing the systems, thus reaching true smartness. To take advantage of the automation and optimization opportunities given by the digitalization of the infrastructure and the load, we though need to rethink the design and operation of energy systems. On the one hand, current energy management systems have functionally focused, hard-coded designs, making them incapable of coping with the evolution of technologies and techniques, and unusable in the new energy setting with digitally controllable loads. On the other hand, current energy management systems often lack mechanisms for recognizing and improving practices of energy consumption, especially in real time.

Software Engineering Smart Energy Systems

1661

As a consequence, a set of new requirements are emerging that are both fundamental and defining of the new type of energy systems at the same time. These requirements often call on specific patterns and techniques for their satisfaction. Data science and machine learning play a crucial role in satisfying the requirements that have to do with monitoring, and, especially, forecasting. Next to these approaches rooted in statistics, Artificial Intelligence (AI) approaches rooted in discrete mathematics, logic, and symbolic reasoning are equally fundamental to address the requirements that have to do with optimization, planning, scheduling, and handling uncertainty. In other words, we put forward that realizing truly smart energy systems is not only a matter of smartly dealing with energy and user data but also an issue of appropriate modeling, planning, and reasoning. The present chapter is not intended to be a review of the state of the art, as such have already been proposed, e.g., Lund et al. (2017). The chapter is rather an illustration of our proposal for general software-engineering requirements and a blueprint for state-of-the-art and future-proof smart energy systems. The proposal of the blueprint is based on more than 10 years of research and development experiences in the field (Pagani and Aiello 2011; Georgievski et al. 2012, 2017; Fiorini and Aiello 2019a, 2020; Fiorini et al. 2020; Nguyen and Aiello 2013; Georgievski et al. 2020). The remainder of the chapter is organized as follows. We generally introduce the main topics relevant to our proposal in Sect. 2. We subsequently present and discuss 16 fundamental software-engineering requirements and the blueprint from a software-engineering perspective in Sect. 3. Then, we present two case studies: the focus of Sect. 4 is on smart office buildings, while Sect. 5 is about residential spaces following grid signals. Final considerations are drawn and presented in Sect. 6.

2

Background

Our proposal rests upon concepts, methods, and techniques from the fields of Software Engineering, AI Planning, and Demand-Side Management. We first introduce the basics of the process of software engineering followed by an introduction to Service-Oriented Computing, which provides modern means to design software systems. We then briefly describe AI planning with an emphasis on Hierarchical Task Network planning as a technique we use in one of our case studies. We finalize this section by explaining Demand-Side Management in a nutshell.

2.1

Software Engineering

Energy infrastructures and utilities are controlled by using software systems. Software systems in general are not governed by physical constraints, making them abstract and intangible entities. They can easily become highly complex, difficult to understand, and expensive to adapt and maintain. Therefore, the need for good practices for engineering software systems arises.

1662

M. Aiello et al.

Software Engineering is an engineering discipline whose objective is to specify, design, implement, and validate software systems. It provides a systematic approach of applying theories, methods, and tools that are appropriate to circumstances specific to the domain or organization for which a software system needs to be produced. There are four main activities in the process of software engineering: software specification, software design and implementation, software validation, and software evolution (Sommerville 2010). These are complex activities and in practice typically include sub-activities. Software specification, and software design and implementation are the most relevant to our treatment, so we focus on these two activities next. Software specification is the process of identifying and defining what is required from a system, and which properties are constraining the operation and development of such a system, and what (Sommerville 2010). The former yields functional requirements, while the latter gives non-functional and user-related requirements. In other words, functional requirements describe the functions the system should provide and how should the system react to particular inputs and in particular situations. Non-functional requirements describe constraints on the functions offered by the system. These typically include performance requirements (e.g., response time), maintainability requirements (e.g., system complexity), interoperability requirements (e.g., standard interfaces), reliability requirements (e.g., system failures), and others (Bourque et al. 2014). User-related requirements describe needs or expectations from users, such as security, comfort, productivity, and sustainability. The outcome of the software specification is a list of requirements that involved stakeholders agree upon. Typically, requirements are gathered through observation of existing software systems, discussions with relevant stakeholders, analysis of core tasks, and similar activities. Requirements can be defined in the form of abstract statements, which are appropriate for end users, or detailed descriptions, which are needed for system developers. The next activity is to convert the software specification into an executable system. This is accomplished using software design and implementation. The software design involves a description of the architecture, interfaces, and components of the system to be implemented. To design architectures, we need to identify the structure of the system, including core components and their interrelationships (Pressman and Maxim 2015). Architecture designs are typically modeled using the Unified Modeling Language (UML) (Booch et al. 2005). To design interfaces, it means to define how system components can be used without having to know their implementation details. To design components, we need to define how each component will operate, including models and algorithms. The software design is then followed by writing programs that implement the system.

2.2

Service-Oriented Computing

Service-Oriented Computing (SOC) is a prominent paradigm for designing and developing modular, scalable, loosely coupled, and heterogeneous software systems

Software Engineering Smart Energy Systems

1663

(Papazoglou and Georgakopoulos 2003). Prevalent and essential is the concept of service, which requires software functions that address specific functional requirements to be encapsulated, hiding implementation and hardware details away. As services are modular, they can be used as blocks to build a system’s architecture. In SOC, such an architecture is termed a Service-Orientation Architecture (SOA). An SOA provides design principles that allow interoperable services to be discovered, utilized, and composed in a way suitable to a context of interest. The architecture is so flexible and dynamic that the blocks can be brokered and composed at runtime (Lazovik et al. 2006). Among the design principles, the most relevant to our treatment are (Erl 2007): • Interoperability. Service should have standardized well-defined interfaces. The exposed functionality is independent of the underlying implementation and hardware. Interfaces can be described in several formats using different specification languages, such as the Web Service Definition Language (Chinnici et al. 2007) and RESTful Service Description Language (Robie et al. 2013). • Loose coupling. Services run and execute independently of each other. Services publish and expose their interfaces often resorting to a broker. This results in services that are dependent on the interface exposed, but not on the internal classes of other services. Services can be discovered and bound to at execution time. • Autonomy. Services implementations and their versions are independent of any other service. The execution of a service does not affect the execution of any other service, except by direct service invocation. • Reusability. Services are designed as functional units and can therefore be invoked by any service. Services can be part of several implementations, even at the same time. • Scalability. Scalability manifests itself at two levels. At the service level, it requires the handling of any number of requests. At the architectural level, it requires the system to manage any possible workload. Since implementations are hidden, one can think of elastic service implementations that increase the number of resources as invocations surge. The potential benefits of using services and SOAs for the software design include the possibility to organize software systems in large-scale environments due to the high-level abstraction of services, interoperation of software developed by different organizations due to standardization, and development of general-purpose tools to support and manage the whole life-cycle of a system, including design, development, maintenance, monitoring, and other aspects (Singh and Huhns 2005).

2.3

AI Planning

AI planning is a flourishing research and development discipline that provides computational techniques for searching and selecting actions that satisfy some user goal (Ghallab et al. 2004). AI planning is highly suitable for complex and changing

1664

M. Aiello et al.

situations in domains that require guarantees of efficiency, safety, and reliability. Popular examples of such domains include space missions (e.g., Cesta et al. 2007), logistics (e.g., Tate et al. 1996), robotics (Karpas and Magazzeni 2020), video games (e.g., Kelly et al. 2008), ubiquitous computing (Georgievski and Aiello 2016), and others. Among the AI planning techniques, Hierarchical Task Network (HTN) planning is a prominent one due to its rich domain knowledge, support for hierarchies, and speed of computation (Georgievski and Aiello 2015). HTN planning solves a planning problem P that consists of an initial state s0 , an initial task network tn0 , and a set of primitive and compound tasks T, all described using a set of predicates Q as defined in predicate logic. Primitive and compound tasks can be accomplished by operators or methods, respectively. An operator represents a transition from a state to another, while a method describes how to decompose a compound task into greater details. Planning is performed by repeatedly decomposing tasks from the initial task network until operators executable in the initial state are reached. A solution, called a plan, is a sequence of operators that when executed in the initial state accomplish the initial task network.

2.4

Demand-Side Management

The concept of Demand-Side Management (DSM) refers to a portfolio of technologies and measures to influence the consumption of electricity in order to optimize the planning and operation of energy systems (Gelazanskas and Gamage 2014). DSM comprises two main categories of measures, namely, energy efficiency programs and Demand Response Programs (DRPs). The former, also known as conservation programs, aim at reducing the consumption of electricity, while supplying the same services. The latter, also known as load shifting programs, aim at flattening the load curve, shifting part of the demand from peak periods to off-peak ones. Once involving mostly industrial consumers by means of load shedding mechanisms (Vallés et al. 2015), today it is increasingly considered also for residential end users. DSM can increase the responsiveness of a large part of the energy consumption in buildings (Goy and Sancho-Tomás 2019). With the dramatic increase of variable renewable generation, DSM and DRPs, in particular, are gaining more and more importance as a source of flexibility to guarantee the balance between supply and demand (Strbac 2008; Mathiesen et al. 2015). Users may adjust their loads during times of system stress in response to signals coming from the grid, possibly receiving a monetary revenue for their flexibility. Incentive-based programs involve the consumers in load shaping by offering participation payments, such as bill credits or discount rates, in return for giving up control of certain appliances or allowing load limitation during periods of local reliability-threatening peak demand or high energy prices. Price-based programs engage the consumers by offering dynamic electricity pricing rates that reflect the energy price and availability at the time of consumption. The goals of DRPs are twofold: on the one hand, the users can

Software Engineering Smart Energy Systems

1665

reduce their energy bills by modifying their normal consumption patterns according to price variability; on the other hand, the utility can reduce the risk of bottlenecks along lines, improving the system reliability, and postponing expensive investments in new generation plant and increasing of infrastructure capacity (Siano 2014).

3

Software Engineering Future-Proof Energy Systems

Our proposal is a shift from coarse-grained, functionally focused designs of energy systems to a software design and realization that consider both functional and nonfunctional perspectives at any scale. We argue that this shift is possible by looking at the capabilities of energy management systems through the prism of service orientation given its benefits, such as extensibility, reusability, interoperability, and scalability. Service orientation is the architecture design style of the blueprint we propose for realizing smart energy systems that can sustain the test of time. The components of a system’s architecture are a set of services, which may take various forms and designs, depending on a system’s needs (Georgievski et al. 2020).

3.1

Smart Energy Systems’ General Requirements

We begin our journey towards a blueprint by identifying the general softwareengineering requirements of smart energy systems. In what follows, we consider functional, user-related, and non-functional requirements.

3.1.1 Functional Requirements Functional requirements define the basic behavior of a system. To elicit them, we need to understand what the system should do, or must not do. A primary function of smart energy systems is to minimize operation costs and environmental impact while satisfying the power needs of the users (Fiorini and Aiello 2019a). There are two essential components to this: the first component is that of understanding the current state of the world in terms of energy needs and availability, also in relation to the expectation of future states; the second component is the ability to act according to the current state and expected ones. The first component is a data-intensive one. It requires the collection and processing of large amounts of data points for the reliable determination of the current state and the best possible forecast of a future state. The second component is a planning and scheduling one. A (set of) optimal future state(s) is defined, and the system exhibits smartness by planning and scheduling all possible actions that bring the system as close as possible to the optimal state. In other terms, an energy system must first have historical and current information about the production, consumption, and energy prices, and then decide on a course of actions that balance the supply and energy demand, optimize DERs operation, reduce energy consumption, and/or minimize greenhouse-gas emissions. Having this in mind, we discuss several high-level functional requirements of energy systems in the following.

1666

M. Aiello et al.

An energy system should be able to gather data from a wide range of devices, including DERs, electrical batteries, smart meters, inverters, appliances, thermostats, and other sensors that help define the user’s state. The data is to be used by the other functions of the energy system for forecasting and decision-making. FR 1. A smart energy system gathers data about weather, energy production of DERs, energy consumption of consumers, and other energy-relevant environmental and behavioral parameters. An energy system should be able to forecast the production of energy of the involved DERs using historical data and weather data. Similarly, for the energy drawn by the distribution networks, it should be able to forecast the expected costs and relevant environmental parameters, e.g., the emission per kWh that depends on the used energy mix. The system can also forecast energy consumption using data coming from smart metering infrastructures. These can work at the building, unit, or even appliance level. Lastly, the energy system may need to forecast relevant weather data, such as wind speed, outdoor temperature, solar radiation, and other weather parameters. FR 2. A smart energy system forecasts and uses energy production of DERs, energy consumption and load demand of consumers, and weather data. An energy system should be able to interact with the electricity market and react to dynamic pricing and other types of network signals. In addition, it has to balance the operation of DERs and respond to the demand of consumers. The expectation is that future power grids have full duplex communication with the load and can provide signals to incentivize and even control the load. Such signals are most likely based on the price of energy, may this be the spot price, hour-ahead, day-ahead, but can potentially be of other types, such as emission-based. The signals can also refer to overall energy consumption or the marginal changes. That is, kWh of energy could have a price that is not fixed but rather based on the discrepancy from a set value. FR 3. A smart energy system makes decisions based on signals coming from the smart grid. The task of the smart energy system is thus one of optimization for a given temporal horizon. Based on the forecasts and component models, the system balances the demand and response taking into account the constraints coming from the user needs, DER generation, smart grid signals, and planning, scheduling, and acting on the controllable loads. For example, the use of batteries can be managed by planning and scheduling the charge and discharge actions, the interaction with the electricity market can be orchestrated by planning the actions for buying from and selling energy to the grid, and the use of controllable loads can be managed by planning and scheduling their switching operations. An energy system operates in real time by considering possible adjustments of forecasted plans with respect to

Software Engineering Smart Energy Systems

1667

the actual data coming in real time from the market (e.g., price signals), the grid (e.g., congestion), DERs (e.g., power output), and demand side. FR 4. A smart energy system computes feasible plans based on forecasted data and is always ready to intervene with corrective actions based on real-time data regarding users and possible demand-response unbalances. An energy system should be able to arrange actions of controllable devices on the demand side to increase energy efficiency by, for example, decreasing peak load or reducing energy consumption. FR 5. In its balancing decisions, a smart energy system always opts for the ones that increase the overall energy efficiency. Functional requirements FR 1–FR 5 are high-level mandates that define the behavior or business logic of a smart energy system. Such requirements are often referred to as business requirements. Business requirements need to be made more specific during the development process of an energy system. Other categories of functional requirements are also important, such as administrative requirements (e.g., reporting), user requirements (e.g., specification of the operation time of a household appliance), and system requirements (e.g., system actions and responses). The treatment of such functional requirements is out of the scope of the current work.

3.1.2 User-Related Requirements A smart energy system revolves around the energy needs of users. While typical user requirements are functional requirements specified by the users, there is a number of general requirements that are relevant for the user. Similarly to Asimov’s laws of robotics (Asimov et al. 1984), stating that robots should never harm humans in their operation, we have the following user-related requirements. Clearly, no optimization of energy can come at the price of putting at risk the safety of people. UR 1. No decision made by a smart energy system can affect the safety of any human. Collecting user data for energy saving implies also being able to make deductions about the users, such as their location, their activity, their working patterns, and so on. One should therefore be very careful with what data is collected, stored, and who can access it. The “principle of least privilege” or “need to know” should drive the implementation of the privacy requirement. UR 2. Privacy-sensitive user data collected by a smart energy system for its operation should be kept to the strictly necessary and kept safe from accidental disclosure.

1668

M. Aiello et al.

In an office environment, the economic value of the amount of energy that one can save around a user is several orders of magnitude lower than his/her daily costs for the organization. In other terms, hindering the productivity of someone for saving a few kWh of electricity is counterproductive. UR 3. The effectiveness of user actions/tasks/intentions cannot be affected by the operation of a smart energy system. Similar to productivity is the requirement for comfort. UR 4. The comfort of users cannot be affected by the operation of a smart energy system. At most, it can be improved. Then comes the sustainability requirement that establishes that a smart energy system should strive for improving the sustainability of all user actions. It is related to the functional requirement FR 5. UR 5. The operation of a smart energy system improves the sustainability of user actions. Smart energy systems are very complex systems that gather great amounts of data and operate many actuators. Often the operations of these might be hard to understand by the user. Why did a light go on, or why was the temperature of a room raised? Ideally, the user should always have the possibility to be given an explanation on why a certain action was taken, based on current or expected states and with which final goal. UR 6. The decisions and subsequent actions taken by a smart energy system can be explained to users. The user-related requirements should be satisfied in strict total order. Take safety, for instance. It is above all and cannot be compromised by satisfying other requirements such as sustainability or comfort. In other terms, one usually considers: UR 1 > UR 2 > UR 3 > UR 4 > UR 5 > UR 6.

3.1.3 Non-functional Requirements Non-functional requirements define the characteristics that affect the experience of using a system. To elicit them, we need to understand how a software system should perform when meeting the functional requirements and how easy such a system should be to use. The early detection of non-functional requirements enables various constraints to be considered and addressed into early architecture designs of energy systems as opposed to being refactored at a later time during the software life-cycle of the systems. We want to identify the main non-functional requirements that an energy system under development should meet. We cannot elicit such non-functional requirements

Software Engineering Smart Energy Systems

1669

by interviewing stakeholders, because we are not trying to develop a specific energy system, or examining existing studies, because they do not consider them. What we can do is, also based on our past experience with specific applications, review the most common design choices of current energy systems and along that identify the challenges related to the quality of those systems as software applications. Current energy systems are commonly designed as monolithic applications (Lee et al. 2016; Georgievski et al. 2020). These are single-tiered software systems in which functional requirements are implemented into a single, often large, program unit. As such a unit implements all the functionality, an energy system represents a rather complex application that limits the understanding of its internal working and dependencies. A countermeasure to this complexity is to partition the design of the software system into logical components that correspond to appropriate functional requirements so that the system becomes easy to understand and maintain. NFR 1. The architecture of a smart energy system shall be partitioned into logical components that encapsulate its functions and implementation choices. In a monolithic energy system, it might be difficult to understand the dependencies of separate software parts dedicated to addressing corresponding functional requirements. To put it differently, the dependencies of tightly coupled software parts might be unknown. This makes energy systems entangled with specific techniques and smart-grid technologies, which directly affects the modifiability of the systems: addressing a new functional requirement or modifying one of FR 1–FR 5 is almost impossible without impairing the rest of requirements and disrupting the overall system functionality. A countermeasure to such software tightness is to increase the system’s degree of coupling. NFR 2. A smart energy system shall have well-defined and minimal dependencies between its components. As with other software systems, the performance of energy systems directly depends on their workload. For example, adding more DERs or load profiles may affect the responsiveness of an energy system while meeting FR 1 and FR 2. When energy systems are monolithic and tightly coupled, it is often difficult to appropriately scale them. Continuing with the example, if the energy system needs to handle more DERs and load profiles, one may need to deploy and run multiple instances of the system for the sake of increasing the availability of capabilities that implement FR 1 and FR 2. This results in also scaling system capabilities that do not need to be scaled, thus wasting computational resources. NFR 3. A smart energy system shall maintain its performance and function under an increase of the workload without a corresponding increase in its internal complexity.

1670

M. Aiello et al.

Designing and developing energy systems for specific purposes is currently common practice, but it raises the issue of being able to port the software solution to new environments or to adapt to different uses. As the need for using an energy system in more settings increases, such specificity cannot offer much as its reconfiguration will probably require substantial development time, thus slowing down the time to market. A countermeasure to this is designing energy systems as reusable software applications. NFR 4. A smart energy system shall allow for the use of its existing software components to compose subsequent and new types of energy systems. Consider, for example, that a bug has been found in the code dedicated to satisfying FR 3. What will happen with a monolithic energy system in this case? The most probable scenario when fixing the bug is that the operation and even the availability of the energy system as a whole will be affected. A countermeasure for this, in addition to NFR 1–NFR 3, is to account for software reliability. NFR 5. A smart energy system shall be made to perform as expected even under failures.

3.2

The Blueprint

Smart energy systems are composed of many components both at the physical and software levels. These are typically distributed, independent, and heterogeneous. Therefore, to create a general blueprint that satisfies the software-requirements requirements just reviewed, one needs an architecture built upon modern design principles. To that end, we employ an SOA and provide a more concrete illustration of possible services and architecture implementation. We use commercial buildings as a case study while considering FR 1–FR 5 and NFR 1–NFR 5. DERs are represented by three types of (simulated) resources: solar panels, wind turbines, and electrical batteries. Load demand should come from (historical or simulated) load profiles of buildings. The dynamic pricing structure comes from the actual energy market regulation, say a day-ahead pricing model. Since energy providers publish energy prices one day in advance, this model allows for planning the use of loads for the upcoming day. Day-ahead prices are commonly available in a time window of 1 h. Considering the design principles mentioned in the previous section, the core set of services is: • DER Service: it provides current and forecast data about solar panels, wind turbines, and batteries. • Load Service: it provides current and forecast data about the overall building consumption. • Price Service: it provides the current price of energy and the forecast of energy prices for the next 24 h.

Software Engineering Smart Energy Systems

1671

• Demand Management Service: it provides energy efficient plans to regulate the variable load and the use of the energy storage to satisfy the expected load. • Optimization Service: it computes plans that balance the operation of DERs and satisfy the demand. • Coordination Service: it executes and monitors the execution of the plans coming from the demand and optimization services. • Weather Service: it provides current and forecast data about the local weather. • Database Service: it provides data storage services for power, load, prices, plans, and weather. It is an ancillary service. For each service, there is a design of an Application Programming Interface (API) with one or more operations associated with a description of the operation’s purpose. An operation has a specified return type, and a set of parameters that need to be provided by a consuming service. For example, an API of the DER Service might be dedicated to providing generated or available power on an hourly basis via three operations: one for measuring the output of solar panels, another for measuring the output of a wind turbine, and a third one for measuring the available power in a battery. A consuming service that wants to measure, let say, the output of solar panels, needs to provide several parameters to the corresponding DER Service’s operation. The parameters and their description are given in Table 1. The APIs of other energy system services can be defined in a similar way. A high-level overview of the architecture design for our energy system in the form of a component diagram described in UML is shown in Fig. 1a. Except for the Coordination and Database Services, all other services are intended to be stateless. The Coordination Service acts as a supervisor, making sure our system is correctly executed. The practical benefit of such a design choice is that if and when the logic of the execution needs to be modified, only the implementation of the Coordination Service will be affected. Consider now we want to extend our energy system with a new functionality of empowering users to interact with the system using a dashboard. This new requirement does not call for the restructuring of the existing system. Instead, we could create two new services, one for the dashboard itself, and another one that will act as middleware between the existing services and the Dashboard Service. In

Table 1 Parameters required by the DER Service operation for measuring the output of solar panels Parameter panelArea panelYield panelAngle panelLatitude solarRadiation temperature

Data type Float Float Integer Float Float Float

Description Total area of the solar panels in m2 Efficiency of the solar panels in % Angle of the installation of the solar panels in degrees Latitude of the solar panels in degrees Solar radiation on a tilted surface Current outdoor temperature in degrees Celsius

1672

M. Aiello et al.

Fig. 1 UML component diagram of the energy system architecture

this way, we are reusing existing services to extend the functionality of the system. The extension of the system architecture is depicted in Fig. 1b. When it comes to scalability, we could increase the number of services as needed. For example, if the Middleware Service starts to become a bottleneck of the system, we replicate it and balance the load coming from the large numbers of Dashboard Service clients.

4

Case Study: Smart Energy Offices

The management of energy in buildings needs to be aligned with the experiences and behavior of building occupants. Energy systems in buildings must first understand the situations occupants are in and activities they perform, and then react appropriately. The reaction should improve the occupants’ productivity and comfort or at least not disrupt them (cf. UR 4), while the building environment becomes as energy efficient as possible (cf. UR 5). And the relationship between people’s behavior and the demand for energy is known: their behavior has a large impact on the demand for space heating, cooling and ventilation, lighting, and appliance usage (Page et al. 2008). While current energy systems for buildings offer at most reaction control (feedback loops) and centralized human-based control, future-proof smart energy systems for buildings should understand buildings’ situations and occupant activities, and coordinate the use of building devices such that the occupant activities are supported and the energy consumption is reduced. Searching through all possible contextual situations every time occupants perform some activities and

Software Engineering Smart Energy Systems

1673

finding appropriate solutions to device coordination under the constraints of activity support and energy efficiency is a complex task. We present an approach based on AI planning that automatically and dynamically finds such solutions as soon activities of occupants have been recognized. Our approach satisfies the software requirements of Sect. 3.1 and adopts the design principles in its design.

4.1

Design

The design revolves around the Demand Management Service. It is based on two assumptions related to building environments. First, buildings need to have some fraction of their physically bound space logically organized in a way that occupant activities can be observed. Second, observations about building environments and occupant activities need to be made, and the environments are controllable using IoT actuators embedded in the building. In the following, we formally define the problem that the Demand Management Service is faced with, and how it relates to an AI planning problem. Such a formal approach allows us to (1) make the Demand Management Service suitable for many possible implementations of energy systems as the formalism is not limited to a specific building environment; (2) improve the maintainability and evolvability of the Demand Management Service and the overall energy system (Bettini et al. 2010); and (3) use plans computed by AI planning as sound solutions for efficiency improvement and overall smartness.

4.1.1 Formal Approach A building environment E is a tuple V , L, where V is a set of variables each v of which varies over D v and L is a set of locations (Georgievski et al. 2017). Variables represent data acquired by IoT devices, and locations represent the logical organization of the building space. One or more variables can be associated with a single location (e.g., room0353Lamp is at desk), while locations themselves may have some organizational relationships (e.g., desk is a sub-location of room0353). We refer to such associations and relationships as spatial properties (Georgievski and Aiello 2016). In addition, many of the building locations are associated with specific activities. For example, a kitchen is where people have coffee or eat. We refer to this association as an activity area. An occupant activity act is a tuple n, l, where n is the name of the activity and l ∈ L is the activity area. Each act is derived from the correlation of all v ∈ V that are associated with the location of act. The set of recognized activities, Act, together with the building environment E provide a snapshot of the building at a particular point in the time. We refer to such a snapshot as a context. Certain types of buildings may need to adhere or maintain some quality conditions enforced by building standards, corporate policies, health protocols, national laws, etc. For example, the European standard for lighting in indoor work places defines that basic requirements, such as light level, should be considered for existing and future buildings in general situations and diverse specific activities taking place there (European Committee for Standardization 2011). Conditions may specify a

1674

M. Aiello et al.

recommended state (e.g., the light level when working on a computer should be between 450 and 500 lux), or limit alterations (e.g., up to a certain amount of carbon can be emitted). So, buildings need to maintain such quality conditions while trying to reduce the energy consumption as they affect the activities and performance of occupants. The environmental perspective of such conditions encompasses not only the social dimension, such as the quality of life of occupants and health of people, but also the economic dimension, including resource management, and energy dimension, including demand management (cf. UR 3–UR 5). We express a building condition ψ as a propositional formula over building variables and their values. For example, consider that when someone is working on a computer in office room0353PC, the light level should be within the recommended range of 450 and 500 lux. The building condition that must be satisfied and maintained by the building when trying to reduce the energy demand is defined as (room0353PC = active) ∧ (room0353Lux1 > 450) ∧ (room0353Lux1 < 500). The recognized occupant activities may require an immediate actuation on the environment so that the building can support the occupants in performing their activities, but also to make sure energy is not consumed unnecessarily while providing such support. The building needs to coordinate devices, i.e., actuators, that can act upon their values using one or more actions. The coordination must be accomplished according to the building conditions and considering the building properties. A building coordination problem P B is a tuple E, , P rop, Act, where E is the building environment, is a set of building conditions, P rop is a set of building properties, and Act is a set of activities. Given a set of device actions A, we say that α is a satisfying coordination for P B if and only if α ⊆ A that transforms V into one that supports Act according to . So, a satisfying coordination is a sequence of device actions that changes the values of building variables such that the building supports the performance of occupant activities and reduced energy consumption without violating building conditions. For example, turning off all lamps in someone’s office is a satisfying coordination for the activity of working with a computer when the light level would be 480 lux, which is in accordance with the building conditions (between 450 and 500 lux). This clearly reduces energy consumption, too. A request for building coordination is achievable if and only if there exists at least one satisfying coordination for it. This means that if there is no sequence of device actions that can support the performance of current occupant activities while satisfying building conditions, then the occupant activities cannot be supported as expected and the energy demand cannot be influenced. The building coordination problem has a close correspondence with problems that HTN planning is designed to solve. Given a building coordination problem P B and an HTN planning problem P, we say that P corresponds to P B if and only if s0 is the initial state corresponding to V ; tn0 is the initial task network corresponding to Act; V ⊆ V ; Q is the set of predicates corresponding to Boolean variables in V and P rop; and T is the set of tasks such that primitive tasks correspond to device actions in A and compound tasks correspond to activities in Act and environmental process based on P rop and . There are two important properties of this correspondence. The first is about correctness: if a request represented by Act is achievable, then

Software Engineering Smart Energy Systems

1675

there exist a plan π for P. The second one is about soundness: if there exist a plan π that is a solution to P, then we can construct a sequence of acting services α based on π that is a satisfying coordination for P B . For the full formal correspondence, we refer to Georgievski et al. (2017).

4.1.2 Separation of Concerns The component design of Demand Management Service meets NFR 1 and NFR 2. The service represents a composition of two services, one responsible for the recognition of occupant activities, and another one for solving the building coordination problem. The service depends on the functionality provided by two other services: a service for collecting data from IoT deployed in a building environment, and a service for storing and retrieving data. The Demand Management Service is designed to support indirect communication – it uses message queues to interact with other services without necessarily knowing them and/or their physical location. This component design choice can significantly simplify the implementation of separate services and also improve the performance, scalability, and reliability of the energy system. The Activity Recognition Service is responsible for recognizing occupant activities. It provides an API whose operation requires sensor data and provides a set of activities performed at all locations of interest at any time. The operation does not impose any technical restriction, so various methods can be employed to recognize activities. In addition, the interface design can be easily extended with an operation to predict occupant activities (which occupant activities may occur at what time) (Georgievski et al. 2019). The Building Coordination Service is composed of three services of separate concerns: Problem-Conversion Service, which translates data about a building environment into an HTN planning problem instance, a Problem-Solving Service, which is the main functionality used to solve HTN planning problems, and a Utility Service that provides supporting capabilities, such as message conversion, exception handling, and other functionality satisfying system requirements. The Building Coordination Service depends on a service that provides real-time sensor data, a database service that provides historical data and storing capabilities, and the Activity Recognition Service.

4.2

Implementation

We provide insights into the implementation choices related to each service part of the Demand Management Service. All services support communication using JSON, an open standard file format using text and loose object orientation. For the sake of traceability with asynchronous communication and readability and consistency in generic messages, a basic template is specified to which all the messages of components should adhere. Depending on the data that a component requires, the body of the JSON object may change. We partly address NFR 5 by using the

1676

M. Aiello et al.

RabbitMQ messaging framework, which is a highly reliable enterprise messaging system based on the emerging AMQP standard (Videla and Williams 2012). The Activity Recognition Service is able to recognize the following activities of occupants in office buildings: working with a computer, working without a computer (e.g., reading), having a meeting, drinking (e.g., having coffee), eating, presence, and absence. Two techniques have been used to implement the service, ontological reasoning and regression. For more details on the actual implementation of each technique, see Georgievski et al. (2019, 2017). Each of the sub-services of the Building Coordination Service is implemented using the capabilities of the HTN-based planner we developed (Georgievski 2015). The Problem-Solving Service uses the planning capability and requires two pieces of input: a domain description and a problem specification, including an initial state and initial task network. Both inputs should be described in the Hierarchical Planning Definition Language (HPDL) (Georgievski 2013). While the domain description is manually specified by a domain expert, the problem specification is automatically generated by the Problem-Conversion Service (cf. NFR 4). The service transforms the information coming from the building environment into an HPDL problem specification and transforms it together with the HPDL domain description further into programming-level constructs. Upon receiving a request with appropriate arguments, the Building Coordination Service may check, for example, the correctness of an HPDL domain or problem, the consistency of a required problem and domain, or search for a solution. All services are implemented as RESTful Web services.

4.3

Benefits

We show the benefits of our Demand Management Service by evaluating its energy efficiency in a restaurant of an office building at the University of Groningen, The Netherlands, and the performance of its two sub-services, Activity Recognition Service with respect to accuracy in a living lab of our own building at the University of Stuttgart, Germany, and Building Coordination Service with respect to scalability in an empirical setting.

4.3.1 Energy Efficiency The restaurant covers a total area of 251,50 m2 with a capacity of 200 sitting places. It is used for lunch from 11:30 a.m. until 2:00 p.m. and for work, meetings, and other events outside lunchtime. The restaurant environment is an open space divided in two sections by construction. Each section has 15 controllable lamps with either 38 or 18 W of power consumption each. We attached smart plugs to the lamps to make them controllable and collect their power consumption. We installed 15 more sensors, one to measure the natural light level, and the rest to detect people’s movement. To make a more meaningful use of the restaurant space, we divide each section into smaller areas, not necessarily of the same size. We embedded movement sensors in each area in positions that cover most of its space.

1677

0.06 0.04 0.00

0.02

Power (kWh)

0.08

Software Engineering Smart Energy Systems

00:00

00:55

01:50

02:45

03:40

04:35

05:30

06:25

07:20

08:15

09:10

10:05

11:00

11:55

12:50

13:45

14:40

15:35

16:30

17:25

18:20

19:15

20:10

21:05

22:00

22:55

23:50

Time of day manual control

movement sensors

automated control

Fig. 2 Comparison of daily average electricity consumption between using manual control, movement sensors, and our system to control lamps, from (Georgievski et al. 2017)

We conducted tests over the course of 5 weeks in the months of February, March, and April 2015, involving measurements from Monday to Sunday. In the Netherlands, these months are dark months, meaning that some of the worst possible conditions to reduce energy demand would be encountered. In calendar weeks 7 to 9, we recorded measures of energy consumption to understand the effect of manual control of lamps in the restaurant. In calendar weeks 13 and 14, we allowed for automated control of the environment by using our energy system. For the purpose of comparison, we simulate control of lamps based only on the use of movement sensors for the setting and period when automated control was run. We consider two types of activities, namely, presence and absence. We observed the lamps are usually turned on by the building cleaners at 6:30 a.m. and stay on until around 8 p.m. when they are switched off by the security personnel. The average consumption per working day in the restaurant is 14 kWh. In weekends, the lamps are always off. The use of our system results in intelligent management of electricity consumption with respect to the natural light and presence of people. Figure 2 shows the daily average electricity consumption when the lamps are controlled manually, when the lamps are triggered by movement sensors, and when our system is used in the restaurant. The average reduction of electricity consumption between the scenario of manual control and movement sensors is 71%, the average reduction between the scenario of manual control and the one with our system is 89%, and the average reduction between the scenario of movement sensors and our system is 61%. In addition, we noticed our system was able to deal with special events without any intervention occurring in the restaurant at evenings and during weekends. This demonstrates that our system makes the restaurant truly adaptable to the happenings within.

4.3.2 Accuracy We have implemented the Activity Recognition Service using two techniques. In the following, we describe the experiment and results of the implementation based on machine learning (Georgievski et al. 2019). For experiments and results of using ontological reasoning, we refer to Nguyen et al. (2014) and Georgievski et al. (2017).

1678

M. Aiello et al.

We used one large space of our own office building at the University of Stuttgart. The space covers a total area of 140 m2 and is divided into five sub-locations. The relevant locations for our experiment are one working space and a kitchen corner. The working space has a desk equipped with a laptop and a desk lamp. The kitchen has a toaster and a coffee machine. Smart plugs are attached to all the appliances to get their electricity consumption. The working space is equipped with sensors for movement detection, temperature and humidity, ultrasonic range, and sound. Another movement sensor is deployed in the kitchen. We conducted tests during working hours over 4 weeks in the months of January and February 2019. We collected 1440 samples on average per day with a sampling rate of 1 min and after preprocessing. We consider six types of activities: working with a laptop, working without a laptop, reading, drinking, eating, presence, and absence. The majority of samples belong to the absence activity, and 5806 samples represent the rest of the activities. We evaluated the technique using four metrics: accuracy, precision, recall, and F1 score. The accuracy score of activity recognition, which gives us the ratio between the number of correct predictions and total number of predictions made, is 0.9971. This high score is because the metric does not consider the total samples of each of the activity classes. In other words, given that the class of absence activity is bigger than the classes of other activities and is correctly predicted most of the times, the accuracy score is high even though the actual accuracy for other classes is worse. So, the other three metrics should give us more insights about accuracy. We use the precision score to understand whether the Activity Recognition Service only labels positive samples as positive. We use the recall score to understand the inability of the Activity Recognition Service to find all positive samples. Finally, we use the F1 score, which represents a weighted average of precision and recall. Table 2 shows the results of precision, recall, and F1 score of the Activity Recognition Service for each of the activity classes, and the last column indicates the total number of samples per activity class. It can be observed that recognizing occupant activities using long short-term memory neural networks (Hochreiter and Schmidhuber 1997) has almost perfect scores for all types of activities. Among all

Table 2 Evaluation results of activity recognition Absence Presence Working with computer Working without computer Reading Drinking Eating

Precision 1.00 0.91 0.99 1.00 1.00 1.00 1.00

Recall 1.00 0.96 1.00 0.95 1.00 1.00 1.00

F1 score 1.00 0.93 0.99 0.98 1.00 1.00 1.00

Support 7789 134 680 200 95 11 19

Software Engineering Smart Energy Systems

1679

activities, our approach only struggles when recognizing presence. Looking at the overall scores of activity recognition, the precision is 0.9849, the recall is 0.9880, and the F1 score is 0.9868. We also look at the cross entropy loss, which is the negative log-likelihood of true labels given the probabilities of predictions of the activity recognition. The cross entropy loss of activity recognition is 0.0110 (the lower the value, the better predictions are).

4.3.3 Scalability Our idea is to get insights about the performance of the Building Coordination Service when used in building environments larger and more complex than the restaurant (cf. NFR 3). We have, therefore, executed a set of performance tests on two synthetically generated sets of HTN planning problems using the domain model created for the restaurant. In the first set of tests, we evaluate the performance in terms of scalability of the number of tasks in the initial task network under a constant number of lamps per area. We are interested in the behavior of the service when the size of the building environment increases. Figure 3 shows the scaling of the Building Coordination Service. Figure 3a, b and c indicate the planning time when the number of lamps per area is fixed. In addition, moving from left to right charts, we increase the number of rooms too. Looking at Fig. 3a, it is difficult to assess the behavior of the planning system. From Fig. 3b and c, one can notice more or less a gradual rise of the planning time up until 70 simultaneous activities in Fig. 3b, and until 100 simultaneous activities in Fig. 3c where the number of rooms is higher. After these points, the planning time increases steeply where the worst case is just above 1.5 s. In the second set of tests, we evaluate the performance in terms of scalability of the number of lamps when the number of areas (and therefore simultaneous activities) is constant. Considering that lamps can be exclusively in one area or shared with other areas, this increases the difficulty of solving a planning problem. Thus, the number of predicates related to location properties increases, too. Figure 4 depicts the results. One can notice that the number of lamps and their relationships with the environment layout affect the performance of the service. When the number of lamps is 200, the planning time is almost the same in all three charts despite the fact that the number of rooms across which the decision is made is increased (from 2 through 5 to 8). The curves are gradually rising, and the service needs around 3 s to deal with 800 lamps. In summary, Figs. 3 and 4 show that the Building Coordination Service can successfully compute plans in just few milliseconds for large planning problems representing real-world situations. The experiments in which the Building Coordination Service takes up to 3 s to find a solution are much larger but unrealistic and to the extreme. In reality, there will be more than one instance of the service to cover such a space. Nevertheless, one instance of the service takes, for example, only 3 s to find out which lamps to turn on or off from a pool of 800 lamps.

1680

M. Aiello et al.

15

20

25

30

30

40

50

60

Number of activities

(a)

(b) 1400

Number of activities

70

Rooms=8, Lamps=7

200

600

1000

Planning time (ms)

10

200 400 600 800

150

200

Planning time (ms)

Rooms=5, Lamps=7

100

Planning time (ms)

Rooms=2, Lamps=7

40

60

80

100

120

Number of activities

(c)

1200 800

Rooms=5, Activities=5

0

400

Planning time (ms)

100

200

300

400

Rooms=2, Activities=5

0

Planning time (ms)

Fig. 3 Performance results of when scaling the number of activities per area, from (Georgievski et al. 2017)

100

150

200

100

200

300

400

Number of lamps

(a)

(b)

500

Rooms=8, Activities=5

0

1000

2000

3000

Number of lamps

Planning time (ms)

50

200

300

400

500

600

700

800

900

Number of lamps

(c) Fig. 4 Performance results when scaling the number of lamps per area, from (Georgievski et al. 2017)

Software Engineering Smart Energy Systems

5

1681

Case Study: Following Smart Grid Signals

Office buildings, or more generally, buildings are not systems that operate in isolation their energy needs. They rely and are interconnected on energy grids. While in the previous section we have studied a smart energy system for office buildings, we now consider how buildings can become active players of the energy infrastructure by coordinating with the grid. This is typically achieved by following signals coming from the grid and using them to optimize energy-relevant operations. Therefore, the planning problem that we have seen for office buildings now is extended with an extra dimension, that of variable costs for the provisioning of power. It is the main task of the Optimization Service to plan the use of available resources, such as generators and storage, as well as flexible loads, with the aim of optimizing economic and environmental objectives, while satisfying the energy demand and users’ comfort, based on signals coming from the grid. While current energy systems offer at the most demand response programs based on dynamic energy prices, future smart energy systems for buildings will be able to consider multiple dynamic signals, such as carbon emissions and prices, to reduce the costs and, at the same time, to decrease the environmental footprint of buildings. We present an approach that allows for optimal scheduling of flexible resources in buildings based on the objectives set by the users while satisfying users’ comfort preferences.

5.1

Optimal Operation Scheduling

Optimal operation scheduling is the short-term planning of available resources, such as generators and storage, with the aim of optimizing an objective function while covering the energy demand. When loads are flexible, they become part of the resources to be optimally scheduled. To generalize the operation scheduling problem within the energy context, we propose a general definition for the planning of energy resources to satisfy the load demand. Generally speaking, a scheduling problem consists of the allocation of resources to a set of requests over time. Formally, given a set D of requests to satisfy, a set K of resource types, and a discrete representation of time T, a time-discrete scheduling of typed resources to satisfy requests is a mapping s : D × T → K × R, which associates with each request d ∈ D and each time step t ∈ T the type and quantity of resource(s) required to satisfy the request. The scheduling problem consists of a set of variables X; a set of domain values V = {D, T , K, R} such that x ∈ V ; a set of constraints C that restricts the values

1682

M. Aiello et al.

that the variables can take. In particular, a constraint cj over a subset of variables Xj ⊂ X is a relation Rj (Xj ) on the corresponding subset of domains Vj ⊂ V . A feasible solution to a scheduling problem is an assignment to each variable in X such that every constraint in C is satisfied. We denote the set of all feasible solutions to a scheduling problem as I. A cost function f is a mapping f : I → R that associates with each feasible solution i ∈ I a cost value. The optimal cost function fopt of a scheduling instance is defined by fopt = min{f (i)|i ∈ I } and the set of optimal solutions to a scheduling problem is denoted by Iopt = {i ∈ I |f (i) = fopt }. Within the energy management context, D is the set of power demands to be satisfied, T is a finite set of ordered time steps, and K is the set of types of available resources, which include distribution grids (e.g., gas) and DERs. A solution to the derived scheduling problem is a set of pairs (type, quantity) of resources that satisfy a power demand di ∈ D at time step tj ∈ T .

5.1.1 Objectives The cost function to be optimized by the scheduling algorithm may represent a highlevel goal set by the users, e.g., minimization of daily energy costs or daily carbon emissions, as well as a technical goal set by the utilities, e.g., minimization of peak demand. According to our survey on energy management in buildings (Fiorini and Aiello 2019a), the most common approach is to formulate the optimal scheduling problem as an economic single-objective problem over a determined time horizon. The smart energy system finds the optimal solution to the problem that determines how much of a certain resource has to be supplied at each time step in order to satisfy the demand while minimizing the operation costs or maximizing the profit based on signals coming from the smart grid. Typical signals of price-based DRPs reflect the day-ahead, hour-ahead, or real-time energy market prices, enabling users to reduce their energy bills by proactively modifying their normal consumption pattern. In Fiorini and Aiello (2020), we propose a multi-objective optimal scheduling approach based on both dynamic prices and dynamic carbon intensities. Carbonbased signals may be based either on the average emission intensity of the composition of the generation mix, or on the emission intensity of the power plant that supplies the marginal changes in the load demand. Regardless of the nature of the signals, such an approach satisfies the functional requirement FR 3, and promotes more efficient and sustainable energy systems, satisfying the functional requirement FR 5 and the user-related requirement UR 5. 5.1.2 Constraints The optimal scheduling problem is subject to several constraints that guarantee the correct operation of the energy system. For the sake of conciseness, we mention a couple of the most significant ones, while for a more detailed formulation, we refer to Fiorini and Aiello (2018, 2019b, 2020). The balance between supply and demand must be guaranteed at any time; if there is a shortage of supply, the smart energy system may discharge a storage device, if any, or curtail as much load as necessary to balance the system. If there is a surplus of supply, the smart energy system may

Software Engineering Smart Energy Systems

1683

charge a storage device, if any, turn on a flexible load, or export it to the main grid. This constraint is related to FR 5. The smart energy system monitors the current status of the controllable devices and knows their technical characteristics, such as the current, minimum, and maximum values of the power output of a cogeneration system or of the level of energy stored into the battery of an electric vehicle. Such technical characteristics are translated into constraints of the scheduling problem and the smart energy system makes decisions about the future status of the devices accordingly, as required by FR 1. The smart energy system cannot affect the comfort of the user, as stated in UR 4. User’s preferences with respect to, for instance, acceptable indoor temperature range and allowed delay of flexible appliances, are translated into additional constraints that the optimal solution to the scheduling problem must satisfy. The optimal scheduling problem is subject to several uncertainties owning to, for instance, changes in weather conditions, renewable production, demand, electricity emission factors, and electricity prices. To be able to ensure balancing while optimizing the objectives and satisfying user’s preferences, the smart energy system has to promptly correct the operating point of flexible units as new information on uncertain parameters becomes available. A rolling horizon or model predictive control approach, the terms are often used interchangeably within the literature on operation scheduling. Approximates a long-horizon optimal scheduling problem with a series of sequential short-horizon problems. At each time step, the algorithm estimates the future behavior of the system based on current forecasts and determines its optimal state. Once new forecasts are available, the procedure is repeated. In other words, the smart energy system addresses the forecast uncertainties by sequentially making short-term decisions and taking corrective actions based on new short-term forecasts, thus satisfying FR 4.

5.2

Cost-Efficiency and Sustainability

We investigate the benefits of our multi-objective scheduling for smart energy buildings by evaluating the potential savings in CO2 emissions and energy costs of a group of 300 German smart homes. The smart homes are modeled realistically by considering size, season, having up to six common appliances, and varying thermal and electric loads. We use statistical data on ownership probabilities and usage of the most common appliances in German households (Destatis 2018; Mauser et al. 2016), historical data on electricity consumption (Open Power System Data 2017), weather conditions in Stuttgart (SOLCAST 2020), and German electricity prices and carbon emissions (ENTSO-E 2020; Tomorrow 2020). We propose a smart home model that couples multiple energy vectors, namely, electricity, natural gas, and hot water, hence considering the concept of smart energy system in its more general multi-energy variant, where space heating and domestic hot water demands, as well as some appliances can be supplied by multiple

1684

M. Aiello et al.

technologies or energy carriers, either in parallel or sequentially. Therefore, the management of electric and thermal energy consumption becomes interdependent. Smart homes equipped with solar panels can manage their loads so that they minimize the import of energy from the main grid and, consequently, the daily carbon emissions. On the other hand, they can sell the solar energy to the main grid. We estimate the average daily costs and emissions in two cases, namely, when the smart home energy system follows dynamic price signals and when it follows dynamic CO2 signals. In the former case, the average daily costs for a smart home varies between approximately 0.85 e and 1.95 e, while the daily carbon emissions varies between 8.8 and 9.4 kg, depending on the type of heating system and the available appliances. In the latter case, the average daily costs increase to approximately 1.6–2.1 e, i.e., up to +88%, while the daily emissions drop to 6.9–9.2 kg, i.e., −22%. When the economic and the environmental goals are equally important, the optimal solution to the multi-objective scheduling problem corresponds to 7.1–9.3 kg of CO2 and 0.95–1.95 e per day for an average smart home, depending on the available technologies (Fiorini and Aiello 2020). The results show that a smart energy system following a combination of the two signals finds a trade-off between minimizing the costs and minimizing the carbon emissions, thus enabling cost savings and, at the same time, improving the sustainability of the home. Similar reasoning can be applied to other types of non-residential buildings, such as office buildings, universities, and hospitals.

6

Conclusions

The energy sector is experiencing a fast-paced transformation. A mostly hierarchical, passive system is becoming more complex, dynamic, and distributed. Information technology is playing a central role in enabling this transformation and an equally important role in managing it. Smart energy systems are emerging as energy systems that collect large amounts of data which is used for their automation and to increase their efficiency. These software systems have specific requirements and associated engineering patterns and techniques to satisfy them. In the present chapter, we have identified 16 core software-engineering requirements of smart energy systems and, based on 10 years of research and development in the field, we proposed a generic blueprint for smart energy systems. We have illustrated these design principles on two case studies: smart office buildings and residential buildings following grid signals. The emergence of smart energy systems will have an essential role in increasing the sustainability of the energy sector. Mostly this will happen following the adage “the most sustainable kWh is the one that is saved.” But this is not all: much more control will be shifted at the periphery of our energy systems. Users will be able to express their goals and preferences in terms of energy management and automation; then systems will act on their behalf controlling their homes, offices, cars, and negotiating with the smart grid directly. For this to happen, regulations must follow technological innovation, and solutions to privacy and security concerns must be

Software Engineering Smart Energy Systems

1685

available. Data science, machine learning, optimization, knowledge representation, reasoning, planning, and scheduling complete the areas that are enabling and will continue to support the current transformation in the energy sector.

7

Cross-References

Application of Machine Learning in Occupant and Indoor Environment Behavior

Modeling: Sensors, Methods, and Algorithms Data Analytics Applications in the Energy Systems Concerning Sustainability Data-Driven Techniques for Optimizing the Renewable Energy Systems Opera-

tions Design and Operational Strategies for Grid-Connected Smart Home Economical and Reliable Design of a Hybrid Energy System in a Smart Grid

Network Energy Management of Smart Homes by Optimizing Energy Consumption

Scheduling Machine Learning for Building Energy Modeling Smart Buildings in the IoT Era: Necessity, Challenges, and Opportunities

References I. Asimov, P.S. Warrick, M.H. Greenberg, Machines That Think: The Best Science Fiction Stories About Robots and Computers (H. Holt & Co, New York, 1984) C. Bettini, O. Brdiczka, K. Henricksen, J. Indulska, D. Nicklas, A. Ranganathan, D. Riboni, A survey of context modelling and reasoning techniques. Pervasive Mob. Comput. 6(2), 161–180 (2010) G. Booch, J. Rumbaugh, I. Jacobson, The Unified Modeling Language User Guide, 2nd edn. (Addison-Wesley Professional, Boston, 2005) P. Bourque, R.E. Fairley, I.C. Society, Guide to the Software Engineering Body of Knowledge (SWEBOK(R)): Version 3.0, 3rd edn. (IEEE Computer Society Press, Piscataway, 2014) A. Cesta, G. Cortellessa, S. Fratini, A. Oddi, N. Policella, An innovative product for space mission planning: an a posteriori evaluation, in International Conference on Automated Planning and Scheduling (2007), pp. 57–64 R. Chinnici, J.-J. Moreau, A. Ryman, S. Weerawarana, Web Services Description Language (WSDL) Version 2.0 Part 1: Core Language (2007) Destatis, Wirtschaftsrechnungen – Einkommens- und Verbrauchsstichprobe Ausstattung privater Haushalte mit ausgewählten Gebrauchsgütern und Versicherungen. Technical Report 1 (2018) ENTSO-E, ENTSO-E Transparency Platform (2020) T. Erl, SOA Principles of Service Design (Prentice Hall PTR, 2007) European Committee for Standardization, Light and Lighting – Lighting of Work Places – Part 1: Indoor Work Places. European standard, Official Journal of the European Union (2011) L. Fiorini, M. Aiello, Household CO2 -efficient energy management. Energy Inform. 1(Suppl 1), 21–34 (2018) L. Fiorini, M. Aiello, Energy management for user’s thermal and power needs: a survey. Energy Rep. 5, 1048–1076 (2019a) L. Fiorini, M. Aiello, Predictive CO2 -efficient scheduling of hybrid electric and thermal loads, in 2019 IEEE International Conference on Energy Internet (ICEI), Nanjing (IEEE, 2019b)

1686

M. Aiello et al.

L. Fiorini, M. Aiello, Predictive multi-objective scheduling with dynamic prices and marginal CO2 -emission intensities, in 11th ACM International Conference on Future Energy Systems (e-Energy’20) (2020), pp. 196–207 L. Fiorini, L. Steg, M. Aiello, Sustainability choices when cooking pasta, in The 11th ACM International Conference on Future Energy Systems (e-Energy’20) (2020), pp. 161–166 L. Gelazanskas, K.A. Gamage Demand side management in smart grid: a review and proposals for future direction. Sustain. Cities Soc. 11, 22–30 (2014) I. Georgievski, HPDL: Hierarchical Planning Definition Language. JBI Preprint 2013-12-3, University of Groningen (2013) I. Georgievski, Coordinating services embedded everywhere via hierarchical planning. PhD thesis, University of Groningen (2015) I. Georgievski, M. Aiello, HTN planning: overview, comparison, and beyond. Artif. Intell. 222, 124–156 (2015) I. Georgievski, M. Aiello, Automated planning for ubiquitous computing. ACM Comput. Surv. 49(4), 63:1–63:46 (2016) I. Georgievski, V. Degeler, G.A. Pagani, T.A. Nguyen, A. Lazovik, M. Aiello, Optimizing energy costs for offices connected to the smart grid. IEEE Trans. Smart Grid 3(4), 2273–2285 (2012) I. Georgievski, T.A. Nguyen, F. Nizamic, B. Setz, A. Lazovik, M. Aiello, Planning meets activity recognition: service coordination for intelligent buildings. Pervasive Mob. Comput. 38(1), 110– 139 (2017) I. Georgievski, P. Gupta, M. Aiello, Activity learning for intelligent buildings, in IEEE Annual Ubiquitous Computing, Electronics and Mobile Communication Conference, UEMCON 2019 (2019), pp. 916–923 I. Georgievski, L. Fiorini, M. Aiello, Towards service-oriented and intelligent microgrids, in International Conference on Applications of Intelligent Systems, APPIS 2020 (2020), pp. 1–6 M. Ghallab, D.S. Nau, P. Traverso, Automated Planning: Theory and Practice (Morgan Kaufmann Publishers Inc, Burlington, Massachusetts, 2004) S. Goy, A. Sancho-Tomás, Load management in buildings, in Urban Energy Systems for LowCarbon Cities, ed. by U. Eicker (Academic publishing, Cambridge, Massachusetts, 2019), pp. 137–179 S. Hochreiter, J. Schmidhuber, Long short-term memory. Neural Comput. 9(8), 1735–1780 (1997) E. Karpas, D. Magazzeni, Automated planning for robotics. Ann. Rev. Control Robot. Auton. Syst. 3(1), 417–439 (2020) J.P. Kelly, A. Botea, S. Koenig et al., Offline planning with hierarchical task networks in video games, in Artificial Intelligence for Interactive Digital Entertainment Conference (2008), pp. 60–65 A. Lazovik, M. Aiello, M. Papazoglou, Planning and monitoring the execution of Web service requests. J. Digit. Libr. 6(3), 235–246 (2006) E.-K. Lee, W. Shi, R. Gadh, W. Kim, Design and implementation of a microgrid energy management system. Sustainability 8(11), 1143 (2016). https://doi.org/10.3390/su8111143 H. Lund, P. Østergaard, D. Connolly, B. Mathiesen, Smart energy and smart energy systems. Energy 137(C), 556–565 (2017) B.V. Mathiesen, H. Lund, D. Connolly, H. Wenzel, P.A. Østergaard, B. Möller, S. Nielsen, I. Ridjan, P. Karnøe, K. Sperling et al., Smart energy systems for coherent 100% renewable energy and transport solutions. Appl. Energy 145, 139–154 (2015) I. Mauser, J. Müller, F. Allerding, H. Schmeck, Adaptive building energy management with multiple commodities and flexible evolutionary optimization. Renew. Energy 87, 911–921 (2016) T.A. Nguyen, M. Aiello Energy intelligent buildings based on user activity: a survey. Energy Build. 56, 244–257 (2013) T.A. Nguyen, A. Raspitzu, M. Aiello, Ontology-based office activity recognition with applications for energy savings. J. Ambient Intell. Humaniz. Comput. 5(5), 667–681 (2014) Open Power System Data, Data Package Household Data. Version 10 Nov 2017 (2017)

Software Engineering Smart Energy Systems

1687

G.A. Pagani, M. Aiello, Towards decentralized trading: a topological investigation of the medium and low voltage grids. IEEE Trans. Smart Grid 3(2), 538–547 (2011) J. Page, D. Robinson, N. Morel, J.-L. Scartezzini, A generalised stochastic model for the simulation of occupant presence. Energy Build. 40(2), 83–98 (2008) M.P. Papazoglou, D. Georgakopoulos, Introduction: service-oriented computing. Commun. ACM 46(10), 24–28 (2003) R.S. Pressman, B.R. Maxim, Software Engineering: A Practitioner’s Approach, 8th edn. (McGraw-Hill, Inc, New York, 2015) J. Robie, R. Cavicchio, R. Sinnema, E. Wilde, RESTful Service Description Language (RSDL), Describing RESTful services without tight coupling, in Balisage: The Markup Conference (2013), pp. 6–9 P. Siano, Demand response and smart grids – a survey. Renew. Sustain. Energy Rev. 30, 461–478 (2014) M.P. Singh, M.N. Huhns, Service-Oriented Computing: Semantics, Processes, Agents (John Wiley Sons Ltd., Hoboken, 2005) SOLCAST, Global solar irradiance data and PV system power output data (2020) I. Sommerville, Software Engineering, 9th edn. (Addison-Wesley Professional, Boston, 2010) G. Strbac, Demand side management: benefits and challenges. Energy Policy 36(12), 4419–4426 (2008) A. Tate, B. Drabble, J. Dalton, O-Plan: a knowledge-based planner and its application to logistics, in Advanced Planning Technology: Technological Achievements of the ARPA/Rome Laboratory Planning Initiative (1996), pp. 259–266 Tomorrow, electricityMap (2020) M. Vallés, P. Frías, T.G. Gómez, Regulatory and market framework analysis: a working document assessing the impact of the regulatory and market framework on the IndustRE business models. IndustRE deliverable 2.2 (2015) A. Videla, J.J. Williams, RabbitMQ in Action (Manning Publications, Shelter Island, 2012)

An Intelligent Decision Support System for an Integrated Energy Aware Production-Distribution Model Soulmaz Rahman Mohammadpour and Seyed Habib A. Rahmati

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Mathematical Model and Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Solution Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 GA and BBO Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Solution Representation and Decoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Migration and Crossover Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Mutation Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Hypothesis Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Discussion and Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1690 1693 1698 1698 1699 1699 1699 1700 1704 1704 1707 1707 1707

Abstract

Nowadays, the green aspect of sustainable development is being a global requirement. Words such as CO2 emission, zero waste, or energy consumption are repeated day in day out. Besides, intelligent or smart system development is also the chief concept of modern life. Meanwhile, different sources of pollution are driving the world into global warming and perhaps production and logistic systems are foremost among them. Developing intelligent systems alongside the global prohibiting laws and green altering or culture-making campaigns are the future expected ways of coping with this serious obstacle of humankind.

S. Rahman Mohammadpour · S. H. A. Rahmati () Department of Industrial Engineering, Faculty of Industrial and Mechanical Engineering, Qazvin Branch, Islamic Azad University, Qazvin, Iran e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_77

1689

1690

S. Rahman Mohammadpour and S. H. A. Rahmati

This chapter focuses on introducing an intelligent decision support system (DSS) for controlling affairs of a complex production-distribution system (PDS) in which reducing CO2 emission and enhancing company economical profit are in conflict. Both aspects of manufacturing and vehicle logistic pollutions are modeled according to real coefficients and calculations of CO2 emission per electricity consumption (kg/kWh) and CO2 emission per fuel consumption (kg/L), respectively. The introduced smart DSS provides PDS owners with an ifthen system in which they are capable of assessing different scenarios of planning to improve their environmental responsibility and profit simultaneously. These systems implement natural inspired intelligent modules to suggest solutions for different decisions of system owners in minimum time and without the need to real case try and error. Various analytical and experimental results are also provided to introduce the performance of the DSS and PDS systems more explicitly. Keywords

Smart energy scheduling · Decision support system · Production-distribution · Intelligent optimization · CO2 emission reduction

1

Introduction

Today, researchers have a concern regarding the greenhouse effects. Humans have been artificially raising the concentration of greenhouse gases in the atmosphere at an ever-increasing rate by burning fossil fuels. One of the largest uses of fossil fuels is in the production of electricity energy. On the other hand, over 65% of the world’s electrical energy used today is generated by steam turbine generators burning fossil fuels. For example in the United States, about 65% of total electricity generation in 2018 was produced from fossil fuels (coal, natural gas, and petroleum) (Pasandidehpoor et al. 2022). In addition, the world’s population continues to soar; energy demand is growing at a dramatic pace. As a result, reducing greenhouse gas emissions, especially reducing CO2 emissions, is one of the most important goals of researchers. In general, energy consumption can be categorized into four main sectors, including industrial consumption, transportation consumption, commercial consumption, and residential consumption. The industrial sector as the first rank of energy consumption uses the delivered energy more than other end-use sectors and consumes about 54% of the total delivered energy in the world (Nokhbeh Dehghan et al. 2022); then the second rank is transportation. Therefore, optimizing energy consumption in the production and transportation levels in the industry has particular importance. Accordingly, the subject of this study is reducing costs, energy consumption, and CO2 emission by providing a novel green production-distribution model. The

An Intelligent Decision Support System for an Integrated Energy Aware. . .

1691

novelty of this study consists of considering parallel machine scheduling production (PMSP) systems as production model, presenting new model for estimating cost of production energy consumption, as well as distribution cost from manufacturer to distribution center (DC) and DC to customer. Finally, this study by using genetic algorithm (GA) and biogeography-based optimization algorithms (BBO) provides a practical methodology for solving the problem. In addition, the relation of cost, energy consumption, and CO2 emission will be discussed. The rest of the section reviews the literature and presents the chief gap filling contributions of our work. Harris et al. (2011) evaluated the effect of traditional cost optimization approach on overall logistics costs and CO2 emissions in the supply chain. Meng and Niu (2011) modeled the CO2 emission through fossil fuel combustion in the industries transportation process. Ulku (2012) proposed a strategic solution to mixed different cargoes in one truck and manage the transportation cost and energy. Liotta et al. (2015) present a production and transportation model with the subject of reducing production costs, transportation costs, as well as reducing carbon dioxide emissions in the transportation sector. Mariano et al. (2016) assessed the relation between transport logistics performance and CO2 emissions from the transport sector. They used slacks-based measure of data envelopment analysis to introduce low carbon logistics performance index. Yao et al. (2018) analyzed green production performance (GPP) in China. This study examined the effects of implementing a new approach aimed at saving energy and reducing emissions in GPP. Zhang et al. (2018) presented an optimization model to reduce the total cost and raise the total profit of green transportation. Cai et al. (2019) proposed a new concept entitled “Energy Saving and Emission Reduction” is presented as a way to improve energy efficiency and reduce waste emissions more effectively. Wang et al. (2020) studied the direct relationship between the transportation and the amount of CO2 emissions in different countries. This study highlighted the effects of carbon and transport intensity on CO2 emissions, while factors of economic structure and population size help increase CO2 emissions. Babagolzadeh et al. (2020) provided a new model to determine re-optimization policies and transportation schedules to minimize operating costs and CO2 emissions. Table 1 investigates the literature from two perspectives of studies that used coefficient energy consumption in their estimations or have introduced a specific model for energy consumption. Based on the studies analysis, due to the complexity of presenting the model, most studies have used the coefficient as energy consumption. Therefore, the calculation of energy consumption in the production sector involves a more complex process compared to transportation; for this reason, research on optimization of energy consumption in manufacturing has a smaller portion than transportation. In addition, due to the greater complexity of calculating energy consumption in production compared to logistics, most of the studies chose the transportation sector to optimize the energy consumption in their supply chain management. On the other hand, transportation as one of the main causes of CO2 emissions is one of the attractive and important issues to research for improving

Year

2011 2011 2012 2015 2016 2018 2018 2019 2020 2020

Author(s)

Harris et al. (2011) Meng and Niu (2011) Ulku (2012) Liotta et al. (2015) Mariano et al. (2016) Yao et al. (2018) Zhang et al. (2018) Cai et al. (2019) Wang et al. (2020) Babagolzadeh et al. (2020) Our study

*

* * * * *

*

Energy consumption coefficient

*

* *

* *

Mathematical energy model

Table 1 Comparison of the contributions of different authors

*

*

*

Metaheuristic algorithms

*

*

Supply level Production Distribution * * * * * * * * * * * *

*

Customer

*

*

Green part Production Logistics * * * * * * * * * *

1692 S. Rahman Mohammadpour and S. H. A. Rahmati

An Intelligent Decision Support System for an Integrated Energy Aware. . .

1693

Fig. 1 Concentration of production-distribution problem

logistics performance and reducing environmental impact. The share of studies in the field of logistics in Table 1 approves the validity of this claim. Another topic that has been studied in the literature review is the use of meta-heuristic algorithms to solve the models presented in the studies. According to the assessments, only a limited number of articles have used meta-heuristic algorithms to solve their model. As mentioned in the literature review in Table 1 and Fig. 1, this study presents a novel production-distribution model on the subject of total optimization, reducing total costs, and reducing total energy consumption. Moreover, using metaheuristic algorithms for solving the model has a complexity that just a few researches used these algorithms, but this study solves the model by two metaheuristic algorithms that included GA and BBO algorithms. The rest of the study is organized as follows. Section 2 presents the mathematical model. Section 3 describes the development of GA and BBO algorithms for solving the developed model. It presents the solution structure and neighborhood structures alongside of the main flowcharts of the algorithms. Section 4 concentrates on computational results of GA and BBO and provides various statistical and graphical analysis. It also discusses the findings and implications of the chapter. Section 5 concludes the chapter.

2

Mathematical Model and Assumptions

The model presented in this study consists of two parts including productiondistribution and energy. The following nomenclature shows assumptions, which is used in this chapter.

1694

S. Rahman Mohammadpour and S. H. A. Rahmati

Nomenclature Part 1. Production-Distribution Decision variables Order quantity which produced by manufacturer m for distributor d Amount of sent order from distributor d to customer c If one job i is done after another job j on machine k by manufacturer m = 1; Otherwise 0

ℎ

If a particular machine k does a specific job j by the manufacturer m = 1; Otherwise 0 If a machine k is selected by the manufacturer m = 1; Otherwise 0 Earliness of job i by manufacturer m Tardiness of job i by manufacturer m Completion time of job i by manufacturer m

Parameters c: Customer number, c = 1, 2, …, C d: Distributer number, d = 1, 2, …, D m: Manufacturer number, m = 1, 2, …, M

: Amount of customer demand c Cost of sending each order unit produced by the manufacturer m to the distributor d Cost of sending each order unit from distributor d to customer c Capacity of Distributor d Production capacity of manufacturer m Total cost of production for the manufacturer m

i, j: Number of jobs on machines, i, j = 0, 1, 2, …, N

k: Machine number, k = 1, 2, …, K

: Setup time to switch from job i to job j on machine k for manufacturer m

: Process time of job i on machine k for manufacturer m, (i = 0, 1, 2, …, N, 1, 2, …, K, m = 1, 2, …, M) : Due date of job i for manufacturer m : Tardiness penalty of job i for manufacturer m : Cost of machine k for manufacturer m, (1, 2, …, K, m = 1, 2, …, M) Large number

This section develops production-distribution model. Moreover, this research provides an innovative green approach in the field of the PMSP model by considering energy consumption at the manufacturing and also presents model for energy consumption at the distribution sector. Perhaps, one of the chief contribution of this study is green production modeling with a novel practical energy consumption model. This energy optimization leads to lower consumption of limited resources of the world and reduces CO2 emission. Moreover, in the practical energy consumption formula of the study, a sinus coefficient controls the energy cost according to the region of the placement of manufacturers. Generally, this concerned study properly provides a sustainable supply chain model because it is capable to minimize total production-distribution cost (as economical concern), reduces energy consumption and CO2 emission (as environmental concern), and creates social justice for energy payments by considering the place of production (as social concern). Finally, places

An Intelligent Decision Support System for an Integrated Energy Aware. . .

1695

Nomenclature Part 2. Energy

Production

Distribution

Parameters

Parameters : Voltage of manufacturer m

t : Truck number, t = 1, 2, …, T

: Power of manufacturer m

α = 0.2 : Truck Consumption per liter (L/km)

: Ampere of manufacturer m

μ = 1.46 Unit fuel cost ($/L)

(Sinφ)

Coefficient location of manufacturer m

: Distance between manufacturer m to distributer d

: Total Cost of Energy for manufacturer m : Distance between distributer d to customer c : Total Cost of Energy for truck t

with lower suitability index of living are more valuable for developing companies since their energy cost coefficient are considerably lower. In the long term, low developed regions around these companies can take many social benefits. The rest of this section introduced the model as follows: Min T C =

M m

T P Cm +

k k=1

+ T CE m + T DE t +

βkm bkm +

M D m

d

N i=1

(eim EJ im + tim T J im )

SP md QO md +

D C d

c

SP dc OS dc (1)

In this proposed model, Eq. (1) is the objective function consisting total production cost, total earliness and tardiness production cost, total production energy cost, total transportation energy cost, and total distribution cost. Constraints and their explanation are discussed as follows. T P Cm =

D d

QOmd ∗ P C m

∀m

(2)

Equation (2) represents the total production costs of each unit. Equation (3) shows the total cost of energy which consists of energy cost, voltage as a specific number (220 volts), ampere, and a location of the manufacturer. T CE m =

√ 3 ∗ Cost m ∗ Vm ∗ Im ∗ (Sinϕ)m

m = 1, . . . , M

(3)

Equation (4) shows the amount of customer demand coverage by distributors.

1696

S. Rahman Mohammadpour and S. H. A. Rahmati

D d

OS dc ≥ AD c ;

c = 1, 2, . . . , C

(4)

Equation (5) models the volume of production by considering the maximum capacity of the distributor. C c

M m

OS dc ≥ DC d ;

QO md =

D d

C c

d = 1, 2, . . . , D

OS dc ;

QO md ≤ P C m ;

d = 1, 2, . . . , D

m = 1, 2, . . . , M

(5)

(6)

(7)

Equation (6) creates a balance between production and distribution with the producer and consumer. Equation (7) shows the production capacity. Equation (8) calculates the amount of earliness and tardiness of job time based on the completion time. CJ im + EJ im − T J im = dim

i = 1, . . . , N; m = 1, . . . , M

(8)

Equation (9) ensures that a job should be allocated to only one machine, independent of the previous job. N K gij km = 1 i=0 k=1 i=j

j = 1, . . . , N ; m = 1, . . . , M

(9)

Equation (10) shows the prerequisite for job j and allocating it to machine k so that job j must be done after job 0. N i = 0 gij km = hj km i = j

j = 1, . . . , N; k = 1, . . . , K; m = 1, . . . , M

(10)

N j =1 gij km ≤ hj km j =i

i = 1, . . . , N ; k = 1, . . . , K; m = 1, . . . , M

(11)

K k=1

hj km =1 i=1, . . . , N ; m=1, . . . , M

(12)

Equation (11) represents that if job j is allocated to machine k, at least one job will be done immediately. Equation (12) ensures that each job is allocated to only one machine. Equation (13) ensures that if no machine is selected, nothing should be done on it.

An Intelligent Decision Support System for an Integrated Energy Aware. . .

N j =1

g0j km ≤ bkm

k = 1, . . . , K; m = 1, . . . , M

1697

(13)

Equation (14) ensures that if a machine is chosen, at least one job must be allocated to it. N j =1

g0j km > bkm − 1

k = 1, . . . , K; m = 1, . . . , M

(14)

Equation (15) represents the total cost of energy transportation for each truck between manufacturer to distributer and distributer to customer. T DE t = (Dmd ∗ 2 ∗ α ∗ μ) + (Ddc ∗ 2 ∗ α ∗ μ)

(15)

Equations (16) and (17) create the relationship between the completion time of jobs i and j concerning the start-up time as long as both jobs are allocated to one device Cj m − Cim + Z 1 − gij km ≥ Pj km + Wij km

i = 1, . . . , N ;

j = 1, . . . , N, i = j ; m = 1, . . . , M Cim ≥

K k=1

(W0ikm + Pikm ) × hikm

i = 1, . . . , N ;

(16)

(17)

k = 1, . . . , K; m = 1, . . . , M Equations (18) and (19) define our variable types hj km , gij km , bkm ∈ {0, 1}

i = 1, . . . , N ; j = 1, . . . , N ; k = 1, . . . , K;

m = 1, . . . , M; i = j (18) Cim , EJ im , T J im ≥ 0

i = 1, . . . , N; m = 1, . . . , M

(19)

Please notice that in Eq. (3) TCEm is a function of consumed ampere as Eq. (20). Moreover, according to Eq. (21) the consumed ampere is a function of time, and finally according to Eq. (22) ampere is a function of complementation time (Cmax ). T CE m = f (Im )

(20)

Im = f (t) = f (Cmax )

(21)

Im = a (Cmax )

(22)

1698

S. Rahman Mohammadpour and S. H. A. Rahmati

Table 2 Definition of vehicle typical parameters Notation μ α η

Description Unit fuel cost ($/L) Truck consumption per liter CO2 emitted by unit fuel consumption (kg/L)

Typical value 1.46 0.2 2.669

ƛ

Unit CO2 emissions price ($/kg) 1 Pound (£)

0.44 1.79 AUD dollars

$

Source: Cachon (2014), Demir et al. (2012), Koç et al. (2014), Soysal et al. (2015), Cheng et al. (2017), and Babagolzadeh et al. (2020)

Additionally another noticeable point is regarding the calculation of Eq. (15) that it used the coefficients from Table 2 to provide the transportation model. Moreover, taking into account the slope, speed, type of truck, and other effective factors, this study considered α = 20 km/L as the coefficient of energy consumption of each truck. Furthermore, it considered the round trip to the destination for obtaining total transportation cost. TDEt = (Dmd (km) ∗ 2 ∗ α (L/km) ∗ μ (AU $/L) ∗ t) + (Ddc (km) ∗ 2 ∗ α (L/km) ∗ μ (AU $/L) ∗ t)

3

Solution Methodology

In this section, the model is developed by the two solution methodologies consisting of meta-heuristic methods such as GA and BBO algorithms. These algorithms optimize the developed model (Rahmati and Zandieh 2011).

3.1

GA and BBO Algorithms

GA is an effective meta-heuristic for solving combinatorial optimization problems. GA is a special type of evolutionary algorithm that uses evolutionary concepts such as inheritance, biological mutation, and Darwin’s principles. The modeling of this algorithm is a programming technique that uses genetic evolution process. The problem to be solved has inputs that are converted into solutions during a modeled process of genetic evolution, and then the solutions are evaluated as candidates by the fitness function, and the algorithm terminates if the exit condition is met. In general, it is an iterative algorithm and most of its parts are selected as random processes. This algorithm is based on population, and each chromosome is known as a single chromosome and has its own level of fitness. The operators used include migration, mutation, and reproduction operators. A good solution is characterized

An Intelligent Decision Support System for an Integrated Energy Aware. . .

1699

by high fitness. In addition, GA operators can discard initial individuals during iterations (Rahmati and Zandieh 2011). BBO algorithm is another optimization method that is based on the concept of biogeography-based migration. Generally, biogeography studies the behavior of different biological species over time and space. The algorithm is population-based, and each habitat is identified as a single solution and has its own habitat suitability index (HSI). The operators used include migration and mutation operators. This algorithm does not have a reproduction operator but control the exploitation operator by its own migration process. In this algorithm, the habitat or solution with more HIS is more suitable. Moreover, initial population is not discarded during iterations but it is modified (Rahmati and Zandieh 2011). Both algorithms start with a randomly generated initial population.

3.2

Solution Representation and Decoding

According to the problem structure, Fig. 2 provides a combination of two vectors for production and distribution. The first vector shows the amount of manufacturer production for distributors. The second vector demonstrates the amount of distribution for each distributor to customers. The third vector represents the information on production units.

3.3

Migration and Crossover Operator

The crossover consists of two sections of production and distribution. Figure 3 represents the crossover in production and Fig. 4 shows the crossover in distribution. In this study, there are two types of vectors to represent the responses (sequence vector and allocation vector) that each of them has a specific crossover operator.

3.4

Mutation Operator

The mutation also consists of two sections including production and distribution. Figure 5 represents the mutation in production and Fig. 6 shows the mutation in distribution.

Fig. 2 Solution structures related to production-distribution in meta-heuristic algorithms

1700

S. Rahman Mohammadpour and S. H. A. Rahmati

Fig. 3 Feasible crossover/migration neighborhood structure for the production part

4

Results and Discussion

This section presents the outputs and improve the accuracy of the study; we created ten problems with different sizes as shown in Table 3. In the second stage, in Minitab software we got the required Taguchi design; from ten selected problems, we chose the fifth problem as the basic problem to get the best answer. In the third stage, according to the design that is obtained from the Taguchi method, we entered the parameters of a number of repetitions, population size, internal coefficient, and percentage of reduction on the MATLAB program. Then, we analyzed the output of the designs in two sections of best costs and time. Finally, according to the best answer, we investigated the results of the ten problems. Table 3 not only represents the problem attributes, but also it shows the final outputs of GA and BBO algorithm. Figure 7 represents a comparison of the best costs between GA and BBO results. It shows the close results of GA and BBO on best cost. Of course, GA’s best cost is a little bit better than BBO.

An Intelligent Decision Support System for an Integrated Energy Aware. . . Fig. 4 The structure of crossover and migration by uniform operator in distribution vector

Fig. 5 Feasible mutation neighborhood structure for the production part

Fig. 6 The reversion mutation in the response structure of distribution part

1701

Number of plant

1 1 1 2 2 2 2 3 3 3

#

1 2 3 4 5 6 7 8 9 10

3 5 5 3 5 5 10 3 5 10

Number of distributer

30 50 100 50 100 500 200 100 200 1000

Number of customers 10 11 12 13 13 14 14 17 20 23

Number of works

Table 3 Final output of ten problems by GA method

3 5 6 6 6 7 7 8 9 9

Number of machines GA Best cost 166,492,902.7246 170,134,028.8531 401,646,202.431 176,205,121.1153 418,445,947.8546 1,525,066,666.1219 743,243,906.7019 386,258,092.3772 790,793,067.8466 4,597,767,023.874 Time 22.42 23.06 25.19 24.24 27.77 37.76 32.48 29.41 29.14 79.67

BBO Best cost 166,477,784.8425 170,103,271.4331 480,609,929.9538 176,171,669.3001 418,370,005.8594 1,989,536,823.9641 733,159,694.6092 426,165,462.4721 965,706,318.0655 4,890,081,686.8002

Time 14.74 16.64 18.01 16.14 19.07 33.51 23.54 18.83 23.2 75.06

1702 S. Rahman Mohammadpour and S. H. A. Rahmati

An Intelligent Decision Support System for an Integrated Energy Aware. . .

1703

6E+09 5E+09

Best Cost

4E+09 3E+09 2E+09 1E+09 0 1

2

3

4

5

6

7

8

9

10

Problem Number Best Cost (GA)

Best Cost (BBO)

Fig. 7 Comparison between best cost (GA) and best cost (BBO) results for ten problems

90 80 70 60

Time

50 40 30 20 10 0 1

2

3

4

5

6

7

8

9

10

Problem Number Time (GA)

Time (BBO)

Fig. 8 Comparison between time (GA) and time (BBO) results for ten problems

Figure 8 represents a comparison of CPU time for obtaining the results of GA and BBO. It demonstrates the time consumption for gaining the best cost in most problems by the BBO algorithm is lower than the GA. By considering the problem data, it will be clear that the number of customers has a direct impact on time consumption. Whenever the customer number increases, the time consumption will rise too.

1704

S. Rahman Mohammadpour and S. H. A. Rahmati

Fig. 9 Hypothesis test by considering confidence interval for GA and BBO algorithm

Table 4 Hypothesis test results on best cost and time of the algorithms Metric Best cost Time

4.1

Difference −8,699,141 7.64080

Confidence interval for difference (−563,025,666, 372,423,062) (1.64493, 13.6542)

Results No significant difference No significant difference

Hypothesis Test

We implemented Mann-Whitney hypothesis test since our data are non-normal. Figure 9 is the results of the best cost and time from the hypothesis test, and Table 4 shows that both of statistics are in their confidence interval and there is no significant difference between GA and BBO algorithm. In this study, we illustrate the convergence plots based evolutionary algorithms for the first time. These plots create better clarification to show the convergence process as well as possible. The strength, competitive, and positive point of this study compared to other studies is the vast range of numerical examples, which are proposed in Table 3. Figures 10 and 11 demonstrate the results of convergence plot on ten test problems by GA and BBO methods. The most important point in comparing the GA and BBO algorithm is their convergence speed.

4.2

Discussion and Implications

This research focuses on developing a production-distribution model for extending the existing related literature (Liotta et al. 2015; Mariano et al. 2016; Yao et al. 2018; Zhang et al. 2018; Cai et al. 2019; Babagolzadeh et al. 2020). The competitive point of this research is that both production and distribution sectors are green while most of the literature just has green production (Yao et al. 2018; Cai et al. 2019) or green distribution (Ulku 2012; Meng and Niu 2011; Liotta et al. 2015; Harris et al. 2011; Mariano et al. 2016; Zhang et al. 2018; Wang et al. 2020;

An Intelligent Decision Support System for an Integrated Energy Aware. . .

1705

Fig. 10 Convergence plot of GA method on ten test problems

Babagolzadeh et al. 2020). Our developed model does not only consider a simple energy consumption coefficient (Liotta et al. 2015; Harris et al. 2011; Mariano et al. 2016; Yao et al. 2018; Zhang et al. 2018; Cai et al. 2019) and go deeper by adapting real-world formulation of energy consumption (Ulku 2012; Meng and Niu 2011; Wang et al. 2020; Babagolzadeh et al. 2020) method by production-distribution literature in Eqs. (3), (15), (20), (21), and (22). To achieve this compatibility, PMSP and logistics are coordinated with modules as green production-distribution modelers. The managerial consequence of this formula is that it creates social justice for the payment of energy by considering the place of production as a social concern. According to sin (Φ) coefficient in Eq. (3), if a place has a lower suitability index or is located in a less developed area, its energy cost coefficient is significantly reduced. In the long run, underdeveloped areas are likely to develop because of the

1706

S. Rahman Mohammadpour and S. H. A. Rahmati

Fig. 11 Convergence plot of BBO method on ten test problems

large number of companies established due to low energy costs. The proposed model is in the category of complex models; hence it was developed by metaheuristic algorithms as an NP-Hard problem. However, only a limited number of studies use meta-heuristic algorithms to solve and develop their models (Zhang et al. 2018; Babagolzadeh et al. 2020). These novel algorithms are another implication of this

An Intelligent Decision Support System for an Integrated Energy Aware. . .

1707

study that can be used by other researcher for any other NP-Hard optimization environment problems.

5

Conclusion

This study developed a sustainable intelligent and smart system to reduce total production-distribution costs, energy consumption, and CO2 emission. Actually, the main contribution of this chapter is introducing an intelligent DSS for controlling affairs of a complex PDS in which reducing CO2 emission and enhancing company economical profit are in conflict. The developed model surrounds the mathematical formulation of manufacturing and vehicle logistic pollutions according to real coefficients and calculations of CO2 emission per electricity consumption (kg/kWh) and CO2 emission per fuel consumption (kg/L). Introduced Smart DSS provides PDS owners with a system in which they are able to evaluate various planning scenarios to improve their environmental responsibility and profit simultaneously. The mathematical operators are provided according to the structure of our problem as a platform for designing the solving methodology alongside the developed adapted solution and neighborhood structures of the GA and BBO algorithm. The algorithms are accompanied with powerful graphical illustrations of the operators to have explicit description. In a total view, there are no significant differences among the results of the algorithms on different test problems of the study. This output proves the repeatability and reproducibility capability of the algorithms’ operators and bears out the potential of implementation of them in other similar cases. These systems implement intelligent modules inspired to offer solutions to various decisions of system owners in the shortest time and without the need for real trial and error. Various analytical and experimental results were presented to more explicitly introduce the performance of DSS and PDS systems. Future work of this chapter can concentrate on other terms of a green supply chain management such as green supplier selection and green purchase. Developing model by considering inventory management and queue theory in production is also of interest.

6

Cross-References

Predicting US Energy Consumption Utilizing Artificial Neural Network US Natural Gas Consumption Analysis via a Smart Time Series Approach Based

on Multilayer Perceptron ANN Tuned by Meta-heuristic Algorithms

References M. Babagolzadeh, A. Shrestha, B. Abbasi, Y. Zhang, A. Woodhead, A. Zhang, Sustainable cold supply chain management under demand uncertainty and carbon tax regulation. Transp. Res. Part D: Transp. Environ. 80, 102245 (2020). https://doi.org/10.1016/j.trd.2020.102245

1708

S. Rahman Mohammadpour and S. H. A. Rahmati

G.P. Cachon, Retail store density and the cost of greenhouse gas emissions. Manag. Sci. 60, 1907– 1925 (2014). https://doi.org/10.1287/mnsc.2013.1819 W. Cai, K.H. Lai, C. Liu, F. Wei, M. Ma, S. Jia, Z. Jiang, L. Lv, Promoting sustainability of manufacturing industry through the lean energy-saving and emission-reduction strategy. Sci. Total Environ. 665, 23–32 (2019). https://doi.org/10.1016/j.scitotenv.2019.02.069 C. Cheng, P. Yang, M. Qi, L.M. Rousseau, Modeling a green inventory routing problem with a heterogeneous fleet. Transp. Res. Part E: Logist. Transp. Rev. 97, 97–112 (2017). https://doi.org/10.1016/j.tre.2016.11.001 E. Demir, T. Bekta¸s, G. Laporte, An adaptive large neighborhood search heuristic for the pollution-routing problem. Eur. J. Oper. Res. 223, 346–359 (2012). https://doi.org/10.1016/j.ejor.2012.06.044 M. Pasandidehpoor, J. Mendes-Moreira, S. Rahman Mohammadpour, R.T. Sousa, Predicting US Energy Consumption Utilizing Artificial Neural Network. Handbook of Smart Energy Systems (2022). https://doi.org/10.1007/978-3-030-97940-9_136-1 I. Harris, M. Naim, A. Palmer, A. Potter, C. Mumford, Assessing the impact of cost optimization based on infrastructure modelling on CO2 emissions. Int. J. Prod. Econ. 131, 313–321 (2011). https://doi.org/10.1016/j.ijpe.2010.03.005 K. Nokhbeh Dehghan, S. Rahman Mohammadpour, S.H.A. Rahamti, US Natural Gas Consumption Analysis via a Smart Time Series Approach Based on Multilayer Perceptron ANN Tuned by Metaheuristic Algorithms. Handbook of Smart Energy Systems (2022). https://doi.org/10.1007/ 978-3-030-97940-9_137-1 Ç. Koç, T. Bekta¸s, O. Jabali, G. Laporte, The fleet size and mix pollution-routing problem. Transp. Res. Part B: Methodol. 70, 239–254 (2014). https://doi.org/10.1016/j.trb.2014.09.008 G. Liotta, G. Stecca, T. Kaihara, Optimisation of freight flows and sourcing in sustainable production and transportation networks. Int. J. Prod. Econ. 164, 351–365 (2015). https://doi.org/10.1016/j.ijpe.2014.12.016 E.B. Mariano, J.A. Gobbo Jr., F. de Camioto, D. Rebelatto, CO2 emissions and logistics performance: a composite index proposal. J. Clean. Prod. 163, 166–178 (2016). https://doi.org/10.1016/j.jclepro.2016.05.084 M. Meng, D. Niu, Modeling CO2 emissions from fossil fuel combustion using the logistic equation. Energy 36, 3355–3359 (2011). https://doi.org/10.1016/j.energy.2011.03.032 S.H. Rahmati, M. Zandieh, A new biogeography-based optimization (BBO) algorithm for the flexible job shop scheduling problem. Int. J. Adv. Manuf. Technol. 58, 1115–1129 (2011). https://doi.org/10.1007/s00170-011-3437-9 M. Soysal, J.M. Bloemhof-Ruwaard, R. Haijema, J.G. van der Vorst, Modeling an inventory routing problem for perishable products with environmental considerations and demand uncertainty. Int. J. Prod. Econ. 164, 118–133 (2015). https://doi.org/10.1016/j.ijpe.2015.03.008 M.A. Ulku, Dare to care: shipment consolidation reduces not only costs, but also environmental damage. Int. J. Prod. Econ. 139, 438–446 (2012). https://doi.org/10.1016/j.ijpe.2011.09.015 C. Wang, Y. Zhao, Y. Wang, J. Wood, C.Y. Kim, Y. Li, Transportation CO2 emission decoupling: an assessment of the Eurasian logistics corridor. Transp. Res. Part D: Transp. Environ. 86, 102486 (2020). https://doi.org/10.1016/j.trd.2020.102486 Y. Yao, J. Jiao, X. Han, C. Wang, Can constraint targets facilitate industrial green production performance in China? Energy-saving target vs emission-reduction target. J. Clean. Prod. 209, 862–875 (2018). https://doi.org/10.1016/j.jclepro.2018.10.274 D. Zhang, Q. Zhan, Y. Chen, S. Li, Joint optimization of logistics infrastructure investments and subsidies in a regional logistics network with CO2 emission reduction targets. Transp. Res. Part D: Transp. Environ. 60, 174–190 (2018). https://doi.org/10.1016/j.trd.2016.02.019

Design and Operational Strategies for Grid-Connected Smart Home Manimuthu Arunmozhi, S. Senthilmurugan, and Viswanathan Ganesh

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Smart Devices and Sensors at Grid Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Smart Devices and Sensors at Home Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Designing of Existing Household Devices in PSCAD . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Data Flow Between Smart Grid and Smart Homes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1710 1712 1718 1720 1727 1731

Abstract

The progressing pace of the world is phenomenal, and to hold the levels of sustainability, many aspects of smart terminology come into existence such as smart grid, smart lights, smart meters, etc. The level of depreciation of available fossil fuels has made everyone search for alternatives that can be renewable and save the environment. Hence, the overall generation, transmission, and distribution system requires a revamp from the traditional methods. This chapter focuses on the consumer’s end where the domestic load can be converted to smart appliances and work coherently with the smart meters followed by the distribution side. This chapter discusses the above aspects being implemented in

M. Arunmozhi () Energy Research Institute (ERIAN), Nanyang Technological University, Singapore, Singapore e-mail: [email protected] S. Senthilmurugan SRM Institute of Science and Technology, Kattankulatur, Chennai, India e-mail: [email protected] V. Ganesh Department of Sustainable Electric Power Engineering and Electromobility, Chalmers University of Technology, Gothenburg, Sweden e-mail: [email protected] © Springer Nature Switzerland AG 2023 M. Fathi et al. (eds.), Handbook of Smart Energy Systems, https://doi.org/10.1007/978-3-030-97940-9_78

1709

1710

M. Arunmozhi et al.

software like PSCAD and OMNET++, and a custom community with collections of houses are combined to check the various parameters for a smart grid and a smart home. Keywords

Smart devices · Sensors · Simulation & modelling · Smart grid

1

Introduction

The impact of energy production and consumption has been in the positive inclination for the past four decades (Sakib et al. 2021; Bambrik 2020; Manimuthu and Ramadoss 2019). The need for energy didn’t rise until the development of extensive appliances for both domestic and commercial purposes; considering the development of certain devices in the field of semiconductors, they have created a positive impact on the individual and efficient consumption of energy (Shin et al. 2020; Mylonas et al. 2020). For example, the groundbreaking invention in the field of lighting was LEDs which consumed only a partial amount of energy consumed by conventional incandescent and CFL bulbs. Similarly, there are certain grid systems where the load demand has risen drastically, therefore breaking the limit of the current-carrying capacity of the cable designed for transmission systems; hence, the overall efficiency of the total system is reduced significantly (Khalid and Shobole 2021; Kashem et al. 2020; Ahmadian et al. 2020; Sovacool and Del Rio 2020). Therefore, there exist significant ways possible for the efficient generation and consumption of energy, by increasing the size of the current-carrying conductor, generating the required power locally near the load which reduces the losses that occurred due to transmission. The mode of generation, transmission, and distribution can be in either complete DC or complete AC, else a hybrid system comprising both DC and AC. When the system is completely DC, there is no need for inverters and transformers as they have high losses. On the other hand, when the system is completely AC, it comprises of transformers and inverters that can convert DC to AC. Therefore, a combination of AC and DC grid will be an optimal selection for the designing of a new grid. The concept of smart grid has been initiated to enhance the reliability, resilience, and efficiency of overall energy consumption. Smart grid is a combination of all terminology and devices such as smart meter, smart load sharing algorithm, peak shaving algorithm, etc. (Rashid et al. 2020; Ceglia et al. 2020; van Summeren et al. 2020; Barone et al. 2020). The classification of smart grid can be expanded with the mode of generation, transmission, and distribution of energy, for example, the conventional and classical electrical grid can be compared to a one-way flow of power which comprised of central generation method where energy would be generated at far distances from the load. Therefore, the mode of transmission plays a crucial role which can cause power losses and overall reduction of efficiency from generation to consumption. Similarly, on the other hand, when we arrange a smart

Design and Operational Strategies for Grid-Connected Smart Home

1711

Table 1 Classification of smart grid on the basis of capacity Classification of smart grid Pico Nano Micro Mini

Capacity (kW) C