Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning 0128238992, 9780128238998

Citation preview

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning Edited by

GIORGIO GRADITI Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy

MARIALAURA DI SOMMA Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy

Elsevier Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States Copyright © 2021 Elsevier Inc. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBN: 978-0-12-823899-8 For Information on all Elsevier publications visit our website at https://www.elsevier.com/books-and-journals

Publisher: Joe Hayton Acquisitions Editor: Graham Nisbet Editorial Project Manager: Chris Hockaday Production Project Manager: Prasanna Kalyanaraman Cover Designer: Christian Bilbow Typeset by MPS Limited, Chennai, India

Contents

List of contributors

1.

Overview of distributed energy resources in the context of local integrated energy systems

xi

1

Gianfranco Chicco, Marialaura Di Somma and Giorgio Graditi

2.

Abbreviations 1.1 Introduction 1.2 Distributed energy resources 1.3 Grid side aspects 1.4 Emergent paradigms and solutions References

1 1 3 7 16 26

Architectures and concepts for smart decentralised energy systems

31

Andrei Z. Morch, Chris Caerts, Anna Mutule and Julia Merino Abbreviations 2.1 Introduction 2.2 Why decentralizing the energy system? 2.3 Development of the decentralized architecture 2.4 Grid-secure activations for ancillary services (real-time control) 2.5 ELECTRA Web-of-Cells control concept 2.6 Post-primary voltage control 2.7 Balance restoration control 2.8 Balance steering control 2.9 Adaptive frequency containment control 2.10 Inertia control 2.11 Decentralizing the DA/ID energy market clearing and grid prequalification of ancillary services 2.12 What is next: evolution of roles and responsibilities necessary for decentralization the European regulatory framework 2.13 Conclusions References

3.

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

31 32 33 36 41 45 46 47 48 48 49 49 57 58 59

63

Nilufar Neyestani Abbreviations Nomenclature

63 63

v

vi

Contents

4.

3.1 Introduction 3.2 Internal multicarrier dependency in a smart local system 3.3 External dependencies in a smart local system 3.4 Interdependency modeling 3.5 A case study on interdependent MES model 3.6 Conclusions References

64 66 76 78 82 84 85

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

89

Bing Yan, Marialaura Di Somma and Giorgio Graditi Abbreviations Nomenclature 4.1 Importance of multiobjective operation optimization for short- and long-run sustainability of local integrated energy systems 4.2 Multiobjective optimization for the operation of a local integrated energy system 4.3 Case study: eco-exergetic operation optimization of a local integrated energy system for a large hotel in Beijing 4.4 Operation optimization of multiple integrated energy systems in a local energy community 4.5 Conclusions and key findings References

5.

Impact of neighborhood energy trading and renewable energy communities on the operation and planning of distribution networks

89 89 92 96 103 108 119 120

125

Alberto Borghetti, Camilo Orozco Corredor, Carlo Alberto Nucci, Ali Arefi, Javid Maleki Delarestaghi, Marialaura Di Somma and Giorgio Graditi Abbreviations Nomenclature 5.1 Introduction 5.2 A distributed approach for the day-ahead scheduling of the LEC 5.3 Implementation and numerical tests 5.4 Distribution network planning model considering nonnetwork solutions and neighborhood energy trading 5.5 Application of the planning model to case studies and analysis of the results 5.6 Conclusions Acknowledgment References

125 126 128 129 138 148 160 170 171 171

Contents

6.

Fostering DER integration in the electricity markets

175

Julia Merino, Inés Gómez, Jesús Fraile-Ardanuy, Maider Santos, Andrés Cortés, Joseba Jimeno and Carlos Madina Abbreviations 6.1 Distributed energy resources as providers of flexibility services 6.2 The regulatory framework for the participation of distributed energy resources in different electricity markets 6.3 Flexibility needs in power systems 6.4 The market value of flexibility in the distribution system 6.5 Local energy markets 6.6 Conclusions References

7.

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

175 176 181 188 193 198 202 203

207

M. Caruso, P. Gallo, M.G. Ippolito, S. Nassuato, N. Tomasone, E.R. Sanseverino, G. Sciumè and G. Zizzo Abbreviations 7.1 Introduction 7.2 The blockchain technology 7.3 The energy blockchain: current trends and possible evolutions 7.4 Laboratory setup for energy blockchain testing 7.5 Conclusions Acknowledgment References

8.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

207 208 209 218 223 228 229 229

231

Anna Pinnarelli, Daniele Menniti, Nicola Sorrentino and Angel A. Bayod-Rújula Abbreviations Nomenclature 8.1 Introduction 8.2 Energy storage systems as distributed flexibility 8.3 The energy storage system in a nanogrid: the configuration 8.4 Optimal energy management for virtual nonsumers nanogrid community 8.5 The energy storage systems for grid ancillary service 8.6 Case study 8.7 Conclusions References

231 232 233 234 239 259 263 268 276 276

vii

viii

Contents

9.

Demand response role for enhancing the flexibility of local energy systems

279

Seyed Amir Mansouri, Amir Ahmarinejad, Mohammad Sadegh Javadi, Ali Esmaeel Nezhad, Miadreza Shafie-Khah and João P.S. Catalão Abbreviations Nomenclature 9.1 Introduction 9.2 Demand response programs for local energy systems 9.3 Flexibility assessment of local energy systems in the presence of energy storage systems and DR programs 9.4 Energy management framework for DER integrated distribution networks 9.5 Simulation results 9.6 Conclusion remarks Acknowledgment References

10. The integration of electric vehicles in smart distribution grids with other distributed resources

279 279 282 286 292 292 304 310 311 311

315

Morris Brenna, Federica Foiadelli, Dario Zaninelli, Giorgio Graditi and Marialaura Di Somma Abbreviations Nomenclature 10.1 Introduction to electric vehicles and charging infrastructures 10.2 Integration of electric vehicles in smart distribution grids 10.3 Vehicle-to-Grid 10.4 Conclusions References

11. Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

315 315 316 322 333 344 345

347

C.N. Papadimitriou, M. Patsalides, V. Venizelos, P. Therapontos and V. Efthymiou Abbreviations Nomenclature 11.1 Introduction 11.2 RES uncertainties description and assessment 11.3 Uncertainties affecting system resilience 11.4 Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER 11.5 Discussion and conclusions References

347 348 348 352 363 370 385 385

Contents

12. Load forecasting in the short-term scheduling of DERs

389

Jiajia Yang, Fengji Luo, Weicong Kong and Zhao Yang Dong Abbreviations Nomenclature 12.1 Introduction 12.2 New trends in load forecasting 12.3 Trans-active energy systems with DERs 12.4 Short-term scheduling of DERs in demand side 12.5 Conclusions and future thoughts References

13. Conclusions and key findings of optimal operation and planning of distributed energy resources in the context of local integrated energy systems

389 390 390 392 399 406 412 414

419

Giorgio Graditi and Marialaura Di Somma Index

427

ix

List of contributors

Amir Ahmarinejad Department of Electrical Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran Ali Arefi College of Science, Health, Engineering and Education, Murdoch University, Perth, WA, Australia Angel A. Bayod-Rújula Department of Electrical Engineering, University of Zaragoza, Zaragoza, Spain Alberto Borghetti Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy Morris Brenna Department of Energy, Politecnico Di Milano, Milan, Italy Chris Caerts VITO/EnergyVille, Genk, Belgium M. Caruso Exalto Energy & Innovation Srl, Palermo (PA), Italy João P.S. Catalão Institute for Systems and Computer Engineering, Technology and Science (INESC TEC), Porto, Portugal; Faculty of Engineering of the University of Porto, Porto, Portugal Gianfranco Chicco Politecnico di Torino, Torino, Italy Andrés Cortés TECNALIA, Basque Research and Technology Alliance (BRTA), Derio, Spain; Department of Electrical Engineering, University of the Basque Country, Bilbao, Spain Javid Maleki Delarestaghi College of Science, Health, Engineering and Education, Murdoch University, Perth, WA, Australia Marialaura Di Somma Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy Zhao Yang Dong School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney, NSW, Australia V. Efthymiou FOSS Research Centre for Sustainable Energy of University of Cyprus, Nicosia, Cyprus Federica Foiadelli Department of Energy, Politecnico Di Milano, Milan, Italy

xi

xii

List of contributors

Jesús Fraile-Ardanuy Information Processing and Telecommunication Center (IPTC-SISDAC), Universidad Politécnica de Madrid, Madrid, Spain P. Gallo Department of Engineering, University of Palermo, Palermo (PA), Italy Inés Gómez TECNALIA, Basque Research and Technology Alliance (BRTA), Derio, Spain Giorgio Graditi Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy M.G. Ippolito Department of Engineering, University of Palermo, Palermo (PA), Italy Mohammad Sadegh Javadi Institute for Systems and Computer Engineering, Technology and Science (INESC TEC), Porto, Portugal Joseba Jimeno TECNALIA, Basque Research and Technology Alliance (BRTA), Derio, Spain Weicong Kong School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney, NSW, Australia; ALDI Stores, Sydney, NSW, Australia Fengji Luo School of Civil Engineering, University of Sydney, Sydney, NSW, Australia Carlos Madina TECNALIA, Basque Research and Technology Alliance (BRTA), Derio, Spain Seyed Amir Mansouri Department of Electrical Engineering, Yadegar-e-Imam Khomeini (RAH) Shahre Rey Branch, Islamic Azad University, Tehran, Iran Daniele Menniti Department of Mechanical, Energy and Management Engineering, University of Calabria, Rende, Italy Julia Merino TECNALIA, Basque Research and Technology Alliance (BRTA), Derio, Spain; Department of Electrical Engineering, University of the Basque Country, Bilbao, Spain Andrei Z. Morch SINTEF Energy Research, Trondheim, Norway Anna Mutule Institute of Physical Energetics, Riga, Latvia S. Nassuato Regalgrid Europe S.r.l., Treviso (TV), Italy

List of contributors

Nilufar Neyestani Centre for Power and Energy Systems, INESC TEC, Porto, Portugal Ali Esmaeel Nezhad Department of Electrical, Electronic, and Information Engineering, University of Bologna, Bologna, Italy Carlo Alberto Nucci Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy Camilo Orozco Corredor Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy C.N. Papadimitriou FOSS Research Centre for Sustainable Energy of University of Cyprus, Nicosia, Cyprus M. Patsalides FOSS Research Centre for Sustainable Energy of University of Cyprus, Nicosia, Cyprus Anna Pinnarelli Department of Mechanical, Energy and Management Engineering, University of Calabria, Rende, Italy E.R. Sanseverino Department of Engineering, University of Palermo, Palermo (PA), Italy Maider Santos TECNALIA, Basque Research and Technology Alliance (BRTA), Derio, Spain G. Sciumè Department of Engineering, University of Palermo, Palermo (PA), Italy Miadreza Shafie-Khah School of Technology and Innovations, University of Vaasa, Vaasa, Finland Nicola Sorrentino Department of Mechanical, Energy and Management Engineering, University of Calabria, Rende, Italy P. Therapontos Electricity Authority of Cyprus (EAC), Nicosia, Cyprus N. Tomasone Regalgrid Europe S.r.l., Treviso (TV), Italy V. Venizelos FOSS Research Centre for Sustainable Energy of University of Cyprus, Nicosia, Cyprus Bing Yan Department of Electrical and Microelectronic Engineering, Rochester Institute of Technology, Rochester, NY, United States

xiii

xiv

List of contributors

Jiajia Yang School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney, NSW, Australia Dario Zaninelli Department of Energy, Politecnico Di Milano, Milan, Italy G. Zizzo Department of Engineering, University of Palermo, Palermo (PA), Italy

CHAPTER 1

Overview of distributed energy resources in the context of local integrated energy systems Gianfranco Chicco1, Marialaura Di Somma2 and Giorgio Graditi2 1

Politecnico di Torino, Torino, Italy Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy 2

Abbreviations DER DG DMES DR DS DSM DSO EH ENTSO-E ESP EV ICT IES MES MPER PV RES RTP SNET TES TOU V2G VPP

Distributed Energy Resources Distributed Generation Distributed Multi-Energy Systems Demand Response Distributed Storage Demand Side Management Distribution System Operator Energy Hubs European Network of Transmission System Operators for Electricity Electricity Shifting Potential Electric Vehicle Information and Communication Technology Integrated Energy Systems Multi-Energy Systems Maximum Profit Electricity Reduction Photovoltaic Renewable Energy Sources Real-Time Pricing Smart Networks for Energy Transition Thermal Energy Storage Time-of-Use Vehicle-to-Grid Virtual Power Plants

1.1 Introduction Distributed Energy Resources (DER) are technologies and means that can be deployed at the supply side or demand side of a Low Voltage or Medium Voltage electric distribution system to meet the energy and reliability needs of the user(s) served by that system. The DER components are partitioned into Distributed Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00002-9

r 2021 Elsevier Inc. All rights reserved.

1

2

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Generation (DG), Demand Response (DR), and Distributed Storage (DS). A particular type of DG is the one composed of Renewable Energy Sources (RES). From another point of view, DER contain both local generation resources connected to the distribution system or at the customer side of the meter, as well as demand-side resources such as load management systems or local energy storage. The latter are aimed at changing the shape of the electrical demand curve and reducing the internal demand of the consumer. The development of local energy systems for DG and storage is not new. The conceptualisation of most of the relevant aspects was already conducted in the early Eighties [1,2]. Likewise, the principles of Demand Side Management (DSM) were defined in the Eighties as an attempt to build structured solutions for promoting the consumer participation to making the load patterns variable in time [3]. However, the development of DG, DS, and DSM solutions has been initially limited by the lack of suitable technologies available at accessible costs to the generality of the users. With respect to the situation that was in place in the Eighties in many industrialized countries, in the following years there have been significant changes that have introduced new perspectives and drivers for the successive evolution. The most significant drivers have been technical, economic, and social. On the technical side, relevant advances have been conducted under the smart grid framework, in which Information and Communication Technologies (ICT) and distribution automation have been exploited for the modernization of the electrical infrastructure. In the smart grid framework, ICT plays a primary role, introducing increasingly higher capabilities for system monitoring and control, as well as data management (communication, storage, security, and analysis tools). In addition, the technologies for energy generation and storage have enhanced their performance and extended the range of the available sizes, up to making available small-scale and micro-power sources. The transition towards a new energy system characterized by a multitude of players has requested the creation of regulatory authorities to upgrade and extend the set of rules, to solve new issues concerning grid connection and more detailed protection system settings. On the economic side, the advent and evolution of the competitive electricity markets, and the growing refinement of the regulation of the quality of supply, have created the unbundling of the electricity business with the separation of generation, transmission, distribution and retail, and have established new rules to incentivise or penalize the operators of the electrical system. On the social side, the increasingly high attention towards environmental aspects has pushed the strong development of RES, also with the consequent improvement in the technical solutions, while the growing interest to promote consumer awareness and engagement has led to the introduction of new regulatory provisions concerning the demand side.

Overview of distributed energy resources in the context of local integrated energy systems

Most of the changes mentioned above have occurred during the restructuring of the electrical system. However, during the years there has been a progressive integration of the electrical system with other (nonelectrical) energy systems. This has led to develop and apply multi-energy solutions, in which there is a stronger integration among the energy network infrastructures and energy carriers, enabling the definition of proactive solutions to provide energy services. These solutions also include further interaction with the mobility infrastructure to encompass the diffusion of electric vehicles (EVs). Recent developments towards enabling the formation of energy communities, also supported by appropriate policies, are driving the energy systems towards further integration. These developments are shaping the contours of an energy transition with a stronger cross-sector integration strengthening the link among climate, energy and mobility, in which the role of electricity is expected to increase in the final energy uses.

1.2 Distributed energy resources 1.2.1 Distributed generation based on different energy sources In general, the choice of the DG technology to be installed is largely conditioned by the availability of the primary energy source. For a local energy system, this is even more a limiting factor. Another key aspect is the modularity of the DG plant, which has a double impact, (1) making it possible to construct the plant at different steps, while the portion of the plant already installed can be already profitable, and (2) enabling the operation of the plant as a cluster of modules, in which each module can be activated or not depending on the needs. On the input side, the DG solutions can be partitioned by considering their type of supply, as DG supplied with: a. System-based energy, such as network-based energy (e.g., electricity and fuels), or stored energy. The fuel supply is considered available in the limits of the continuity of supply of the networks, and the power that can be supplied is limited by the capacity of the network connection. Availability of a storage system allows time shifting of the supplied energy provision; however the power to be supplied has to be scheduled by taking into account the time-dependent constraints on the storage system. b. Environment-based energy: in this case, the energy comes from a primary source such as solar irradiance or wind speed and direction, which are by nature uncertain and fluctuating in time. A positive aspect of DER is the possibility of exploiting a mix of local energy sources, to benefit from the complementarities among these sources. This is particularly true for RES-based generation, for which the coupling with a storage system

3

4

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

enables smoothing the fluctuations of the power output. The environmental impact of DER is assessed by considering the effects concerning emissions, noise, and visual impact. Technical and environmental aspects are then merged to establish the cost effectiveness of the DER solutions, in which the costs involved include installation, operation, and maintenance, with possible incentives and penalties established on the basis of the regulation in place. On the modeling side, some details appearing in local energy systems cannot be approximated easily. For example, unbalanced loads are very likely to be found in local energy systems, making three-phase power flow analysis necessary in several cases. In addition, specific modeling is needed for many local units (generators and loads) controlled by power electronics, which cannot be simply modeled by using standard models (e.g., constant impedance, constant current, or constant power). In particular, aggregate models of loads such as the ones controlled by thermostats, electric vehicles, or storage units such as water heaters, have to be formulated in an effective way to represent the physical behavior of these units, also in dynamic conditions. For this purpose, the power node model [4] has been introduced as a modeling framework in which different components, including generic energy storage units, are represented in such a way to be used for both power system analysis and dynamic simulations.

1.2.2 Combined production of different energy carriers DG applications such as cogeneration (typically with electricity and heat) and multigeneration (e.g., with electricity, heat, and cooling) are based on the combined generation of different energy carriers. The relevant paradigms referring to combined generation include: • Integrated Energy Systems (IES), which focuses on the integration of DG equipment with thermally activated technologies. The IES program was launched in 2001 by the US Department of Energy, and includes laboratory-based applications, e.g., for composed systems including microturbine with heat recovery, air conditioning and ventilation, desiccant, and absorption chiller units [5]. • Energy Hubs (EH), which refers to the integrated delivery of different energy carriers in multi-carrier energy systems [6]. The EH framework was developed within the project “Vision of Future Energy Networks,” focusing on the long-term evolution of the energy systems. • Multi-Energy Systems (MES), which address the efficient exploitation of the combined production of different energy carriers connected to energy networks [7 9]. The EH framework has been adopted and extended for MES and distributed multi-energy systems (DMES) to deal with environmental impact and flexibility aspects [10]. A detailed discussion on MES is provided in Chapter 3 of this book.

Overview of distributed energy resources in the context of local integrated energy systems

Common aspects of the IES, EH, and MES paradigms are the formulation of mathematical models of the individual components and of the whole system, and the use of these models for energy efficiency maximization. The matrix form of the EH model has become a very successful structured framework for the mathematical representation of the interactions among the components of the energy system, useful for defining an input output model that takes into account the efficiencies of the components and the topological connections. Storage has been included in the EH model in such a way that its contribution is beneficial with respect to the basic EH model [11]. Another additional contribution has been included in the EH model to represent the load shifted for DSM purposes [12]. The EH model has been further extended to account for the dependencies among different energy carriers, enabled from availability of alternative solutions, based on different energy carriers, to provide the same final service [13]. In this case, the consumers may choose the energy carrier, with choices variable in time. The combined production of multi-energy outputs, integrated into local energy systems, is an essential component for maximizing energy efficiency. Furthermore, it may allow improving the environmental impact by reducing the emissions of local and global pollutants. Suitable indicators have been formulated to calculate the relative variation of the equivalent fuel input (for energy efficiency purposes) or the relative variation of the mass of pollutant (for environmental impact calculations) with respect to the separate production of the same useful energy outputs [14]. Energy efficiency and environmental impact indicators can be used as objective functions to determine the energy inputs that maximize the energy and environmental performance of the MES. The interactions between the different energy carriers may enable better reliability of the energy system. Reliability has to be evaluated by considering specific approaches, which can be partitioned into model-driven modeling and data-driven modeling [15]. In a MES, the interactions among the energy carriers have been studied by determining the multienergy feasible operating regions [16] and by defining the multi-energy node model as an extension of the power node model [10].

1.2.3 Demand response DR is the evolution of a number of early approaches referring to the demand side. Among them, DSM was based on a set of principles, including peak shaving, valley filling, load shifting, flexible load shape, and strategic load reduction or growth [17]. The studies on DSM evolved in parallel with the analysis of possible time-dependent tariff rates, aimed at using the tariff lever to induce the consumers to shifting their electricity consumption to off-peak hours. Among these rates, Time-Of-Use (TOU) rates are predefined for different periods (e.g., peak hours, intermediate hours, off-peak

5

6

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

hours) are kept constant for a relatively long period (e.g., one year). An evolution of the TOU concept is Real-Time Pricing (RTP), with variation of the electricity rates during time (e.g., at each hour) defined in a closer advance (e.g., daily). The further evolution is Spot Pricing, where the instant at which the rate is conceptually determined is immediately before the instant of consumption [18]. In all these early approaches, the role of the consumer is to take decisions considering the rate or price information provided by an external entity. Since the beginning of the third millennium, the restructuring of the electricity business has started some actions intended to involve the consumers to a larger extent. DR has been introduced as the voluntary reduction of the electricity use in response to price signals or incentives, aimed at improving the effectiveness of the electricity supply in response to specific needs (e.g., reducing the demand peaks in critical periods, with respect to a suitably defined baseline). Indeed, the introduction of social and psychological components for representing the behavior of the user makes the DR models more accurate [19] and enables better assessment of incentive-based DR [20]. The variety of the users that may be more or less interested in DR programmes [21], as well as the possible missing information due to communication inaccuracies, requires the adoption of tools able to handle incomplete information [22]. A detailed view on the DR programmes for DER integration in local energy systems is presented in Chapter 9 of this book.

1.2.4 Distributed storage The classical view of the electrical service considers a just-in-time commodity, for which the generation follows the load (including the system losses). In this view, storage is generally not included, or has a minor role. For large electrical systems, this view is still valid, as the size of the storage systems available today is rather limited. However, for small local energy systems the situation is different. The size of the locally available storage in some cases is sufficient to change the paradigm from justin-time to time-adjustable commodity, conceptually reaching the condition in which load follows generation. Ideally, with a sufficient amount of storage there would be no net power exchanged with the grid. However, in practice the capacity, energy, and ramp-rate constraints of the storage units impose limits to the storage system operation. Furthermore, the time-adjustable paradigm has to cope with the uncertainty and fluctuation of the generation from RES, which needs higher demand-side flexibility to exploit more generated power from RES. The main distinction among the storage technologies is based on their powerbased or energy-based usage [23]. The power-based usage implies fast provision of the service, with short times and relatively high power needed from the infrastructure. As such, it is more suitable for ensuring continuity of supply (in local energy systems

Overview of distributed energy resources in the context of local integrated energy systems

through electric batteries, supercapacitors or flywheels), or for automotive applications. The energy-based usage implies slow provision of the service, with long times and relatively low power needed from the infrastructure. This usage is more suitable for applications that provide energy system services. In local energy systems, batteries, or hydrogen-based systems with fuel cells, can support frequency control. An interesting solution is the adoption of flow batteries, in which power and energy can be decoupled depending on the size of the stack to provide more power, and on the size of the electrolyte tanks to provide more energy [24]. Other promising solutions come from vehicle-to-grid (V2G) applications, with the deployment of the electricity stored in the batteries of the EVs, provided that appropriate infrastructure to manage the network connection with battery charging and discharging is in place. The integration of EVs as a DER component in the distribution grids is addressed in Chapter 10 of this book. For long-term energy storage, in local energy systems there are power-to-X solutions (e.g., power-to-liquid and power-to-gas), even though availability of these solutions for small-scale applications is still rather limited. In addition to the storage of electricity, in local energy systems there are viable applications for thermal energy storage (TES) solutions, in which thermal energy is stored to create heat or cooling buffers [25,26]. TES systems have a slower response with respect to electrical storage systems, because their thermal capacity increases the thermal time constant. In addition, the stored heat or cooling cannot be used at high distances from the storage location, because of the increase in the temperature variations and of the corresponding thermal losses at higher distances, while a form of TES can be provided by the interaction with district heating and district cooling systems [27]. Furthermore, the efficiency of the TES system could be relatively low, and thermal standby losses between the storage medium and the environment are not negligible. However, in local applications, TES systems may have benefits from their relatively low cost of installation and maintenance, and low pollutant emissions. Moreover, TES systems can be used to enhance the operational flexibility of buildings [28] and distributed systems [29]. In addition to the many applications of MES-integrated heat storage, cold TES systems can be used for shifting the peak loads in buildings applications [30]. As such, TES systems are an important asset for local energy systems, also enabling distributed energy storage at specific locations within energy communities. The management of energy storage systems in an energy community is discussed in Chapter 8 of this book.

1.3 Grid side aspects 1.3.1 Evolution of the grid connection issues and standards Since the beginning of the diffusion of DER, grid connection has been the major issue for enabling the development and deployment of DER solutions, in particular for

7

8

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

DG. Starting from a situation in which the electrical distribution networks were designed and operated in a centralized way, the connection of new DG plants introduced power injections not controllable by the distribution system operator (DSO). Because of this, new technical rules had to be issued by the regulatory authorities to establish viable conditions for authorizing the grid connections. In the early period of DER integration (before the end of the Nineties), the power injected in the grid was relatively low, and did not cause particular technical problems to the voltage profiles, nor led to excessive growth in the short circuit currents. Hence, the perturbations to the distribution network were low, and there were even benefits given by the reduction of the net power load (i.e., demand minus generation), which reduced the currents flowing in the lines and the corresponding losses and voltage deviations from the rated voltage. In the further decade, there was a growth of the DER installations, in particular DG, with growing impact on the grid, even though critical cases with high node voltages or high short circuit currents were found only occasionally. In these periods, the regulatory requirements imposed to switch off the DG immediately in case of fault in the distribution network. The Standard IEEE 1547 [31] first introduced the technical specifications criteria and requirements applicable to all DER technologies with aggregate capacity of 10 MVA or less at the point of common coupling, interconnected to distribution systems. The Standard IEEE 1547 set up the basis for further regulatory developments at the national and International level. In the following years, the DER diffusion had an advanced growth, also because of significant incentives to promote RES generation for DG, and the number of cases reaching technical limits for the DER diffusion increased considerably. In particular, the overall effect of the disconnection of the DG after a fault was seen at the transmission system level as the sudden steep increase of the net load, which could lead to critical situations for the security of the national electrical systems, impairing the stability of the transmission system. On 18 July 2011, a letter from the President of ENTSO-E launched the alert on the security issues caused by the automatic frequency disconnection settings of PV systems, encouraging the national authorities to implement remedial actions. This letter started several actions that led to modify the national standards, passing, in case of fault in the distribution network, from the situation of “must switch off” DG to the opposite need for keeping DG connected in parallel with the grid also in emergency and grid restoration conditions, with the introduction of the so-called fault ride-through capability limits. In practice, the protection systems at the grid interface and the inverters had to become less sensible to frequency variations, to keep the local systems connected to the grid also for frequency variations between 47.5 and 51.5 Hz. After this radical change in the way to consider DER connection to the networks, most national standards were upgraded, and included the explicit definition of the user (e.g., the subject that uses the network to inject and/or take electricity), with its

Overview of distributed energy resources in the context of local integrated energy systems

classification into active user (for production) when the user exploits rotating or static equipment to convert any form of useful energy into electricity and is connected to the network, and passive user otherwise (the specific standards contain more refined definitions, also indicating possible exceptions). The national standards also indicated the details for the settings of the protection devices, while actions for global harmonization of the standards at the international level are in progress, and the Standard IEEE 1547 series was enriched with more detailed documents during the years. In particular, the standards define the feasibility of islanding operation of a portion of the grid, in which the islanded part of the grid is only supported by DER. Islanding may be nonintentional (i.e., with the formation of an island after an outage), or intentional (allowing independent operation of the island during an interruption of the external supply). The main issue is that, in general, the DER could not ensure adequate frequency and voltage support, stability and quality of supply. In particular, the possible voltage support from DG units is limited by the reactive power capability of the local generation system. For this reason, nonintentional islanding is typically prohibited, and the nonintentional islands have to be detected and eliminated as fast as possible, also to avoid the presence of energized portions of the network during the maintenance carried out in the service restoration process. In case of intentional island, the DER connected to the island have to satisfy the loads and the load variations without experiencing dynamic problems in their voltage and frequency control systems.

1.3.2 Microgrids and local energy networks The development of local energy systems is challenging the paradigm of providing the main supply from large distribution networks managed by the DSO. Further paradigms currently in use are: • Virtual Power Plants (VPP): coordinated control and possible optimization of the operation of the resources in a local system, based on a framework conceptually established in 1997 [32]. • Microgrids: possibility of constructing and operating small distribution systems in a way independent of the grid [33,34]. • Nanogrids, as smaller microgrids supplied in direct current [35], which serve a local system, even a single building [36]. The nanogrids may also be interconnected among them or with larger systems [37]. More details are provided in Chapter 8 of this book. • Web of cells: a decentralized control scheme is applied to portions of the network that can operate in grid-connected or islanded modes [38]. More details are provided in Chapter 2 of this book. In the approaches mentioned above, the connection with the main grid is not necessarily absent, and may be considered especially in emergency conditions. However,

9

10

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

the design of the network has to be carried out in such a way that the operation and survival of the local energy system does not depend on the connection to the distribution system. In this respect, the availability of local energy systems is significant for reducing the vulnerability of the electricity system in case of major events or deliberated malicious attacks. The trend to developing micro-energy systems could raise the concern about the progressive reduction of the users that remain connected to the distribution grid, leading to grid defection issues [39]. This aspect is crucial and depends on the value the users give to the service provided by the grid, mainly linked to reliability aspects, for both electricity and gas networks [40]. The main objectives of VPPs are to enhance the visibility of the DER, provide suitable interfaces among the local components, activate distributed control strategies, promote the adoption of ICT solutions, address the optimal use of the available capacity, and study the interactions with the energy markets. Specific aspects of the VPP include the definition of characteristics similar as a plant connected to the transmission system (with possible meshed structures used during operation and the design of appropriate control and protection schemes), the management of a portfolio of DER, the establishment of generation schedules, the definition of the internal operating cost structure (which is a private information of the owner), and the possible provision of system services. The VPP concept is typically dedicated to control aspects, and does not necessarily imply the presence of a local energy system network independent of the distribution network. Conversely, microgrids are small distribution systems containing generation, load, and storage, which can operate totally separated from the main distribution system (autonomous mode) or connected to it (nonautonomous mode). A microgrid is conceptually different with respect to intentional islands. In fact, intentional islands are normally connected to the network and may operate separately only when required (e.g., to assist the system restoration after a fault). On the other hand, microgrids operate normally as independent systems and may be connected to the network (in nonautonomous mode) in emergency conditions, to improve the continuity of supply after a fault in the network or microgrid, or according to economically convenient electricity prices for energy or reserves [41]. The operation of autonomous micro-grids is critical, because of possible voltage, frequency control, or stability issues. The demand for and supply of energy within the microgrid is monitored from a control center to optimize the use of DER. Appropriate redundancy and diversification in the energy supply mix is beneficial to reduce the critical cases as much as possible. The main advantages of the microgrids are high availability, possible combined management of the energy mix (electricity, heat, cooling, gas, water, hot water, etc.), modular operation planning, possible resource optimization inside the microgrid, synergies for personnel resources, primary energy purchase, maintenance, billing, and supply services, possible increase in

Overview of distributed energy resources in the context of local integrated energy systems

economic efficiency, possible integrated energy bill with fixed cost reductions for the consumers and higher social acceptability, as well as reduced vulnerability of the local energy system due to the presence of many local resources. For these reasons, the microgrids are well suitable to be used in local energy systems, such as rural communities far from the existing power grids, industrial users with many local and scattered sites, industrial parks, island communities, power providers in developing countries lacking of infrastructure, and energy communities. One of the crucial points to ensure viable operation of the local energy systems is to guarantee appropriate coordination among the control systems of the active users. The main operational issues occur when the local generation units are owned and managed independently of each other. This is likely to happen, because the objectives of managing the local units (maximizing the efficiency and profitability of the local system) are generally conflicting with those of a central coordination (e.g., network losses minimization, or voltage support optimization). As a consequence, there could be little or no coordination among the controls, and the local unit protections could not interact within an overall protection scheme, thus creating possible conflicting situations inside the microgrid. This noncoordinated situation could cause substantial worsening of the microgrid operation, e.g., for the voltage profiles. As such, a centralized control strategy would be advisable. An alternative solution is to establish a decentralized control strategy, in which each user is represented as an individual entity (e.g., an agent) that follows its own objectives. In this case, the interactions among the agents have to be structured very carefully, to keep the system operation on the correct line. A further issue is the low inertia that could exist in microgrids due to the many inverter-based microgrid interfaces. Microgrid control strategies can be classified into three levels [42]: (1) the primary control, based on local output variables (voltages, currents, and frequency), which provides power sharing among the generation units through droop control; (2) the secondary control, based on the DER operating status and forecasts, which sends commands to the loads and dispatchable units; and (3) the tertiary control, based on RES and price forecasts, which sends long-term set points to coordinate the operation of the microgrid and of the distribution system to which the microgrids are connected. For nanogrids and web of cells, the reduction of the scale of the system and the presence of a small number of DER connected makes it crucial to solve the issues referring to the controllability of such small systems by using refined power electronics and appropriate control systems.

1.3.3 Integration among energy networks In general, the energy networks of different energy carriers do not share all the nodes, and the paths to reach the nodes are different [43]. This happens mainly because of

11

12

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

the independent planning of the energy networks. However, even in the prospect of planning new multi-energy networks, maintaining different paths is a positive aspect to reduce the energy system vulnerability and improve resilience in case of extreme events, by guaranteeing different paths to reach the users. Interactions between the electrical network and other networks are enabled by the presence of nodes with multi-energy systems, e.g., with MES with or without storage, district heating networks, district cooling networks, or power-to-X solutions. Multiperiod multi-energy scheduling of coupled electrical, heating, and natural gas energy networks can be formulated [44]. In particular, “smart” gas networks have two main features: (1) the presence of smart metering and control systems, and (2) the possibility to allow the injection of nonconventional gases in the pipelines (without exceeding the related limits).

1.3.4 Analysis and optimization of the grid operation with local energy systems Local energy systems are generally characterized by smaller networks with respect to large distribution systems. This simplifies some issues referring to the network analysis. In particular, for relatively small networks with meshed structure and radial operation, it is possible to determine all the radial network structures by using suitable computational techniques [45]. In this way, the optimization problems based on network reconfiguration can be solved by finding out the global optimum, without the need of executing dedicated algorithms of numerical computation or metaheuristics. This possibility is viable for numbers of radial configurations of the order of millions, however there could be local energy networks with more complex structures in which the exhaustive search approach is not effective because of excessively long computation times. In these cases, the dedicated algorithms mentioned above are needed. More importantly, in the presence of RES it becomes essential to model the effects of uncertainty. These effects can be addressed by: 1. executing Monte Carlo analyses, in which the statistical characteristics of the uncertain variables are represented together with the availability of the components, solution methods such as the probabilistic power flow based on a deterministic algorithm are run for a high number of times, and the statistics of the results provide the final outputs; 2. adopting stochastic programming models in the solver [46]; or 3. resorting to robust optimization methods, when the parameters of the problem are known only within given bounds [47]. When the uncertainty on the input parameters becomes too high, the use of probabilistic methods is not justified, because the standard deviation of the results would be so high to make the whole analysis substantially meaningless. The large-scale

Overview of distributed energy resources in the context of local integrated energy systems

uncertainty that occurs in this case [48] is tackled by using scenario analyses. In practice, to represent different but foreseeable alternatives for the uncertain input variables, the analyst constructs a number of scenarios. Each scenario is then assigned a weight that represents its relative importance. Detailed aspects of RES uncertainty are discussed in Chapter 11. The optimization methods are defined by their objective function(s) and constraints. In the presence of many objectives, it is possible to: 1. Formulate an overall single objective function, in which all objectives are represented with comparable quantities (e.g., an economic variable, by assigning a unity cost and an appropriate weight to each objective), and these quantities are summed up together. 2. Consider the objective functions separately, without merging them into a unique objective function. The objective functions are then handled through Pareto analysis, by identifying the nondominated solutions that can be seen as compromise solutions. Multiobjective operational optimization with DER for integrated energy systems is discussed in Chapter 4 of this book. The typical problem for microgrid operation is microgrid scheduling. The objective (to be minimized) is the operation cost of local DERs, including the power exchange with the grid, to supply the microgrid loads in a certain period of time (typically one day). Multiobjective formulations are possible (e.g., considering cost and greenhouse gas emissions as two objectives). The microgrid loads and RES-based local generations are determined from forecasts. The major challenges refer to (1) the management of uncertainty, as microgrids are relatively small, and having a good prediction of RES what happens in a small area is not simple; (2) the coupling in time introduced by unit commitment (e.g., ramping constraints) or storage management; and (3) the use of possible new structures such as loop-based or meshed microgrids, which needs redesigning the protection and control systems. In case of connection to the distribution system, the possibility of reaching the condition in which the local system injects power into the distribution system, causing a reverse power flow in case of excess of local generation, has to be considered. Details of load forecasting for shortterm DER scheduling are presented in Chapter 12 of this book. Where possible, optimization has been formulated as a linear problem. In this case, the existing solvers allow obtaining the true optimal solution, typically without scalability limits for relatively small local systems. In more general problems, lack of convexity of the domain, the presence of integer variables, and nonlinear formulations are more challenging, and require the adoption of suitable solvers. A relevant entity for the management of a local energy system is the aggregator. The task of the aggregator is to build a portfolio of clients (consumers or prosumers) that are globally able to provide a power profile with tradable bands of consumption (or local production), convenient to access profitable market options [49]. In general, a

13

14

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

multienergy player may serve as a DER aggregator that manages the interactions between the wholesale energy market and the local energy systems [50]. In some cases, the aggregators could be dedicated to a single type of users (e.g., EVs) to satisfy specific objectives, such as minimizing the difference between the real and the target load profile, or minimizing the EV charging costs. Depending on the control strategy applied to the EVs, the aggregator can apply (1) a centralized approach based on direct control, in which the central controller knows all the problem variables and constraints, or (2) a distributed approach based on incentives, in which the aggregator sends signals and information on the rewards to the controlled EVs.

1.3.5 Provision of grid services In local energy systems, the provision of grid services is crucial for the viable operation of the system itself. Different situations occur depending on whether or not the local energy network can be connected to the main grid in emergency conditions. In any case, essential grid services are voltage control, frequency control, system stability support (also taking into account inertia-related aspects and possible needs for generation or demand curtailments), as well as power conditioning for ensuring adequate quality of supply. Reliability aspects and the provision of reserve services depend on the possible connection to the main grid, otherwise they have to be procured through local resources, based on their adequacy [51,52]. Advanced services such as energy shifting and operational flexibility depend on the availability of the local resources and on the presence of specific programmes to manage DR and flexibility services. In this book, a detailed analysis on the flexibility services provided from DER is presented in Chapter 6, while Chapter 9 discusses how to enhance flexibility from DER through the adoption of demand response programmes. With multienergy demand, thermally activated technologies and storage contribute in enhancing the operational demand-side flexibility. The addition of electric heating systems (such as heat pumps and auxiliary resistance heaters) provides further variability to the available equipment [53]. The key concept is energy shifting from multienergy sources, determining by starting from the operational baseline, which can be businessas-usual, or calculated as the result of an optimization. Starting from the baseline, it is possible to determine how much electricity input variation can be achieved through energy shifting. If a change (i.e., reduction) in the electricity input to the demand side is requested, the local consumer (or prosumer) could avoid to curtail its electrical demand, identifying some actions to make in the local system to serve the multienergy demand through different energy sources through suitable energy converters, or through energy storage systems (if available). With energy shifting, it is possible to identify solutions in which the multienergy demand is even not curtailed. Similar (appropriately adapted) concepts may be applied in case an increase in the local

Overview of distributed energy resources in the context of local integrated energy systems

electrical demand when requested. In the present situation, the participation of consumers or prosumers in the provision of smart energy services depends on their willingness, driven by profitability aspects. As such, it is key to determine how much different energy services may become profitable, and to what extent. Profitability depends on the difference between revenues and costs. The energy shifting operations have a cost, because the operation point moves away from the initial point. Getting cost information from the demand side enables the service providers formulate viable incentives to enable sufficient rewarding, to attract the consumer of prosumer participation to provide specific energy services. The possibility of shifting energy in a MES is aimed at reducing the input electricity without changing the multienergy demand, that is, by maintaining the same level of comfort for the users. For a given MES, the electricity shifting potential (ESP) has been defined as the maximum electricity reduction available from the local multienergy system [54]. In addition, the maximum profit electricity reduction (MPER) is the electricity input reduction that corresponds to the maximum profit [55]. While the ESP is a purely technical outcome, MPER depends on the revenues associated to the provision of the electricity reduction service; the MPER condition could be found for an input electricity reduction between zero (in case of nonconvenience to reduce electricity) and the ESP. In local energy systems, the uncertainty plays a major role also in the definition of the demand. The demand patterns of the users depend more on local characteristics, and an approach based on general-purpose load profiles is less useful than for large systems. If generation and demand are considered together to for the net power demand, when the number of users is low, the aggregate power patterns are highly variable, and the definition of prosumer profiles could become poorly consistent. For a categorization of the prosumers, probabilistic load profiling techniques are appropriate [56], in which the separate profiling of demand and generation (with storage) could be convenient. Flexibility in power and energy systems refers to the possibility of using the available resources to respond in an adequate way to demand and generation variations during time (taking into account the corresponding uncertainty), at acceptable costs. It is achieved by using components that can adapt their operation and move their production or consumption to different time intervals. For power system applications, operational flexibility has been defined as the technical ability of a power system unit to modulate electrical power feed-in to the grid and/or power out-feed from the grid over time [57]. On the generation side, the approaches used to assess flexibility exploit unit commitment and economic dispatch, with stochastic optimization used to incorporate uncertainty. The limits are based on maximum capacity, minimum stable generation, and up/down ramp rates of generators. Four metrics have been used, namely, the power provision capacity, the energy provision capacity, the power ramp-rate capacity, and the ramp duration [58].

15

16

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

On the grid side, flexibility has been defined as the ability of a power network to deploy its flexibility resources to cope with volatile changes of power system state in operation [59]. The main issue is that the increased uncertain RES generation may cause higher risk of network congestion. The congestion increases the operating costs and limits the usage of available flexibility from generation-side and/or demand-side resources. On the demand side, the definitions of flexibility depend on evaluations carried out for individual appliances or a load aggregation. For individual appliances, flexibility depends on the consumers’ preferences, and the related data are typically gathered from questionnaires and surveys. For aggregate loads, the flexibility assessment is carried out by using different approaches, considering sensitivity functions of shifting the demand by a given time, unit commitment optimization, simulations of the price demand elasticity, or statistics on the demand variations at successive time steps for aggregate residential loads [60]. In energy community districts, flexibility can be assessed by using a stochastic model for demand response resources [61]. Available flexibilities may come from different sides. The main contribution comes from energy vector substitution, also with the presence of an auxiliary gas boiler. The integration of more equipment such as electric heat pumps and storage (from batteries and thermal storage units) provides wider room for more flexible management of the energy mix. Flexibility is assessed by avoiding that the thermal comfort of the occupants falls below agreed limits, for example on the internal building temperature. End-service curtailment and power factor manipulation are further means to increase flexibility [62].

1.4 Emergent paradigms and solutions Addressing climate change and environmental problems represents the major driver for national energy policies worldwide due to the compelling need to decarbonize the energy supply systems and foster the energy transition. With specific reference to the European context, the European Union (EU) set ambitious environmental and energy goals to design a low-carbon energy system by the middle of the 21st century. The targets established for 2030, which envisage a 40% reduction in greenhouse gas emissions (from 1990 levels), 32% share for renewable electricity and 32.5% improvement in energy efficiency, can be achieved by developing energy systems supporting the implementation of three primary goals, namely protecting the environment, creating affordable and market-oriented energy services, and ensuring security, reliability and resilience of the energy supply. These targets become even more ambitious for 2050, with the Energy Roadmap 2050 of the European Commission and the Energy Union strategy supporting the aim of fully decarbonizing the European economy, by reducing greenhouse gas emissions

Overview of distributed energy resources in the context of local integrated energy systems

in developed countries below 80% 95% of 1990 levels by 2050. In such a context, the ETIP SNET1 Vision 2050 [63] is a low-carbon, secure, reliable, resilient, accessible, cost-efficient, and market-based pan-European integrated vision for an energy system that supplies the whole economy with a fully CO2-neutral and circular approach by the year 2050. According to this vision, European citizens are the main actors in the transition from existing fossil fuel-based energy systems to an integrated, lowcarbon and cost-effective energy system by 2050. Energy systems become integrated infrastructures for all energy carriers with the electrical system as a backbone, characterized by a high level of integration between all networks of energy carriers, coupling electrical networks with gas, heating, and cooling networks, supported by energy storage in all forms and conversion processes. Their main feature will be the involvement of the end user in the management of the system itself. Citizens become active consumers and prosumers, using local and user-friendly energy exchanges, as well as peerto-peer exchanges, for a wide range of services and optimal energy prices. Moreover, their active role is fully implemented in the mechanisms of demand side management, through which the user is made participant in the management of network contingencies, as well as in applications such as renewables self-consumption and energy communities. Furthermore, these integrated energy systems are characterized by the advent of distributed poly-generation fully supplied by RES. In such a framework, storage in all its forms and types plays a crucial role and the locally available energy resources are used for their full economic potential, reflecting the upgrading needs of the power transmission and distribution networks and also contributing to maximizing the resilience of supply channels for heating and cooling needs. In line with this scenario, on 30 November 2016, the European Commission published the Clean Energy Package for all Europeans, that is the set of initiatives and directives aimed at making the EU more competitive in the energy transition and redesigning the profile of the European electricity market. It is therefore clear that the EU is calling for a paradigm shift in the energy area by promoting emergent solutions which are prosumer-oriented placing the citizens at the heart of energy transition towards customer empowerment and engagement and local energy systems.

1.4.1 Self-consumption and self-sufficiency Although the self-consumption concept meaning final users that consume the energy they produce on site is not a new concept, it has been re-defined in both the recast of 1

European Technology & Innovation Platforms (ETIPs) have been created by the European Commission in the framework of the new Integrated Roadmap Strategic Energy Technology Plan (SET Plan) by bringing together a multitude of stakeholders and experts from the energy sector. The ETIP Smart Networks for Energy Transition (SNET) role is to guide Research, Development & Innovation (RD&I) to support Europe’s energy transition, with the key mission to set-out a vision for RD&I for Smart Networks and engage stakeholders in this vision.

17

18

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

the Electricity Market Directive 2019/944/EU—EMD II [64] and the revised Renewable Energy Directive 2018/2001/EU—RED II [65]. In both cases, in the framework of self-consumption, final users are allowed to consume and store the electricity produced on site and sell this electricity. The main difference between the two definitions is that the active consumers defined in EMD II can perform activities beyond energy generation such as participation in flexibility and energy efficiency schemes through selling self-generated electricity, whereas the renewable self-consumers defined in RED II are limited to produce electricity from RES. An emergent concept is instead represented by the collective self-consumption, which has been receiving a great attention in recent years thanks to the increased financial viability of self-consumers and the opportunities deriving from electricity sharing between producers or self-consumers [66]. This concept has been formally recognized at EU level through the Clean Energy Package. In fact, in the EMD II, the concept of active consumers includes groups of jointly active consumers, whereas in the RED II, the jointly acting renewable self-consumers are defined separately as groups of renewable self-consumers located in the same building or multiapartment block. While individual self-consumption is possible in most EU countries, collective self-consumption is an emerging concept and only a few countries such as France, Germany, Austria, Belgium, Spain, Slovenia and Switzerland have already put forward legal frameworks to allow this new form of self-consumption [67]. In these promotive countries, collective self-consumption mostly refers to multifamily houses and mixed use of offices and/or small medium enterprises (SMEs), and in some cases this form of energy sharing is also allowed between different buildings.

1.4.2 Development of local energy markets According to the EU policy and vision supporting the growing of DER penetration levels in decentralized energy systems [68], the traditional electricity market faces a number of challenges in integrating emerging technologies such as distributed generation, storage in all forms and demand flexibility. This requires a paradigm shift towards a new market design, which is able to exist in the current market structure, while also guaranteeing cooperation among local generation, storage, and demand response, as well as participation of customers to the market through peer-to-peer energy trading. One of such electricity markets is the local energy market, which is intended as a micro-market that integrates prosumers and consumers as well as storage facilities into the energy supply system [69]. In local energy markets, customers with own generation assets are able to consume the self-generated energy through the optimal usage of DER, thereby becoming prosumers who actively manage their energy needs. In addition, they are also able to trade the excess of generated energy within the boundaries of the energy community through peer-to-peer energy trading [70]. Therefore, the

Overview of distributed energy resources in the context of local integrated energy systems

development of local energy markets represents another example of emergent paradigm of the EU policy towards the customer’s empowerment and engagement, through which the customer becomes aware of efficient energy behaviors and takes parts to demand side management actions. The local energy market development brings a number of benefits also to system operators, by reducing the need of new investments and reinforcements thanks to the higher system’s flexibility and a more efficient operation of the overall network. The evolution from a traditional centralized energy model to a new decentralized and customer-centric energy model, which is behind the development of local energy markets, also allows achieving a higher market transparency and a more balanced allocation of benefits among the different stakeholders. Moreover, this new form of market facilitates the growth of RES penetration in power networks, which is in line with the EU energy and environmental objectives. The development of local energy markets may also take benefits from the definition of a transactive energy system, which include “a set of economic and control mechanisms that allow the dynamic balance of supply and demand across the entire electrical infrastructure using value as a key operational parameter [71].” The GridWise Architecture Council was formed by the U.S. Department of Energy to promote and enable interoperability among the many entities that interact with the electric power system. Four areas of development have been identified as (1) policy and market design, (2) business models and value realization, (3) conceptual architecture guidelines, and (4) cyber-physical infrastructure. The operation of local energy markets within a transactive energy framework is a key topic under discussion [72].

1.4.3 The energy community paradigm The ongoing energy transition brings new opportunities for DER integration and deployment and for the evolution in the role of the final users from passive consumers to active consumers to achieve common goals such as reduction of energy costs, environmental impacts and dependence on the national grid. As compared with traditional centralized energy systems, decentralized local energy systems enable self-sufficiency and sustainability of energy supply through the exploitation of local RES available at local level, the improvement of energy efficiency by combining different sectors such as heat, cooling, electricity and transport, and the reduction of line losses. These systems can also potentially contribute to the achievement of the EU energy and climate objectives, as also highlighted in the Clean Energy Package, where energy communities are recognized as a sustainable way to manage energy at local level, with or without the connection to the distribution grid. The energy community can be seen as a set of energy users deciding to make common choices with the aim to maximize the benefits deriving from this collegial approach, thanks to the implementation of a variety of power and heat technologies and energy storage solutions and the optimized management of energy

19

20

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

flows. Indeed, energy communities are well-placed to meet local energy needs, reduce the need for transmission infrastructure, and bring people together to achieve common goals for well-being. DERs within an energy community can also be employed to support the grid by providing ancillary services and flexibility through different products such as demand response, congestion relief, local energy markets, etc. The Clean Energy Package introduced the energy communities into the EU legislation, by including two definitions: the concept of Citizen Energy Community (CEC), contained in the EMD II, Directive 2019/944/EU [64], and Renewable Energy Community (REC), contained in the RED II, Directive 2018/2001/EU [65]. Both these types of energy communities are legal entities, must be effectively controlled by their shareholders or members and have the primary goal to guarantee environmental, economic, and social community benefits rather than financial profits. In detail, the CEC is defined as a legal entity with the following characteristics: • It is based on the voluntary and open participation of individuals, local authorities or small enterprises; • It has the main goal to offer its members or partners or the territory where operates, environmental, economic or social benefits, without generating any financial profits; • It can participate in the services for generation, including from RES, distribution, supply, consumption, storage, energy efficiency or recharge for electric vehicles or provide other energy services to its members. The CECs must be able to operate on the market on equal and nondiscriminatory terms with respect to the other parties, being able to freely assume the roles of end-customer, producer, supplier or manager of distribution systems. With regard to self-produced electricity consumption, the Directive ensures that CECs are treated as active customers. As for the REC, it is defined as a legal entity with the following characteristics: • It is based on open and voluntary participation, which is autonomous and effectively controlled by shareholders or members who are located in the vicinity of the renewable energy production plants that belong to and are developed by the legal entity in question; • Its shareholders or members are natural persons, SMEs or local authorities, including municipal administrations; • It has the main goal to provide community, environmental, economic or social benefits to its shareholders or members or to the local areas where operates, rather than financial profits. RECs must have the right to produce, consume, store, and sell renewable energy. They are also able to exchange, within the community, the renewable energy produced and access all the appropriate electricity markets, directly or through aggregation, in a nondiscriminatory way [73]. Fig. 1.1 provides an overview of the main features of CECs and RECs based on a comparative analysis on their main activities.

Overview of distributed energy resources in the context of local integrated energy systems

Figure 1.1 Comparative analysis of the main activities of CECs and RECs ([73]—Deliverable developed under the scope of the COMPILE project: integrating community power in energy islands, H2020 824424).

Beyond the differences between CECs and RECs, the primary objectives of both these communities are summarized below: • promoting public acceptance and the development of renewable sources on a decentralized level; • promoting energy efficiency projects; • promoting the participation in the market of users who otherwise would not have been able to do so; • allowing the supply of energy at affordable prices; • combating vulnerability and energy poverty, reducing energy supply costs, and consumption while promoting efficiency. With reference to the enabling technologies for energy communities, the creation of this innovative energy paradigm requires the adoption of distributed generation technologies and technological solutions for the smart management of energy flows and related information. Each of the enabling technological solutions can be characterized in terms of functionality, degree of centralization and level of technological maturity. With reference to the functionality that the specific technology performs in an energy community, it is possible to identify three different categories [74]: • Production and use of energy: This category includes all the technologies allowing to produce the energy needed by the community users on site and to consume this energy efficiently. This category is in turn divided into two subcategories, namely local production and demand-side flexibility, as discussed below. • Management, control, and monitoring of energy flows: this category includes technologies allowing to remotely control the production, distribution, storage, and consumption of energy assets present within the community and to control and monitor energy flows.

21

22

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

These technologies are divided into software and hardware systems. The software systems allow, in the forecast phase, to elaborate the forecasts of energy consumption by the community’s users, and of production by the plants powered by nonprogrammable RES. Moreover, they allow planning the optimal operation strategies of the production, storage and energy consumption assets. During the operation phase, these technologies allow to optimize the operation of the technologies on the basis of the actual operating conditions. The hardware systems for managing, controlling, and monitoring energy flows contribute to the governance of the community, imparting the relative operating methods on the basis of the choices made by the management software and the on-site measurement of the main operating parameters of the community. • Distribution of energy and information flows: this category includes the technologies allowing the distribution of energy and information flows among the production, distribution, storage and consumption assets within the community, and the related management systems. In detail, this category includes the physical networks of electricity and thermal energy distribution (district heating network) and the communication infrastructure that enables the exchange of information between the various assets of the community to ensure their correct functioning. The degree of centralization refers to the field of application of the technology, namely the applicability at a single energy user level (building) and/or at the community level serving multiple energy users. Table 1.1 shows the categorization of the Table 1.1 Enabling technologies for an energy community (based on [74]). Category

Technology Building level

Community level

Local production

Micro-CHP Reciprocating engines Internal combustion engines Fuel cells Solar thermal collectors Rooftop PV Microwind Heat pumps Absorption chiller

Community-CHP Reciprocating engines Internal combustion engines Biomass generators Fuel cells Solar thermal plant Solar PV Wind farm

Demand side flexibility

Flexible loads (appliances) Electric vehicles Thermal and electric storage Battery energy management system Home/building energy Management system

Distributed storage Community battery energy Management system Community energy management system

Overview of distributed energy resources in the context of local integrated energy systems

technological solutions enabling an energy community according to the degree of centralization. The maturity level refers to the expected improvements of both technical and economic performances of the technology as compared to the current performances. Figs. 1.2 and 1.3 show the graphic representations of the technologies for the local energy production and for the flexible demand management, respectively, according to the degree of centralization and technological maturity.

Figure 1.2 Graphical representation of technologies for local energy production according to the degree of centralization and technological maturity (based on [74,75]).

Figure 1.3 Graphical representation of technologies for flexible demand management according to the degree of centralization and technological maturity (based on [74,75]).

23

24

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

In the context of an energy community, many actors can play with different interests and objectives, which can often conflict with each other. For example, community’s users want to have low-cost energy, aggregators try to maximize the value of users’ flexibility in the various market options, while policy-makers are interested in achieving a sustainable and environmental-friendly energy supply, in order to promote the energy transition. Table 1.2 provides a detailed summary of the various actors with their interests, roles, and responsibilities in the energy communities. Table 1.2 Main actors and interests, roles, and responsibilities in energy communities (based on [74]). Actors

Main interests

Roles and responsibilities

Community’s users

Use of affordable and clean energy on site Reduction of energy costs and energy bills

Energy management, local energy exchange, flexibility management

Energy producers Energy suppliers ESCOs

Investment opportunities in local energy systems (profit maximization) Opportunity of profit resulting from a deficit in energy supply Possibility of profit deriving from the implementation of energy efficiency measures and the optimized management of local generation Possibility of profit deriving from the sale of technologies to transform the existing energy landscape both in terms of production and consumption Definition of a business model to generate profits through maximizing the value of flexibility in the various market options Portfolio optimization, balance energy procurement at lowest cost

Technology providers

Aggregators

Balance Responsible Parties Transmission system operators Distribution system operators

Policy makers and regulators

Possibility of larger balance between supply and demand at lower cost for consumers Distribution of energy to users through the use of a safe and reliable network Avoid grid congestion, postpone investments in the grid, balance of energy islands Ensure affordable energy supply for all users Promotion of a sustainable supply system, transition to a low-carbon energy system, and energy security

Supply the deficit energy, energy procurement Financing, supply, and installation of energy efficiency projects Provision of technologies for local generation, flexibility, and energy management systems Aggregate flexibility from community’s users Incorporate flexibility in portfolio Use flexibility for system balance Grid operation, local congestion management

Investments and incentives for energy communities, reduction of barriers

Overview of distributed energy resources in the context of local integrated energy systems

1.4.4 Technical, regulatory, and social barriers The main barriers that hinder an effective deployment of the emergent paradigms and solutions described above are mainly attributable to technical, regulatory, and social aspects. From the technical point of view, one of the main issues is characterized by the different maturity level of DER technologies involved in such emerging solutions. Indeed, some of them are not fully technical-mature and have not yet been widely adopted [70]. Another key technical issue is represented by smart metering, and the different deployment level of the smart meters rollout in European countries, which defines the need to increase standardization in metering schemes [70]. Other barriers are also related to interoperability issues related to communication and control aspects, which result from the large variety of system management and control software options operating for instance in the context of energy communities. In order to overcome these issues, it is needed to make all these systems compatible with components and technologies within an energy community, in order to enable the proper connectivity and efficient operation of all system components. An additional technical challenge is related to the high level of stochasticity of local RES generation and demand response in the context of energy communities. Especially, this latter depends on the time-varying energy demand, the weather and user behavior which is difficult to predict [74]. Moreover, in the context of an energy community, local RES generation not only shows intermittency by nature, but depends also on the choice of use by the plants’ owners. A practical example is the owner of a rooftop PV who may want to transfer excess electricity to the neighbor and not sell back to the grid, thereby causing additional stochasticity. Last but not least, another key technical challenge is related to the collection and exchange of large amounts of data and information, which in most cases are sensitive and of confidential nature, and secure data handling and cyber security become essential for a correct data exchange within energy communities. Beyond technical aspects, the evolution of the regulatory framework in EU countries represents the most relevant factor for an effective deployment of the emerging paradigms discussed above. The absence of clear regulations on energy communities and legal options for collective self-consumptions, as well as the lack of incentives to set up jointly acting renewable self-consumer projects represent the major barrier to remove. Moreover, considering the innovation potential of these solutions and the new business and financial models which are going to emerge, the regulatory framework should also account for the innovation and experimentation aspects, integrating provisions for a periodic evaluation of these emerging solutions, in order to guarantee improvements in the relevant legislation [76]. The regulation aspects represent the major barrier to overcome also for the development of local energy markets and the effective utilization of DER and/or RES and customer empowerment.

25

26

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Finally, social aspects related to user awareness and acceptance are extremely important for the effective implementation of these emerging paradigms and solutions. Indeed, if on the one hand it is evident that the customers’ engagement represents a vital step to implement all these solutions, on the other hand it is most probable that they show a certain resistance to change their role in the energy landscape from passive to active consumers. The main challenge that arises is the quantification from the customers’ side the intangible benefits related to these solutions such as the environmental impact reduction, the promotion of local growth with the opportunities of local jobs and the possibility to achieve a more balances allocation of the benefits deriving from these solutions.

References [1] R.C. Dugan, S.A. Thomas, D.T. Rizy, Integrating dispersed storage and generation (DSG) with an automated distribution system, IEEE Trans. Power Appar. Syst. PAS-103 (6) (1984) 1142 1146. [2] H. Kirkham, D. Nightingale, T. Koerner, Energy management system design with dispersed storage and generation, IEEE Trans. Power Appar. Syst. PAS-100 (7) (1981) 3432 3440. [3] C.W. Gellings, Demand-side load management, IEEE Spectr. 18 (1981) 49 52. [4] K. Heussen, S. Koch, A. Ulbig, G. Andersson, Unified system-level modeling of intermittent renewable energy sources and energy storage for power system operation, IEEE Syst. J. 6 (1) (2012) 140 151. [5] A. Zaltash, A.Y. Petrov, D.T. Rizy, S.D. Labinov, E.A. Vineyard, R.L. Linkous, Laboratory R&D on integrated energy systems (IES), Appl. Therm. Eng. 26 (2006) 18 35. [6] M. Geidl, G. Koeppel, P. Favre-Perrod, B. Klöckl, G. Andersson, K. Fröhlich, Energy hubs for the future, IEEE Power Energy Mag. (2007) 25 30. January/February 2007. [7] G. Chicco, P. Mancarella, Distributed multi-generation: a comprehensive view, Renew. Sustain. Energy Rev. 13 (3) (2009) 535 551. [8] P. Mancarella, MES (multi-energy systems): an overview of concepts and evaluation models, Energy 65 (2014) 1 17. [9] P. Mancarella, G. Chicco, Distributed multi-generation systems. Energy models and analyses, Nova Science Publishers, New York, 2009. [10] G. Chicco, S. Riaz, A. Mazza, P. Mancarella, Flexibility from distributed multienergy systems, Proc. IEEE (2020). in press. [11] M. Geidl, Integrated modeling and optimization of multi-carrier energy systems, Ph.D. dissertation, ETH Zürich, Switzerland, (2007). [12] F. Kienzle, P. Ahcin, G. Andersson, Valuing investments in multi-energy conversion, storage, and demand-side management systems under uncertainty, IEEE Trans. Sustain. Energy 2 (2) (2011) 194 202. [13] N. Neyestani, M. Yazdani-Damavandi, M. Shafie-Khah, G. Chicco, J.P.S. Catalão, Stochastic modeling of multienergy carriers dependencies in smart local networks with distributed energy resources, IEEE Trans. Smart Grid 6 (4) (2015) 1748 1762. [14] G. Chicco, P. Mancarella, A unified model for energy and environmental performance assessment of natural gas-fueled poly-generation systems, Energy Convers. Manag. 49 (8) (2008) 2069 2077. [15] J. He, Z. Yuan, X. Yang, W. Huang, Y. Tu, Y. Li, Reliability modeling and evaluation of urban multi-energy systems: a review of the state of the art and future challenges, IEEE Access. 8 (2020) 98887 98909. [16] J. Hinker, H. Knappe, J.M.A. Myrzik, Precise assessment of technically feasible power vector interactions for arbitrary controllable multi-energy systems, IEEE Trans. Smart Grid 10 (1) (2019) 1146 1155.

Overview of distributed energy resources in the context of local integrated energy systems

[17] C.W. Gellings, J.H. Chamberlin, Demand-side management: concepts & methods, The Fairmont Press, Inc., 1993. [18] F.C. Schweppe, M.C. Caramanis, R.D. Tabors, R.E. Bohn, Spot pricing of electricity, Kluwer Academic Publishers, 1988. [19] N. Pflugradt, U. Muntwyler, Synthesizing residential load profiles using behavior simulation, Energy Procedia 122 (2017) 655 660. [20] Q. Shi, C.F. Chen, A. Mammoli, F. Li, Estimating the profile of incentive-based demand response (IBDR) by integrating technical models and social-behavioral factors, IEEE Trans. Smart Grid 11 (1) (2020) 171 183. [21] N. Good, Using behavioural economic theory in modelling of demand response, Appl. Energy 239 (2019) 107 116. [22] W. Zhong, K. Xie, Y. Liu, C. Yang, S. Xie, Y. Zhang, Distributed demand response for multienergy residential communities with incomplete information, IEEE Trans. Ind. Inform. (2020). [23] V.A. Boicea, Energy storage technologies: the past and the present, Proc. IEEE 102 (11) (2014) 1778 1794. [24] Z. Gary Yang, New generation VRFB for electrical grid and utility applications, IEA workshop on “The Role of Storage in Energy System Flexibility”, Berlin, 29th October 2014, (2014). https:// www.iea.org/media/workshops/2014/egrdenergystorage/Yang.pdf. [25] D. Enescu, G. Chicco, R. Porumb, G. Seritan, Thermal energy storage for grid applications: current status and emerging trends, Energies 13 (2) (2020). [26] M.C. Argyrou, P. Christodoulides, S.A. Kalogirou, Energy storage for electricity generation and related processes: technologies appraisal and grid scale applications, Renew. Sustain. Energy Rev. 94 (2018) 804 821. [27] E. Guelpa, V. Verda, Thermal energy storage in district heating and cooling systems: a review, Appl. Energy 252 (2019) 113474. art. [28] S. Stinner, K. Huchtemann, D. Müller, Quantifying the operational flexibility of building energy systems with thermal energy storages, Appl. Energy 181 (2016) 140 154. [29] IEA-ETSAP and IRENA, Thermal energy storage. Technology brief E17, January 2013. (2013). ,https://www.irena.org/DocumentDownloads/Publications/IRENA-ETSAP%20Tech%20Brief% 20E17%20Thermal%20Energy%20Storage.pdf. [30] Y. Sun, S. Wang, F. Xiao, D. Gao, Peak load shifting control using different cold thermal energy storage facilities in commercial buildings: a review, Energy Convers. Manag. 71 (2013) 101 114. [31] IEEE, IEEE standard for interconnecting distributed resources with electric power systems (Standard 1547), standards coordinating committee 21, The IEEE, New York, NY, 28 July 2003, (2003). [32] D. Pudjianto, C. Ramsay, G. Strbac, Virtual power plant and system integration of distributed energy resources, IET Renew. Power Gener. 1 (1) (2007) 10 16. [33] B. Kroposki, R. Lasseter, T. Ise, S. Morozumi, S. Papatlianassiou, N. Hatziargyriou, Making microgrids work, IEEE Power Energy Mag. 6 (3) (2008) 40 53. [34] R.H. Lasseter, Microgrids, Proc. IEEE Power Eng. Soc. Winter Meeting, 2002, 1, (2002) 305 308. [35] I. Schönberger, R. Duke, S.D. Round, DC-Bus signaling: a distributed control strategy for a hybrid renewable nanogrid, IEEE Trans. Ind. Electron. 53 (5) (2006) 1453 1460. [36] R.P.S. Chandrasena, F. Shahnia, S. Rajakaruna, A. Ghosh, Dynamic operation and control of a hybrid nanogrid system for future community houses, IET Gen. Transm. Distrib. 9 (11) (2015) 1168 1178. [37] A. Werth, N. Kitamura, K. Tanaka, Conceptual study for open energy systems: distributed energy network using interconnected DC nanogrids, IEEE Trans. Smart Grid 6 (4) (2015) 1621 1630. [38] L. Martini, H. Brunner, E. Rodriguez, C. Caerts, T.I. Strasser, G.M. Burt, Grid of the future and the need for a decentralised control architecture: the web-of-cells concept, CIRED - Open. Access. Proc. J. 2017 (1) (2017) 1162 1166. [39] G. Chicco, Customer behaviour and data analytics, Proc. 2016 International Conference and Exposition on Electrical and Power Engineering (EPE), Iasi, 2016, (2016) 771 779.

27

28

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[40] A. van der Mei, J.P. Doomernik, Grid value and defection: a demand perspective, CIRED - Open. Access. Proc. J. 2017 (1) (2017) 2308 2311. [41] S. Parhizi, H. Lotfi, A. Khodaei, S. Bahramirad, State of the art in research on microgrids: a review, IEEE Access. 3 (2015) 890 925. [42] C.A. Cañizares, R. Palma-Behnke, D.E. Olivares, A. Mehrizi-Sani, A.H. Etemadi, R. Iravani, et al., Trends in microgrid control, IEEE Trans. Smart Grid 5 (4) (2014) 1905 1919. [43] E.A.M. Ceseña, P. Mancarella, Energy systems integration in smart districts: robust optimisation of multi-energy flows in integrated electricity, heat and gas networks, IEEE Trans. Smart Grid 10 (1) (2019) 1122 1131. [44] D. Xu, Q. Wu, B. Zhou, C. Li, L. Bai, S. Huang, Distributed multi-energy operation of coupled electricity, heating and natural gas networks, IEEE Trans. Sustain. Energy (2019). [45] H. Andrei, G. Chicco, Identification of the radial configurations extracted from the weakly meshed structures of electrical distribution systems, IEEE Trans. Circuits Systems. I, Regul. Pap. 55 (4) (2008) 1149 1158. [46] M. Shafie-khah, M. Vahid-Ghavidel, M. Di Somma, G. Graditi, P. Siano, J.P.S. Catalão, Management of renewable-based multi-energy microgrids in the presence of electric vehicles, IET Renew. Power Gener. 14 (3) (2020) 417 426. [47] C. Zhang, Y. Xu, Z.Y. Dong, Robustly coordinated operation of a multi-energy micro-grid in gridconnected and islanded modes under uncertainties, IEEE Trans. Sustain. Energy 11 (2) (2020) 640 651. [48] E. Carpaneto, G. Chicco, P. Mancarella, A. Russo, Cogeneration planning under uncertainty. Part I: multiple time frame approach, Appl. Energy 88 (4) (2011) 1059 1067. [49] S.R. Konda, A.S. Al-Sumaiti, L.K. Panwar, B.K. Panigrahi, R. Kumar, Impact of load profile on dynamic interactions between energy markets: a case study of power exchange and demand response exchange, IEEE Trans. Ind. Inform. 15 (11) (2019) 5855 5866. [50] M. Yazdani-Damavandi, N. Neyestani, G. Chicco, M. Shafie-khah, J.P.S. Catalão, Aggregation of distributed energy resources under the concept of multienergy players in local energy systems, IEEE Trans. Sustain. Energy 8 (4) (2017) 1679 1693. [51] W. Kim, J. Shin, J. Kim, Operation strategy of multi-energy storage system for ancillary services, IEEE Trans. Power Syst. 32 (6) (2017) 4409 4417. [52] P. Meneses de Quevedo, J. Contreras, A. Mazza, G. Chicco, R. Porumb, Reliability assessment of microgrids with local and mobile generation, time-dependent profiles, and intraday reconfiguration, IEEE Trans. Ind. Appl. 54 (1) (2018) 61 72. [53] D. Patteeuw, K. Bruninx, A. Arteconi, E. Delarue, L. Helsen, Integrated modeling of active demand response with electric heating systems coupled to thermal energy storage systems, Appl. Energy 151 (2015) (2015) 306 319. [54] P. Mancarella, G. Chicco, Real-time demand response from energy shifting in distributed multigeneration, IEEE Trans. Smart Grid 4 (4) (2013) 1928 1938. [55] P. Mancarella, G. Chicco, T. Capuder, Arbitrage opportunities for distributed multi-energy systems in providing power system ancillary services, Energy 161 (2018) 381 395. [56] Z.A. Khan, D. Jayaweera, M.S. Alvarez-Alvarado, A novel approach for load profiling in smart power grids using smart meter data, Electr. Power Syst. Res. 165 (2018) 191 198. [57] A. Ulbig, G. Andersson, Analyzing operational flexibility of electric power systems, Electr. Power Energy Syst. 72 (2015) 155 164. [58] Y.V. Makarov, C. Loutan, J. Ma, P. de Mello, Operational impacts of wind generation on California power systems, IEEE Trans. Power Syst. 24 (2) (2009) 1039 1050. [59] J. Li, F. Liu, Z. Li, C. Shao, X. Liu, Grid-side flexibility of power systems in integrating large-scale renewable generations: a critical review on concepts, formulations and solution approaches, Renew. Sustain. Energy Rev. 93 (2018) 272 284. [60] I.A. Sajjad, G. Chicco, R. Napoli, Definitions of demand flexibility for aggregate residential loads, IEEE Trans. Smart Grid 7 (6) (2016) 2633 2643. [61] N. Good, P. Mancarella, Flexibility in multi-energy communities with electrical and thermal storage: a stochastic, robust approach for multiservice demand response, IEEE Trans. Smart Grid 10 (1) (2019) 503 513.

Overview of distributed energy resources in the context of local integrated energy systems

[62] N. Good, E. Karangelos, A. Navarro-Espinosa, P. Mancarella, Optimization under uncertainty of thermal storage based flexible demand response with quantification of residential users’ discomfort, IEEE Trans. Smart Grid 6 (5) (2015) 2333 2342. [63] ETIP SNET, VISION 2050 - Integrating smart networks for the energy transition: serving society and protecting the environment, 22 June 2018 (2018). ,https://www.etip-snet.eu/etip_publ/etipsnet-vision-2050/. [64] European Union, Directive 2019/944/EU on the internal electricity market (2019). ,https://eurlex.europa.eu/legal-content/EN/TXT/PDF/?uri 5 CELEX:32019L0944&from 5 EN. [65] European Union, Directive 2018/2001/EU on the promotion of the use of energy from renewable sources (2018). ,https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri 5 CELEX:32018L2001&from 5 EN. [66] CEER, Regulatory aspects of self-consumption and energy communities. Counc. Eur. Energy Regulators, Ref: C18-CRM9_DS7-05-03 (2019). ,https://www.ceer.eu/documents/104400/-/-/ 8ee38e61-a802-bd6f-db27-4fb61aa6eb6a. [67] D. Frieden, A. Tuerk, J. Roberts, S. d’Herbemont, A. Gubina, Collective self-consumption and energy communities: overview of emerging regulatory approaches in Europe, Compile Working paper, June 2019 (2019). ,https://www.compile-project.eu/news/working-paper-collective-selfconsumption-and-energy-communities-overview-of-emerging-regulatory-approaches-in-europe/. [68] European Commission, Proposal for a regulation of the European Parliament and of the council on the internal market for electricity, 2016/0379/COD, Directorate-General for Energy, November 2016 (2016). ,https://eur-lex.europa.eu/legal-content/EN/ALL/?uri 5 CELEX:52016PC0861. [69] M. Khorasany, Y. Mishra, G. Ledwich, Market framework for local energy trading: a review of potential designs and market clearing approaches, IET Generation, Transm. Distrib. 12 (22) (2018) 5899 5908. [70] G. Mendes, J. Nylund, S. Annala, S. Honkapuro, O. Kilkki, J. Segerstam, Local energy markets: opportunities, benefits, and barriers, CIRED Workshop - Ljubljana, 7 8 June 2018, paper 0272 (2018). [71] Grid-Wise Architecture Council, Gridwise transactive energy framework version 1. Grid-Wise Architecture Council, Richland, WA, Tech. Rep. PNNL-22946, January 2015 (2015). [72] F. Lezama, J. Soares, P. Hernandez-Leal, M. Kaisers, T. Pinto, Z. Vale, Local Energy markets: paving the path towards fully transactive energy systems, IEEE Trans. Power Syst. 34 (5) (2019) 4081 4088. [73] J. Roberts, D. Frieden, S. d’Herbemont, Energy community definitions, compile explanatory noteMay 2019. Available from: https://www.compile-project.eu/wp-content/uploads/Explanatorynote-on-energy-community-definitions.pdf. [74] B.P. Koirala, E. Koliou, J. Friege, R.A. Hakvoort, P.M. Herder, Energetic communities for community energy: a review of key issues and trends shaping integrated community energy systems, Renew. Sustain. Energy Rev. 56 (2016) 722 744. [75] Energy & Strategy Group, Smart grid report “Le prospettive di sviluppo delle Energy Community in Italia”, July 2014 (in Italian) (2014). ,http://www.energystrategy.it/eventi/le-prospettive-di-sviluppo-delle-energy-community-in-italia-3072014.html. [76] C. Inês, P.L. Guilherme, M.G. Esther, G. Swantje, H. Stephen, H. Lars, Regulatory challenges and opportunities for collective renewable energy prosumers in the EU, Energy Policy (2019) 111212. art.

29

CHAPTER 2

Architectures and concepts for smart decentralised energy systems Andrei Z. Morch1, Chris Caerts2, Anna Mutule3 and Julia Merino4 1 SINTEF Energy Research, Trondheim, Norway VITO/EnergyVille, Genk, Belgium Institute of Physical Energetics, Riga, Latvia 4 TECNALIA, Basque Research and Technology Alliance (BRTA), Derio, Spain 2 3

Abbreviations ACER ADMM ADSM aFCC aFRC AGC ASM AVR BESS BRC BRP BSC CEC CHP DA DC DER DSO ENTSO-E ENTSO-G EPRI ETIP-SNET EV FCC FHP FP IC ICT ID IEM IGCC

Agency for the Cooperation of Energy Regulators Alternating Direction Method of Multipliers Active Distribution System Management Adaptive Frequency Containment Control Automated Frequency Restoration Control Automatic Generation Control Active System Management Automatic Voltage Regulator Battery Energy Storage System Balance Restoration Control Balance Responsible Party Balance Steering Control Citizens Energy Communities Combined Heat and Power Day Ahead (Market) Direct Current Distributed Energy Resources Distribution System Operator European Network of Transmission System Operators for Electricity European Network of Transmission System Operators for Gas The Electric Power Research Institute European Technology & Innovation Platform—Smart Networks for Energy Transition Electric Vehicle Frequency Containment Control Flexible Heat and Power Framework Program Inertia Control Information and Communication Technology Intraday (Market) Internal Electricity Market International Grid Control Cooperation

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00005-4

r 2021 Elsevier Inc. All rights reserved.

31

32

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

IRP mFRC OLTC P2P PPVC PV PVC R&D RES SAU SGAM T&D TSO TYNDP UC USEF vRES WoC

Integrated Research Program Manual Frequency Restoration Control On-load Tap Changers Peer-to-Peer Post-Primary Voltage Control Photovoltaic Primary Voltage Control Research and Development Renewable Energy Sources Substation Automation Unit Smart Grid Architecture Model Transmission and Distribution Transmission System Operator Ten Years Network Development Plan Use Case Universal Smart Energy Framework variable Renewable Energy Sources Web-of-Cells

2.1 Introduction The concept of decentralization is still under development and it is hard to find a commonly agreed definition of its functionality and technical properties. Several terms, which are used in the present chapter do not have commonly agreed definitions either, or certain contradictions can be discovered in the literature. For the sake of simplicity and to avoid any potential ambiguities, some initial assumptions have been made. For the same reasons it is necessary to set boundaries for the system to be discussed, or its scope. The main scope of this chapter is a grid area, comprising both transmission and distribution parts, operated by a single Transmission System Operator (TSO) and one or several Distribution System Operators (DSOs). The grid area can normally operate as an individual power system or as a part of a larger interconnected system. In the discussed process of decentralization, the system is divided into sections, which are called cells or agents. A “cell” refers to an integral element of a dynamically sizable energy system—from a single home to a region [1]. Furthermore, the chapter uses term decentralized system, which presumes that there is no interaction between the different cells, compared to for example “distributed system,” where a part of the decision-making process is still centralized. Apart from reasoning the necessity and potential benefits of decentralization of power system, the chapter puts emphasis on alternative ways of its operation. This is exemplified by results from EU FP7 project ELECTRA IRP, where Web-of-Cells concept and a set of novel corresponding controls were developed and tested [2].

Architectures and concepts for smart decentralised energy systems

2.2 Why decentralizing the energy system? It is needless to repeat the common knowledge that the present or conventional architecture for power systems initially started as a multitude of small networks, before they converged into national power system and finally into synchronous areas. The ongoing construction of new direct DC interconnections in Europe has taken the process even further, making several synchronous areas to function in principle as one single system commonly managed by the European Network of Transmission System Operators for Electricity (ENTSO-E). It is also necessary to mention that certain level of decentralization units have always been present in the conventional power system. Normally these are consumers, which need to maintain their security of supply on high levels and optimize their energy consumption, as for example airports. Nevertheless, several issues limited the interest towards further development of decentralized systems. The main reason is that the present conventional power system demonstrates very high reliability and security of supply at reasonable costs. Therefore, the most prohibitive issue is seemingly limited benefits potentially coming from development of decentralized system at fairly high costs. Considering isolated microgrids being an outmost version of decentralized system, one can refer to a comprehensive case study done by Electric Power Research Institute (EPRI) in 2016 [3]. The study concludes that going complete off-grid with photovoltaics (PV) and electricity storage as the main power sources while maintaining almost 100% of loads served, will cost 10 (!) times more per kWh for the consumers, compared to similar services from conventional grid. In case the customers are ready to make significant lifestyle changes and willing to accept only 80% of their load needs being met, power from an off-grid system would still cost about five times more per kWh than drawing power from the network. This is of course an extreme case of decentralization, but it gives some indication of costs magnitudes. The situation started to change increasingly after setting the 2020 climate and energy targets in 2007 [4]. This was an important milestone, indicating a gradual paradigm shift for the European power industry, which used to be one of the most conservative sectors. Massive efforts were made to promote an accelerated integration of Renewable Energy Sources (RES) in Europe. The support schemes for RES technologies have been a success story, resulting in substantial volumes of RES added to the generation mix. A considerable share of the renewable generation has been moved into the distribution part of the network, where generation was completely noneexisting before. Apart from the obvious environmental benefits this creates an array of challenges for planning and operation on different network levels, which must be addressed by the system operators in order to maintain the expected reliability levels as for example management of local congestions and voltage violations in distribution network. It reasonable to expect that these challenges will become even more severe and decentralization can be the way to resolve many of these issues.

33

34

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

2.2.1 Decentralization in European future scenarios The path of development towards more decentralized organization of the power systems nowadays is officially considered as one of the probable outlooks for the future. The Ten-Year Network Development Plan 2020 (TYNDP) [5], which was commonly developed by ENTSO-E and ENTSO-G, consider two principal drivers in development of their scenario storylines: decarbonisation and centralisation/ de-centralisation of the energy system and implemented these in three alternative main scenarios for the European power system towards 2040-205: National Trends, Global Ambition and Distributed Energy. European Technology & Innovation Platforms for Smart Networks for Energy Transition (ETIP-SNET), which brings together a multitude of stakeholders and experts from the energy sector, launched in 2018 a bold and holistic outlook for the European Power system called “Vision 2050” [1]. The vision’s main objective was defined as fully carbon-neutral circular economy by the year 2050 in Europe. One of the main features of the vision is close interplay or coupling between different energy carriers in the distribution network for optimal utilization of local resources. The vision is built upon an integrated Pan-European energy system with seamless operation through fully interoperable and networked subsystems, including centrally and locally controlled electricity networks, supported by automated local grids. The vision introduces the “cell” as an element of power system. The vision further applies the socalled subsidiarity principle, meaning that energy systems are operated in such a way that actions are optimized locally (at the most immediate level). Actions that cannot be handled locally are handled at the next, higher system level.

2.2.2 Decentralization in European R&D projects Several European R&D projects explored various aspects of decentralized architecture with different levels of technology maturity, some of these will be presented in detail further on in the chapter. An EU FP7 project Grid4EU (2012 2016) [6] included six large-scale demonstration of advanced Smart Grid solutions. Among them, there was a demonstrator NICE Grid in Carros (FR) [7] focusing on the optimization of photovoltaics (PV) integration into the low voltage (LV) grids and technical solutions for operation in islanding mode, based on battery energy storage, which was successful tested in 2016. Another FP7 project ELECTRA IRP [8] developed a decentralized architecture for the future power system coined Web-of-Cells and corresponding set of novel controls, which will be discussed in detail in following sections. An alternative concept is being developed by the Fractal Grid project (ANR) [9,10] in France, which exploited autosimilarities in power system structures to understand the complexity and emergent properties of power grids, optimize spatial organization of urban patterns and

Architectures and concepts for smart decentralised energy systems

networks, increase the flexibility and resilience of power grids. H2020 project SmartNet [11] developed and made comparative evaluation of several coordination schemes for TSO-DSO interaction with different level of decentralization, one of which was strongly inspired by Web-of-Cells [12]. German research and demonstration project C/sells [13,14] concentrated on balancing the generation and consumption of energy directly at local and regional level in the best possible way and, through this, to stabilize the grid. This list is far from exhaustive but shows that different aspects of decentralization have been given a lot of attention in R&D projects during the recent years.

2.2.3 Pros and cons of decentralization The main argument supporting centralized systems is that after the deregulation they have demonstrated very high level of reliability and cost-efficiency due to their ability to optimize operation. However, during the recent years challenges in the distribution system as local congestions and voltage violations, caused by RES, are becoming more and more visible. This significantly increases the scope and complexity of decisionmaking and operational optimization. In H2020 project SmartNet (2016 2019) five various architectures were developed [15], each of them presenting a different way of organizing the coordination between transmission and distribution system operators (TSOs and DSOs), when distributed resources (production, storage or demand) are used for ancillary services. Even though exploring of decentralization was not objective of the project, this varied across the architectures. To make comparative assessment of the architectures it was developed a simulation platform, modeling in detail Transmission and Distribution (T&D) networks, ancillary services markets and implementing a very detailed dataset of generators and loads. The simulations showed that architectures with high level of centralization had high economic performance, mostly due to finding optimal solutions. However, these required substantial computational efforts to clear the common market with all the network constraints included, despite several simplifications made in the topologies. Although multicarrier energy systems will not be discussed in this chapter, it is however necessary to mention that certain energy carriers presume more local operation for example district heating and probably hydrogen storage systems. Coupling of these energies carriers in a common system with electricity as a backbone as it is envisioned in [1] will require decentralized operation. The major drawback is probably that a decentralized architecture as such is a new model with many unknown and unresolved issues, some of which will be discussed in the following sections.

35

36

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

2.3 Development of the decentralized architecture 2.3.1 Level of decentralization The need of integrating a huge amount of distributed energy resources (DERs) into the power grid is enabling the transition from the traditional centralized power system, build upon a small number of big power plants towards a decentralized architecture based on a large number of small-scale units. These small units, which are located close to the end-users, mainly consist of RES and combined heat and power (CHP) systems. This decentralization process is been fostered by several factors: • The need for decarbonizing the power system through increasing the share of renewable energy to meet the target of reduction of greenhouse gas emissions. This is also related to the increasing interest of the population in environmental issues. • The goals settled about the need for improving energy efficiency. • Rising concern about keeping the security of supply regardless of the intermittency and unpredictability of RES. In this changing context, it is hard to imagine a power system in a 2030 horizon remaining as centralized as it has traditionally been, but it is also unlikely to have a fully decentralized power grid by then. In the coming years, the power system will evolve towards an architecture in between these approaches, running in parallel centralized and decentralized schemes and moving forward towards a power system a higher degree of decentralization as the years go by. The general characteristics of a more decentralized power system include a higher number of generation/storage units per region with a smaller size, smaller number of consumption nodes per number of units, closer distance in average between the distributed resources and the consumption nodes and slighter interactions between the DERs and the consumption units [16]. Different solutions have been proposed as control architectures to be applied to a power system with multiple components. The different control architectures are defined depending on the locations where the data is produced, the DERs are placed and the location where the decision-making entity is arranged [17], ranging from a centralized scheme to a fully decentralized one. In a centralized control scheme, the information, the data processing capabilities and the decision-making tools are managed by a single entity, the central controller, which also sends the commands to the resources. In the decentralized approach, the control is placed locally, and the different agents are independent one from each other. Between those two options, the distributed control arises as another alternative. In the distributed control, the decisions are taken centrally but some data or subprocesses can be executed locally, always supervised by a central entity. However, the border between decentralized and distributed control schemes are not always clear as, depending on the authors, the types of control

Architectures and concepts for smart decentralised energy systems

schemes considered can change, as well as slight differences may be found in their definitions. For example, in [16] decentralized and distributed concepts partially overlap, as it is remarked that all the decentralized systems are distributed but not the other way round. In [18] decentralized and distributed concepts are interchangeable concepts and the scheme referred to as decentralized is called locational or localized. Apart from the centralized, distributed, and decentralized architectures, other complementary approaches can be found, such as the hierarchical or the multiagent schemes that appear in technical literature [19]. A graphical view of these five control schemes can be seen in Fig. 2.1. The fundamental elements of the traditional power plants are the synchronous generators. Regarding the implementation of these schemes in the individual units, local controllers, internal to the devices (e.g., the Automatic Voltage Regulator (AVR)) have been employed. These local controllers are responsible to provide an adequate control signal for the unit regardless of the global control objective of the entire system. However, when replacing the conventional big power plants by smaller DERs, conflicts among the different local controllers may arise. A fully decentralized control architecture can present several drawbacks, such as a loss of information on the system dynamics or the lack of coordination that makes it very difficult for the controllers to respond all at once to a power grid event. This could also occur in the centralized scheme, which is the reason why synchronous generators have also implemented, in a hierarchical approach, centralized control schemes able to efficiently coordinate the various local controllers, such as the

Figure 2.1 Control architectures in power grids. (A) centralized, (B) distributed, (C) decentralized, (D) multiagent, (E) hierarchical.

37

38

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Automatic Generation Control (AGC) used for frequency control of an area comprising several power plants. However, the need for integrating a much larger number of DERs necessarily forces an evolution towards decentralized control architectures. Purely centralized control schemes are no longer valid in this new framework, as the complexity of a centralized control architecture exponentially increases with the number of elements to control. In terms of ICT needs, the deployment cost of a centralized scheme and a decentralized scheme are almost equal. However, the latency in the communications is higher for a centralized scheme and thus, the decentralized option is better in terms of reliability [20]. Additionally, a centralized architecture is more expensive in fixed operation and maintenance costs than a distributed scheme and a decentralized scheme, which is the cheapest one in this scale. Concerning the variable costs, the most expensive architecture is the distributed one, being cheaper the centralized and decentralized approaches. There is also a cost reduction in the distributed and decentralized options due to the higher share of renewable energy of the units. A more extensive analysis of the cost differences between the centralized and decentralized schemes for different cost components can be found in [18]. In a multiagent control scheme, all the controllers are autonomous and share the same responsibility levels, but they operate in a coordinated manner to achieve a common goal. For that reason, this scheme is especially suitable for the management of complex systems. Even though the control actions of the agents depend on their objective, limits are established to their operation to avoid conflicts, smoothing one of the major issues in the decentralized architectures. Nevertheless, the multiagent control schemes are demanding in terms of communication requirements and are difficult to scale up when the number of elements increases. To enhance the possibility of scaling a multiagent system, it can be combined with a hierarchical control, where several layers split the control problem into different subproblems that can follow themselves a different approach (central/decentral). The work developed in [21] summarizes the main architectures investigated in several European R&D projects. Next the application of the abovementioned control schemes in those projects is going to be evaluated. The Web-of-Cells (WoC) concept, developed within ELECTRA IRP [22], is a perfect example of what could be considered a hybrid control scheme, since the classification into one of the five control architecture types previously defined is complex. From the system’s perspective, the cells work independently, using their internal flexibility available, maintaining the agreed exchange among them in the tie-lines and thus, the WoC is considered as a decentralized architecture. However, each cell is centrally managed by a cell operator (centralized scheme at a cell level). Additionally, at a supervisory level, the cells must cooperate to achieve a common control objective, that is, to keep the system stable and secure while running. The links between the cells allow the neighboring cells to support each other in an autonomous collaborative distributed way, behaving as a multiagent system. Last,

Architectures and concepts for smart decentralised energy systems

it follows a hierarchical structure: system-level/decentralized/multiagent, cell level/centralized. Going into the details of the different voltage and balance control mechanisms of the WoC, hierarchical approaches are also followed up, for example the voltage control is done in two hierarchical layers: primary and postprimary [23]. Additional details of these mechanisms and their way to operate will be given in the next section, which will deepen into the control details of a decentralized architecture. The architecture developed in EU FP7 project IDE4L (2013 2016) [24] merges the concepts of hierarchical and distributed control. It has been developed using the Smart Grid Architecture Model (SGAM) [25]. The main advantage of this architecture is that it is not disruptive but looks for a smooth transition from the existing power system scheme and thus, it can be deployed over the existing infrastructures. The power system architecture is distributed by placing part of the control intelligence in a new automation device, the substation automation unit (SAU). The SAU is the main component in this scheme: it controls each MV/LV grid and is responsible for crucial functionalities such as congestion management and network operation optimization and it operates close to real-time. The distribution automation concept developed within IDE4L is also hierarchical, following a classical tertiary/secondary/primary approach. The tertiary control is centralized at the control center and operates on a day-ahead basis depending on the load and production forecasts. The secondary controllers are located in the SAUs while the primary controllers are placed in the DERs and are the most distributed part of the control hierarchy. The deployment of these components can be done sequentially according to the system needs, promoting the efficient use of the network. The SAUs can directly manage the customerowned components or through an aggregator. The aggregator automation system is also hierarchical in two layers and distributed. The lower layer coordinates the DERs and activates the resources. The upper layer manages the DERs data and communicates with the market layer for enabling the participation of the DERs in the electricity markets. The LINK project [26] defines an architecture where all the components are considered as belonging to a single entity, including producers and storages regardless of their technology and size. The architecture components are the Producer-Link, the Storage-Link and the Grid-Link. The proposed control is based on a hierarchical distributed structure where the Grid-Link is responsible for the secondary control and the Producer and Storage Links have the responsibility of the primary controls. It also includes a decentralized approach as in operation in autonomous mode, where each Link can operate independently respecting the obligations agreed with other neighboring links (similar to the relationships between cells in the WoC). In the SmartNet project [11], the concepts developed in the framework of power system operation and control were extended to the electricity market context. The central/ distributed/decentralized approaches were applied to define five different market models (coordination schemes), according to the responsibilities and relationships between the agents involved in the provision of ancillary services to the electricity grid.

39

40

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Some other national projects developed in Europe have also dealt with the design of new power grid architectures. By way of example, projects C/sells and Fractal Grid can again be mentioned. C/sells presents the future grid as an architecture based on the connection of cellular structures, and thus, the idea of the future grid is topologically and operationally very similar to the WoC. However, the cells of the C/sells architecture can have hierarchical dependencies between them [27]. In the Fractal Grid Project, the future power system is envisioned as a fractal structure, consisting of an assemblage of different resources by areas that share common characteristics. The coordination of the different fractal structures is performed by a multiagent based control scheme which is currently under development [9].

2.3.2 How to control a decentralized system Achieving the decarbonization targets of buildings and mobility calls for an accelerated electrification of heating and transport. The resulting increase of electricity consumption will cause distribution grids to be used closer to their limits. In combination with the growing amount of variable renewable generation (vRES: wind and solar) that is being integrated at the distribution grid, and at times may cause high injection peaks, this necessitates a more active control of the distribution grid. Active Distribution System Management (ADSM) will be required not only to forecast and mitigate problems beforehand (Day-Ahead/Intra-Day), but as well as to detect and correct them in real-time [28]. The ever-growing share of—also distribution grid connected—vRES gradually replaces traditional central dispatchable fuel-based generators. This calls for a control paradigm shift to load follows generation to ensure the continuous matching of generation and consumption that is essential to maintain the system balance. Storage in combination with the active control of loads (demand response) will be needed not only to ensure adequacy, but as well to provide the needed flexibility for both ancillary services and congestion management. Such active control of loads can range from simple load shedding to avoid unacceptable peaks, over load shifting to times when it is more appropriate, to load shaping to precisely mimic a requested profile (see Fig. 2.2). The fact that part of the consumption will be actively controlled to offer system services and benefit from demand response incentives (e.g., specific pricing and tariff structures) will make (locational) consumption forecasting based on historical data harder and adds to the challenge. Moreover, increasingly more distribution grid connected resources will be used to provide these real-time ancillary services. Therefore, special care must be taken to ensure that flexibility activations for ancillary services do not themselves cause local congestions, as it was raised in FP7 project EcoGrid EU [29]. As an example: if too many controllable loads like electric heaters or EV charging stations switch on in response to a measured frequency drop, the resulting rapid consumption increase may reach a level that violates the local grid capacity and cause a congestion.

Architectures and concepts for smart decentralised energy systems

Figure 2.2 Flexibility provided by storage and controllable loads will be needed to offer load shedding, load shifting and load shaping.

Flexibility activations are achieved by a combination of market signals embedded in contractual energy prices and grid tariffs (Implicit Demand Response), as well as by ad-hoc requests or incentives (Explicit Demand Response). They may result from a Direct Load Control command by the flexibility requestor, or from a decision by the flexibility owner himself in response to a proposed incentive. Furthermore, the activation may be triggered automatically based on a local measurement and a specific setpoint (e.g., droop control). While such demand response activations offer essential and invaluable system support, they as well make it harder to forecast local consumption and to ensure that no grid violations will result from it. And precautions must be taken to ensure that ancillary services activations themselves do not cause congestions.

2.4 Grid-secure activations for ancillary services (real-time control) Real-time ancillary services are needed to correct physical congestions (over/under voltage, over current) as well as frequency deviations (imbalances). In the past, thanks to central generation and top-down power distribution in strong well-dimensioned grids, there was only a small risk for congestions. For the real-time detection and correction of such congestion problems, only monitoring of a limited number of well-known potential congestion points was needed. Real-time ancillary services were mainly focused on detecting and correcting frequency (balance) deviations. The flexibility for these ancillary services was mostly provided by the generators themselves or by a few large well-known flexibility providers that would adjust their generation or consumption in an automated manner based on secure setpoints (distributed droop control e.g., Frequency Containment Control (FCC), automatic Frequency Restoration Control (aFRC)) or in response to an explicit command (manual Frequency Restoration Control (mFRC)) from the TSO as central controller. The ancillary services providing resources were mostly connected to the

41

42

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

transmission grid whose topology is well-known and static/fixed. To ensure that the activations themselves would not cause a (congestion) problem, grid prequalification that is verifying whether the unit(s) connected to the grid can realize the product delivery, considering the technical characteristics of the unit and the capabilities of the grid (see [30] for explanation of the term), checks were needed for only a limited number of flexibilityproviding resources and a limited number of well-known potential congestion points. Therefore, and because these resources could be contracted for a long time only necessitating very infrequent grid prequalifications, it was relatively easy for the TSO to perform a good grid prequalification of the offered resources and determine for instance secure droop setpoints for them that guaranteed that their activation would not cause a congestion. The grid prequalification was managed centrally by the TSO for the related Control Area, but since there are multiple TSOs so one could say that in that respect there has always been a decentralized control. Though in future we can expect increasing decentralization within the Control Areas. At the same time, the TSO coordination to achieve the common goal (managing balance) in a more effective manner (through better Imbalance Netting) will be improved (see International Grid Control Cooperation (IGCC) [31]). As a protection against the unlikely event that an activation would cause a congestion, only real-time monitoring at the limited number of well-known congestion points by the TSO as central controller was required. In future, the risk of congestions especially in the distribution grid becomes much larger, as grids will be used much closer to their limits, and in ways they were not designed for (e.g., generation connected to distribution grid causing reverse power flows; and generation peaks that are possibly much larger than consumption peaks). The fact that large amounts of flexibility for ancillary services will be provided by distribution grid connected resources, adds to this risk. Ancillary services will be provided increasingly more by storage and controllable loads instead of by generation. Their different nature and characteristics (e.g., energy constraints and rebound effect) complicates the grid prequalification and necessitates more frequent and closer to real-time grid prequalifications that can take into account up-to-date state and forecast information. Besides, there no longer are a small number of well-known congestion points, but congestions may occur virtually everywhere. Therefore, central checking by the TSO will become virtually infeasible. Sufficiently detailed grid models must be used for this; not only are these detailed grid models not available to the TSO, must also the sheer complexity of doing such a check in a central manner for the complete grid would be computationally intractable. Overloading the TSO with (DSO provided) information without addressing the inherent complexity of doing the necessary grid checks in a sufficiently detailed manner is not addressing the problem properly. Therefore, the task of grid prequalification for the proposed resources must be delegated to the DSO, and stronger TSO-DSO coordination must be put in place to support

Architectures and concepts for smart decentralised energy systems

the TSO in only contracting properly prequalified flexibility resources [30]. In the H2020 project SmartNet five different TSO-DSO coordination schemes for ancillary services grid prequalification have been proposed an evaluated [15]. In the Centralized Ancillary Services market model, the TSO contracts distribution grid connected resources directly, without any involvement of the DSO. In the Local Ancillary Services market model, the TSO indirectly contracts resources through a DSO operated local market that aggregates resources and transfers them to the TSO Ancillary Services market. In the Shared Balancing Responsibility model, the TSO transfers the balancing responsibility for the distribution grid to the DSO, meaning that the DSO has to respect a predefined schedule. The Common TSODSO Ancillary Services market model promotes a common flexibility market for System Operators. And finally, the Integrated Flexibility market model promotes the introduction of a market where both regulated (TSO and DSO) and Commercial Market Parties procure flexibility in a common market. In contrast to the transmission grid, the distribution grid characteristics and topology is not always known and up-to-date, and the topology is more complex (e.g., meshed) and more dynamic (reconfigurations). Even if a grid prequalification could be done that ensures that activations do not cause a local problem, it is much harder even computationally intractable (see section above 1.3) to ensure that they—in combination with other activations—would not cause a problem elsewhere for example upstream. Hence, as the grid prequalification of distribution grid connected resources for ancillary services becomes much harder, there is an increased need of real-time monitoring to prevent and correct congestion problems. There ideally must be a mechanism in place that ensures that the actual real-time activations of these resources are grid-secure that is that they do not cause local congestions. Real-time local observations and corrections are needed not only to detect and correct congestions resulting from local consumption and generation deviations from the market-cleared plan, but also to detect and correct congestions resulting from ancillary services activations. As congestion by definition is a local problem that needs local monitoring and can best be solved using local flexible resources taking into account detailed local information (like grid model and state), this can be done most effectively in a decentralized manner. As congestions can only be solved by local flexibility, in contrast to frequency/balance problems for which it does not matter where the activated resources are located, a prioritization is needed in the use and activation of the locally available resources. In a decentralized control approach, for a given location with a limited amount of local flexible resources, the subsidiarity principle must be applied that is a prioritized decision to first activate resources to solve local congestion problems, and only use remaining resources to help in fixing frequency (balance) problems. This is quite different from the traditional centralistic approach that prioritizes activations for frequency/

43

44

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

balance deviations using fast and automatic droop control and decides on activations for congestion secondly. Besides, due to the higher risk of congestions the fact that the grid prequalification process gives less guarantees than in the past (computationally intractable), precautions must be taken to ensure that activations for frequency control do not cause problems. Ideally, activation decisions for frequency control must be taken based on real-time information of the local grid status and forecasts (e.g., resulting in changing/adapting droop setpoints, using other resources at other locations). As an example, instead of resources having a centrally decided distributed droop setpoint causing automatic activations, there could be a decentral droop controller for a cluster/aggregate of resources that explicitly dispatches real-time activation requests to the most appropriate resources taking into account the most recent and accurate grid status and forecast. An illustration of such this principle using an EV fleet for frequency control can be found in [32]. Given the fact that local flexible resources may be scarce, it is advised that all available resources are collected in multipurpose pool under the control of the decentral controller, rather than having dedicated pools for congestion control and dedicated pools for frequency/balance control. This allows for a more effective use of the available resources and ensures that resources that may be most effective for resolving a local congestion problem are not available because they were reserved for solving frequency/balance problems. As an example, one could accomplish this by not having automatic distributed frequency droop control, but instead have a cell central controller that dispatches flex activation commands to resources that have remaining flex and that can activate it in a grid secure manner. Actually, such a decentralized control decision taken by a decentral controller as opposed to an automated distributed control based on setpoints provided by a central controller, could not only increase the grid security of activations, but as well consider other objectives like minimizing losses or maximizing robustness by taking into account local and up-to-date information and forecasts. In the future grid, such a decentralized control paradigm with empowered Cell Controllers makes it easier to ensure the prioritization of flex for congestion first and frequency second, and it pushes intelligence and decision-making authority down to the appropriate level. Both the decentralized grid prequalification as well as the real-time ancillary activations can be done taking into account local grid constraints to ensure local grid security. But it is virtually impossible to assess all possible impacts that ancillary services activations could have in other remote locations (e.g., upstream), and it is virtually impossible to guarantee that they would not cause or contribute to upstream problems. Secure flex activations at distribution grid level should not only concern local grid security, but their upstream effect must be checked as well. This would be an intractable problem. Therefore, for ultimate grid security and robustness, the WoC concept [8]

Architectures and concepts for smart decentralised energy systems

developed in the FP7 ELECTRA IRP project [22] completely removes all such risks and uncertainties by continuously driving the system to a known secure state without risking that local deviations or locally secure activations would cause upstream problems.

2.5 ELECTRA Web-of-Cells control concept In the ELECTRA IRP project, a decentralized real-time control concept was developed and mapped on a cellular grid architecture. An ELECTRA cell is defined as a geographical area with clearly defined electrical boundaries (e.g., portion of electricity grid) that has associated with it a Cell Controller that is responsible for the real-time control using intra-cell flexibility for congestion control and ancillary services. A cell can span multiple voltage levels and is connected to neighboring cells via one or more intercell tie-lines (see Fig. 2.3). In the ELECTRA IRP project, there was no cell hierarchy, even though the concept could be easily extended for a hierarchy of cells. The primarily goal of the WoC control concept is to ensure grid-secureness by continuously drive each cell to a known secure state which ensures that local

Figure 2.3 ELECTRA cells forming a Web-of-Cells for ultimate grid-secure real-time control of ancillary services. Each cell autonomously but collaboratively manages its intra-cell generation and consumption according to system-level cleared secure cell balance.

45

46

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

deviations and locally secure flex activations will not cause congestion problems elsewhere (e.g., upstream). As a starting point for the real-time control, each cell receives a secure power-flow profile for each of its tie-lines as a result of the market clearing process: this assumes that this market clearing process can rely on bids that contain location information that allows to check the corresponding planned power exchanges at the cell boundaries. These tie-line power-flow profiles define the cell’s balance setpoint (actually, the cell balance was defined as the aggregated tie-line power-flow profiles that is the cell’s import/export profile). This cell balance may be disturbed because of deviations from the planned consumption or injection in the cell itself, or by deviations in neighboring cells. Deviations may be the result of consumption or generation forecast errors, or from flexibility (ancillary services) activations. By continuously managing the cell balance, that is the power-flow at each of the tie-lines, at the same time taking care that no intra-cell congestions occur, all cell balances are corrected locally and collaboratively between every pair of cells, without risking that congestions would be caused elsewhere (i.e., no need to concern about what the remote impact could be). Below follows a brief description of the proposed controls that were defined for the ELECTRA decentralized WoC control concept [2]. Each cell has a Cell Controller that uses only intra-cell local resources and act based on local intra-cell observables.

2.6 Post-primary voltage control A Post-Primary Voltage Control (PPVC) is responsible for correcting intra-cell over- and under-voltages by periodically recalculating On-load tap changers (OLTC) and voltage droop control (PVC) setpoints. The setpoints for the OLTCs and AVR/PVR nodes are determined by the Cell Controller using detailed knowledge of the local grid. Small voltage deviations will cause distributed control actions (voltage droop control). Large deviations will cause voltage safe-band violations that trigger a preemptive new setpoint calculation. The setpoints are calculated to not only optimize power-flows but as well provide robustness so that small intra-cell deviations of generation and consumption do not cause voltage safe band violations that would necessitate new setpoint calculations. New optimal setpoints are regularly re-calculated to take into account the updated grid status and forecasts [23]. Besides loss minimization and robustness, the setpoint calculation could as well take fairness into account. By using detailed local knowledge and forecasts, a grid sensitivity model can be created that models the impact of activations at a certain location to the voltage at other locations. This sensitivity model can be used to calculate more fair droop settings that ensure a more collaborative activation of resources to correct for a specific deviation [33].

Architectures and concepts for smart decentralised energy systems

2.7 Balance restoration control A Balance Restoration Control (BRC) is responsible to remove any concerns about the potential impact that local deviations—even if they are tolerable locally—or ancillary services activations may have on potential congestions in other locations (e.g., upstream). By controlling intercell tie-line power-flows, it not only avoids congestions at these tie-lines, but as well avoids that deviations (even if tolerable at the specific tieline) could cause problems (congestions) elsewhere (e.g., upstream). This assumes that secure intercell tie-line power-flow profiles are available as the result of a marketclearing process. For each cell, the collection of its intercell tie-line power-flow profiles is defined as the cell balance setpoint. When there is no hierarchy of cells, the cell balance setpoints are determined by the central controller. If there is a hierarchy of cells, this cell balance setpoint is determined by the higher-level cell controller. The BRC continuously monitors the intercell tie-line power-flows and corrects any deviations from the setpoint profile. As both cells at either side of the tie-line observe the deviation, there will be a local collaborative effort to correct the deviation and avoid that it causes disturbances in other locations (i.e., make sure it does not ripple further). Actually, in the ELECTRA IRP project, the corrections were done based on the aggregated deviation of all intercell tie-lines that is the cell’s net import/export deviations, and flexibility was activated to increase/decrease consumption and generation as needed. But a more secure approach would be to correct the deviation at the level of each individual tie-line: that is not only restore the cell’s consumption/generation balance, but as well actively steer power flows over tie-lines to correct their specific deviation by means of Controllable Network Transformers for instance [34]. The BRC also restores the system’s balance and frequency. The cell balance setpoint (more specifically: its aggregated net consumption/generation profile) reflects its contribution to the system balance as determined by the market clearing process. Therefore, if each cell manages its tie-line power-flows, it manages the cell’s aggregated net consumption/generation and thereby contributes to restoring the system balance. And it does so in a bottom-up responsibalising manner, meaning that corrections for deviations that cause frequency deviations are done in the cell where the deviation occurs. This resembles the aFRC control principle that activates resources in the TSO Control Area where the incident that caused the deviation occurs. By applying this principle at cell level, which is (much) smaller than a Control Area, this means that activations are done close to the source which also reduces losses, next to avoiding long-distance correcting power-flows that may potentially cause in-between congestions in the path. But this bottom-up frequency and balance restoration by the aggregated effect of all cells goes at the expense of reducing imbalance netting. This must be traded off against the increased grid security and robustness. See section 4 for a more elaborate discussion (Fig. 2.4).

47

48

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 2.4 ELECTRA Web-of-Cells decentralized control scheme focused on congestion avoidance by grid secure ancillary services activations and bottom-up frequency restoration.

2.8 Balance steering control To optimize (i.e., increase) the amount of imbalance netting in combination with the BRC, a Balance Steering Control (BSC) was proposed in the ELECTRA IRP project. This control implements the negotiation of an adjusted intercell tie-line profile setpoint by the two affected cells. They can jointly agree to tolerate a deviation from the initially determined secure setpoint as long as it does not violate the specific tie-line’s constraints, has no impact on the other cell tie-lines so that there is no impact on other locations for example upstream and it still contributes to restoring the system balance in the same manner. Such coordination within the tie-line tolerances can reduce the amount of activations that are needed by each of the cells and increases imbalance netting while still ensuring balance neutrality and not risking remote (upstream) congestions. The determination of this new intercell tie-line profile setpoint must not only consider the intercell tie-lines, but as well the intra-cell tie-lines/ busses. So local detailed grid information is required, and quite some checks are needed, which depends on a decentralized approach so that the area of concern is small enough to make this possible.

2.9 Adaptive frequency containment control The BRC—when done with fast resources like storage and fast acting loads—not only restores the system frequency, but as well can contain frequency deviations. Therefore, strictly speaking there would be no need for having a separate frequency containment control (FCC). Being a distributed droop control that acts on a frequency deviation signal in a nondiscriminatory manner, such a frequency containment control would

Architectures and concepts for smart decentralised energy systems

cause cell imbalances because of its activations for correcting deviations in remote cells. However, such a separate frequency containment control may have its merits as a safety net and to support the cell balance restoration for larger deviations/incidents that cannot be corrected (fast enough) by the single cell that caused it and its neighbors. Ideally however, it should be more selective and not be the cause of cell imbalances. Therefore, an Adaptive FCC was proposed in the ELECTRA IRP project [8,22]. Just like the traditional FCC, it is a frequency droop control that acts on a frequency deviation signal. Its response is scaled though with a droop scaling factor that is proportional to the cell’s imbalance. This way, it focusses FCC activations in cells that have an imbalance already. To improve its local intra-cell grid security, it may be implemented by a cell-central frequency measurement and droop factor scaling, that dispatches flex activations to a pool of resources in a grid-secure manner, taking into account the most up-to-date grid status and forecasts. In contrast with the traditional FCC, where the explicit purpose is that all resources jointly act on an observed frequency deviation (global observables) this is not the case with adaptive frequency containment control which is more selective. This is better adapted to the situation where deviations are not the result of a single large event (in one cell) but the combined effect of many deviations in many cells, which may be a more realistic scenario for the future grid where few large central generators are replaced by many distributed (vRES) generators.

2.10 Inertia control For what concerns Inertia control, virtual\synthetic inertia could be supplied in cells and dimensioned taking into account the cell’s location (e.g., central or at outskirts, with strong or weak connections, with a lot or little inertia). Having full control over the amount of added inertia this way, one could ensure that a fixed amount of inertia is available, irrespective of the energy mix, weather conditions etc. that may otherwise impact the available inertia provided by synchronous generators [2,35,36].

2.11 Decentralizing the DA/ID energy market clearing and grid prequalification of ancillary services As a preparation and basis for the decentralized real-time control using ancillary services, two types of grid checks are needed. The first one must ensure that during the energy market clearing process at the DA (and ID) time frame, only energy bids are accepted that do not cause congestions. The second must (try to) ensure that real-time activations of flexibility that was contracted for the ancillary services, do not cause congestions (grid prequalification).

49

50

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

As discussed above, this is a challenging task in the future power system, due to the increased risk of congestions that may occur at virtually any location. Detailed grid checks are needed which cannot be done by a central controller like the TSO: not only because it does not have the detailed grid models, but as well because doing this for the complete grid would be computationally intractable. Therefore, these grid checks call for a decentralized approach where part of the checks are done at a smaller granular level that make such checks feasible. This can be done by breaking up the grid in a number of geographical areas—that we call cells—that delimit a portion of the electricity grid and cluster a number of consumption and injection nodes. This way, through decentralization, a single large unmanageable check can be substituted by multiple smaller manageable checks. In such a decentralized control scheme, each cell has assigned to it a cell controller (most likely associated with the DSO) to which part of the necessary information gathering and checking is delegated. Such a cellular architecture makes complexity manageable. For the system level energy market clearing, each cell could be represented as a single node that represents the aggregated generation/consumption of the cell, and grid clearing can be done by the TSO using a reduced equivalent grid model (see Grid Model Reduction for Large Scale Renewable Energy Integration Analyses in [37]) (Fig. 2.5). The cell controller has the responsibility to create a forecast for the cell’s planned consumption and injection. This means that firstly, intra-cell forecasts at sufficient fine granular level must be created and a grid safety analysis must be done to check for local intra-cell grid violations. This requires a sufficiently good model and knowledge of the local grid to ascertain that the planned consumption and injection of each of the intra-cell nodes does not violate any local grid constraints. If there would be violations, local flexibility activations must be determined and agreed that resolve this, and the consumption and injection forecasts are adapted accordingly. Such local intra-cell re-planning may be accomplished through local energy markets at cell level. The resulting aggregated planned consumption and injection of each cell is communicated to the central controller (master) that receives this information from all cell

Figure 2.5 In a cell-based decentralized DA\ID grid safety analysis, a reduced equivalent grid model could be constructed where each node represents a cell with its aggregated consumption and injection bids/profiles.

Architectures and concepts for smart decentralised energy systems

controllers and determines the resulting power flows over all intercell tie-lines using a reduced equivalent grid model. This resembles the traditional energy market clearing, except that the reduced equivalent grid model of the complete grid is used instead of a full model of a relevant portion of the transmission grid. It as well requires that energy bids that contain location (specifically: which cell) information. If a problem—for example a congestion at one of the intercell tie-lines—is detected by the grid safety analysis, change of plans through flexibility activations can be requested by the central controller through a dialog with the cell controllers. In the simplest case when there is only one cell causing the congestion, an immediate flex activation can be requested to resolve the problem. In the more complex case where there are multiple cells that jointly cause the congestion and mutually impact each other, a distributed optimization algorithm like ADMM (Alternating Direction Method of Multipliers) may be used to negotiate each cell’s contribution to solving the congestion. At the end of this cycle, the accepted grid-secure energy bids of each cell along with the corresponding intercell tie-line power-flows is determined and communicated to the cell controllers as a setpoint for the real-time control (Fig. 2.6).

Figure 2.6 In a cell-based decentralized approach, the complexity of the grid safety analysis can be mastered by delegating part of the checking to the individual cells and combine this with an overall system level check on a reduced equivalent grid model.

51

52

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

An example of such a locational (cfr. intra-cell) interaction scheme for DA (and possibly ID) power-flow checking at LV grid level, was piloted in the REnnovates project [38] using the Universal Smart Energy Framework (USEF) interaction scheme [39]. The USEF interaction scheme prescribes the interactions between an aggregator offering flexibility to for instance BRPs (Balance Responsible Parties) and a DSO, to avoid that planned flexibility activations by the aggregator for the BRP would cause local grid problems. To this purpose, the aggregator informs the DSO upfront about planned flexibility activations, and the DSO combines this information with his own forecasts to do a grid safety analysis using a detailed grid model. When a problem is forecasted this way, like a congestion of a specific size at a specific time and location, the DSO issues a flex request that informs the aggregator about this problem, so that the flexibility offer for the BRP can be altered in such a manner that the problem is avoided. Such a scheme could be applied and extended to a large (e.g., national) power grid represented by a reduced equivalent grid model. The above described cell-based decentralized DA/ID clearing could be done in a “flat” manner (left) or in a hierarchical manner (right). In a flat cell architecture, all cells are at the same level and have a direct interaction and communication with the

Figure 2.7 In a cell-based decentralized approach, the cell hierarchy could be either flat or hierarchical.

Architectures and concepts for smart decentralised energy systems

central controller (e.g., associated with the TSO). In a hierarchical cell architecture (see Fig. 2.7), cells are embedded in larger cells, and at each hierarchical level, a reduced equivalent grid model is constructed and used for doing a grid security analysis for all cells that are embedded in this larger cell. This way, a hierarchy consisting of multiple levels could be used, which makes that approach very scalable. The EU Continental synchronous area could for instance be represented as the higher-level cell that embeds each country—or TSO—as a cell, and each of these could then be subdivided again in a hierarchy of lower level cells. While the above described approach reduces the complexity of the grid-checking by decomposing a complex overarching check in multiple smaller checks that can be done more easily in a decentralized manner, two important challenges remain. Firstly, the intra-cell grid safety check by the Cell Controller requires accurate forecasting of local generation and consumption at individual node/bus level. This is becoming increasingly complex due to increasing amounts of distributed vRES and the active control of consumption (i.e., demand response) that makes traditional forecasting based on historical data less accurate. This challenge can be addressed by making the individual nodes/busses that are doing active control responsible for communicating their own forecast which includes any planned flex activations. This turns the source of the problem (e.g., buildings that actively control their flexibility making it harder for an external party like the cell controller to forecast it) into an opportunity. Such functionality can be expected from future Grid-Interactive buildings, that not only activate flexibility in response to implicit or explicit demand response signals or incentives, but as well pro-actively communicate what they plan to do, so that cell controllers can make better forecasts and request better and more robust flex activations. Such interactions between the cell controller and Grid Interactive Buildings can be seen as adding DSO-consumer coordination on top of better TSO-DSO coordination. Such proactively provided forecasts could serve as the baseline for flex activation settlements and could be ultimately used to give buildings penalties or rewards in relation to their forecasting/planning accuracy. The latter could be seen as delegating part of the BRP’s balancing responsibility to buildings or cells/communities (imbalance netting). Secondly, to make better informed decisions about effective and robust flex activations, it would be helpful if those that must make these decisions are informed about what flexibility is available where. That would require for instance that Grid Interactive Buildings not only communicate what they plan to do, but as well provide information about the flexibility they have that is how they can change their plan, if needed, within their own constraints. This way, those that need to decide about flexibility activations e.g., as the result of a grid safety analysis, would know immediately whether sufficient flexibility is available that can solve the problem, and whom to ask for activating this flexibility. In a more complex case, an aggregated flexibility activation can be disaggregated over multiple flexibility providers using a distributed optimization algorithm like ADMM.

53

54

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

The above two principles were elaborated and demonstrated in the Flexible Heat and Power (FHP) project [33] for a cluster of buildings (heat-pumps) where powerto-heat flexibility was used to mitigate congestions and avoid local grid problems in a preventive (DA and ID) manner. Each building with active control communicated its planned consumption, taking into account the active control of the heat-pump for a specific objective, to a district\community level controller (that could be conceptually seen as a cell controller). The active control decision of the heat-pump is resulting from a building-level optimization that takes into account the building thermal characteristics, user comfort preferences and relevant forecasts like weather and dynamic prices. Next to the consumption plan, also flexibility information in the form of consumption boundaries, which is a format that can be aggregated easily and scales well, was provided to the district\community controller. Using the aggregated information—per known congestion point—a grid safety check was done, and if needed an optimal cluster level flexibility activation was determined within the available aggregated flexibility [40]. This cluster level flexibility activation was subsequently disaggregated into a per-building flexibility activation through a distributed optimization mechanism (ADMM), consisting of an iterative process where in response to a carefully designed incentive signals, adjusted consumption profiles were generated.

2.11.1 Decentralization and markets The transformation of the power sector is changing the landscape of electricity supply. The functioning of decentralized energy systems without a doubt requires the appropriate market structure, which must satisfy the requirements of all parties. At the lower end—microgrids, local energy communities, distributed generation, local battery energy storage systems (BESS), which contribute to enabling the subsidiarity principle when balancing the system locally as far as techno-economically feasible. The expected trends also request to change the approach towards balancing and voltage control issues and addressing them in practice. This is could be achieved through specific markets architecture and other smart grid solutions. Currently, there are no consented models for the market arrangement to enable decentralized approach [41]. However, many trials and pilot projects are exploring potential solutions, or already found appropriate solutions. For example, the project ELECTRA IRP proposed WoC decentralized control architecture for the six use cases (UC) addressed in the previous subsection with a particular emphasis on the market mechanism and conditions required to perform trading of balancing and voltage control products involved in the real-time operation of the grid [22]. On the one hand the strategy for moving towards a decentralized system and related market design must be determined by: • Equilibrium of energy contents (generation, demand, storage, exchanges with other sectors, losses) at all times

Architectures and concepts for smart decentralised energy systems

• • • • •

The variability and degrees of uncertainty in generation and demand The criticality of demand (cost of energy not served) as well as their flexibility To keep system frequency within acceptable limits in case of large disturbances Voltage supporting possibilities, to maintain voltage levels at all the buses in the grid The amount of local resources to provide balancing services (including flexible demand). On the other hand, active customers are the new frontier of the electric market. In article 16 of Internal Electricity Market (IEM) Directive [42] active consumers are recognized, previously known as “prosumers” (producers and consumers). Through this article, the European Commission gives to the final customer the right to generate, store, consume and sell self-generated electricity. Several possible markets germinated by prosumers are discussed in [43] namely: • Peer-to-peer markets involve decentralized, autonomous, and flexible peer-to-peer networks, where a platform is used to bid and directly sell or buy electricity and services. • Organized prosumer groups connect goal-oriented prosumer community clusters, which are located in geographical proximity of each other, to allow efficient energy sharing among local members. According to [1] decentralized control techniques and peer-to-peer (P2P) electricity trade permeates local energy communities and their interconnection to the electricity system should be enabled. This is caused by a number of benefits of P2P energy trading [44] • Those without solar panels are still able to access renewable energy at a reasonable price from their neighbors, and those that sell their excess energy can do so at a price that is more than they’d receive as a feed-in tariff from their retailer. • Energy generation can be created from renewables, which has several benefits alone. • Energy can be bought from a known source (which allows you to choose where your energy comes from, for example from a specific community project you might like to support). • Providing a choice for dealing with other consumers and cutting out the middleman (electricity retailers). • Using blockchain, all transactions are public and once on the blockchain cannot be altered in any way creating full transparency. Customers who participate in a P2P market have physical limits imposed by the technical constraints of the network, which they are connected. The absence of control and management process in the P2P trading may lead to network issues such as overvoltage [45]. Some of the above-mentioned problems could be solved by grouping and aggregating active participants, such as microgrids, virtual power plants, prosumer community groups, energy districts, aggregator entity, prosumer coalitions, clustering power system etc.

55

56

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

The EU is poised to set an unprecedented standard by formalising the role of citizens’ communities in Europe’s energy transition. By 2050, almost half of all EU households could be involved in producing renewable energy, about 37% of which could come through involvement in an energy community. However, the market design initiative must set strong rules to acknowledge, enable and provide rights to citizens that want to be active customers or participate in energy communities. Furthermore, a future challenge should be foreseen according to the [1] by coupling electricity networks with gas, heating and cooling networks, supported by energy storage and power conversion processes. However, the nature of the energy sector requires architectural thinking in terms of commercial market structures, information infrastructures and physical interfaces to be combined in a way that is not necessary in other sectors. Crucially, there is no single authority for the energy system and so it is important to use the methods to help the sector define interfaces and establish value flow mechanisms that enable an actor with a business motive to acquire the required relationships, data and levers of control held by other actors; rather than to use the methods in the way a central planner might [46]. Definitely the clearest development is going on within existing markets, developing them in transformation of energy systems, including over-the-counter markets and electricity exchanges, markets for the trading of energy, capacity, balancing and ancillary services in all timeframes, including forward, day-ahead and intraday markets. The market design needs to provide neutral incentives and allow all kinds of technologies to develop in parallel. The complexity of the structure of the electricity market, which may allow the development of decentralized energy systems, is an important task of general conception. Intra-day markets are more flexible and better adapted to deal with renewable power in decentralized markets. Iterative intra-day trading in a decentralized market can also be used to sort out coordination problems related to nonconvexities in the production. The local market features—peer-to-peer trade and battery storage—are central to decrease the electricity bills for the end-users and to integrate distributed energy resources. The implementation of market features shows that the electricity costs within the community can be decreased by half [47]. Encouraging rapid development of markets is essential to drive social welfare and to ensure that efforts are consistent with consumers’ needs.

2.11.2 Open questions and unresolved issues The scope of the present chapter does not allow to uncover and address all possible challenging issues. One of the main issues is the necessity to make some significant changes in the present set or roles and responsibilities in order to implement decentralized system architectures. This issue will be discussed in the following section in more

Architectures and concepts for smart decentralised energy systems

details. It is however reasonable to expect that the process will be a gradual transition with some incremental changes over a period of time. Already today it is commonly recognized a necessity for better coordination and interaction between TSOs and DSOs. Moving towards decentralization is meant to reduce complexity of this interaction, but it must be extended towards the cell-level for example energy communities or active customers. Therefore, it will certainly require certain efforts, several iterations and learning period to arrive to a new fully functional model. The present ENTSO-E driven regionalization process in Europe and decentralization trends need to be aligned to draw benefits from both approaches and minimize any potential drawbacks. Coupling electricity with other energy carriers in a decentralized power system will require both new optimization methods and market models allowing common trading of different energy products.

2.12 What is next: evolution of roles and responsibilities necessary for decentralization the European regulatory framework A system architecture is characterized by a specific set of roles and responsibilities, which are legally assigned to the key market actors. Until recently the main principles of the regulatory framework was mostly based upon centralized organization of the power system and further encourages the convergence process, as for example introduction of the regional coordination centers, which started in 2008 as a reaction to the major 2006 blackout. The most recent EU Regulation on Internal Energy Market defines very specifically roles and responsibilities for the existing actors and their corresponding associations, both existing as Agency for Cooperation of Energy Regulators (ACER), ENTSO-E and emerging as forthcoming EU DSO entity. Even though the most recent TYNDP [5] includes decentralization as one of probable scenarios, so far it has not been an open debate dedicated to decentralization as prioritized future development path, therefore it is difficult to draw conclusions about position of specific stakeholders. Partial transfer of balancing responsibility from TSO down to DSOs and even further is an essential prerequisite and first step towards establishing of architectures defined in WoC concept or the decentralized coordination scheme from SmartNet project. It is interesting to mention a common document of ENTSO-E and several associations representing European DSOs [48], mentions different points of attention coming from DSOs and TSOs, where DSOs are essentially concerned about possible misalignments of actions between TSOs, DSOs, and other market players, which could lead to loss of control over the distribution grid and drive inefficient grid expansion. DSOs think that certain balancing actions could be delegated to them to procure balancing services on their network as a subsidiary activity to support TSOs. This step will be pivotal for decentralization.

57

58

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

It is also worth reminding the previously mentioned point that the European Commission has started the formalisation process of several new actors, including active customers and so-called Citizens Energy Communities (CEC) by indicating their roles and responsibilities in the IEM Directive [3]. For CECs an array of possibilities appears to be open for example engaging in energy generation, distribution, supply, ownership and management of batteries etc. The introduction of these new actors could change the landscape and roles/procedures applied both in the planning and in the operation phases. Following this, Eurelectric [49] looks at Microgrids and in particular CEC as an important future resource, which can be endorsed with new duties (especially balancing responsibility) when acting either as a supplier, as an active customer, as a DSO, or as any other system user. The final configuration of CECs depends upon many factors and is difficult to foresee, but it has good chances to become a nucleus for the future cells in a decentralized system.

2.13 Conclusions Summarising the chapter above and bearing in mind the overall picture, it seems evident that decentralization is a viable and promising alternative for the future development of the power system. It is however important to remember that decentralization is not an ultimate goal itself, but rather a way to deal with growing complexity of the power system. This means in principle that transition to the decentralized system will follow the overall path of changes in the power system corresponding to Europe’s climate ambitions. Experience from massive introduction of renewables has already shown that changing a well-established system is a very demanding process, where many difficult choices and decisions must be made. • Results from projects like ELECTRA IRP show that decentralized architecture along with the most recent advances in ICT allows changing the whole present paradigm of controlling the power system, making it more robust and resilient, despite the growing share of RES. Transition to a decentralized power system with new controlling paradigm will require a new significant step in the evolution of roles and responsibilities within the power sector. One of the most challenging issues is probably creation of sound market platforms, which will meet needs of the new controlling approaches and support trading of resources necessary for efficient and secure operation of the system. It is also worth mentioning that the EC has started the formalisation process of several new business actors, including Citizens Energy Communities. The introduction of these new actors could change the landscape and roles/procedures applied both in the planning and in the operation phases and thus creating a starting point for further decentralization.

Architectures and concepts for smart decentralised energy systems

The final configuration of the system and the transition path for arriving there will be based on continuous trade-offs between different indicators in order to achieve the optimal combination securing implementation of the energy union strategy in both consumer, environmental and economic dimensions.

References [1] R. Bacher, M. de Nigris, E. Peirano, ETIP-SNET vision 2050, 27 June 2018. [Online]. ,https:// www.etip-snet.eu/wp-content/uploads/2018/06/VISION2050-DIGITALupdated.pdf.. (accessed 30.04.2020). [2] T.I. Strasser et al, D7.1 report on the evaluation and validation of the ELECTRA WoC control concept, AIT Austrian Institute of Technology, December 2018. [Online]. ,https://www.researchgate.net/ publication/329947540_Report_on_the_evaluation_and_validation_of_the_ELECTRA_WoC_control_ concept.. [3] EPRI, Residential off-grid solar 1 storage systems: a case study comparison of on-grid and off-grid power for residential consumers, (2016). [Online]. ,https://www.epri.com/research/products/ 3002009150.. (accessed 30.04.2020). [4] The European Commission, 2020 climate & energy package, (2007). [Online]. ,https://ec.europa. eu/clima/policies/strategies/2020_en#:B:text 5 The%202020%20package%20is%20a,of%20EU% 20energy%20from%20renewables.. [5] ENTSO-E and ENTSOG, TYNDP scenario report, 12 November 2019. [Online]. ,https:// www.entsos-tyndp2020-scenarios.eu/.. (accessed 12.12.2019). [6] Joint Research Centre (JRC), EU FP7 project Grid4EU, [Online]. ,https://ses.jrc.ec.europa.eu/ grid4eu.. [7] “NICE Grid,” [Online]. ,http://www.nicegrid.fr/en/.. [8] C. Caerts, E. Rikos, S. Mazheruddin, E.G. Sansano, J.M. Fernandez, Deliverable D4.2: description of the detailed functional architecture of the frequency and voltage control solution(functional and information layer), Electra IRP Proj. (FP7) (2017). [9] N. Retiere, G. Muratore, G. Kariniotakis, A. Michiorri, P. Frankhauser, J.G. Caputo, et al., Fractal grid towards the future smart grid, CIRED - Open. Access. Proc. J. (2017) 2296 2299. [10] G. Kariniotakis, L. Martini, C. Caerts, H. Brunner, N. Retiere, Challenges, innovative architectures and control strategies for future networks: the Web-of-Cells, fractal grids and other concepts, in 24th International Conference & Exhibition on Electricity Distribution (CIRED), Glasgow, (2017). [11] SmartNet project, [Online]. ,http://smartnet-project.eu/.. [12] A. Morch, D. Siface, H. Gerard, G. Migliavacca, I. Kockar, H. Sæle, Market architecture for TSO-DSO interaction in the context of European regulation, in EEM2019 conference proceedings, Ljubljana, (2019). [13] C/sells project, [Online]. ,https://www.csells.net/en/.. [14] A. Reuter, Otwendige Technik und logische Markterfordernisse einer Energiewende Zwischenergebnisse des C/sells-Projektes, Elektrotech. Inftech 385 (2019) 136. [15] G. Migliavacca, TSO-DSO interactions and ancillary services in electricity transmission and distribution networks, Springer International Publishing, 2020. [16] K. Alanne, A. Saari, Distributed energy generation and sustainable development, Renew. Sustain. Energy Rev. 10 (2006) 539 558. [17] O. Gherke, S. Ropenus, P. Venne, Distributed energy resources and control: a power system point of view, in Risø International Energy Conference, (2007). [18] H.J. Bhatti, M. Danilovic, Making the world more sustainable: enabling localized energy generation and distribution on decentralized smart grid systems, World J. Eng. Technol. 6 (2018) 350 382. [19] S. Mohagheghi, R.G. Harley, G.K. Venayagamoorthy, Making the power grid more intelligent, IEEE VII (2007) 1 10. [20] M. Shahraeini, M.H. Javidi, M.S. Ghazizadeh, Comparison between communication infrastructures of centralized and decentralized wide area measurement systems, IEEE Trans. Smart Grid 2 (2011) 206 211.

59

60

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[21] ETIP-SNET, White paper: holistic architectures for future power systems, (2019). [22] “ELECTRA IRP project,” [Online]. ,http://www.electrairp.eu/.. [23] J. Merino, J.E. Rodríguez-Seco, Í. García-Villalba, A. Temiz, C. Caerts, R. Schwalbe, et al., Electra IRP voltage control strategy for enhancing power system stability in future grid architectures, CIRED - Open. Access. Proc. J. (2017) 1068 1072. [24] S. Repo, F. Ponci, D. Della Giustina, A. Alvarez, C.G. Corchero, Z. Al-Jassim, et al., The IDE4L project: defining, designing, and demonstrating the ideal grid for all, IEEE Power Energy Mag. 15 (2017) 41 51. [25] CEN-CENELEC-ETSI Smart grid coordination group, smart grid reference architecture, (2012). [26] A. Ilo, “Link”—The smart grid paradigm for a secure decentralized operation architecture, Electr. Power Syst. Res. 131 (2016) 116 125. [27] M. Kühnbach, F. Guthoff, A. Bekk, L. Eltrop, Development of scenarios for a multi-model system analysis based on the example of a cellular energy system, Energies 13 (2020) 773. [28] EURELECTRIC, Active distribution system management a key tool for the smooth integration of distributed generation, February 2013. [Online]. ,https://www.eurelectric.org/media/1781/ asm_full_report_discussion_paper_final-2013-030-0117-01-e.pdf.. [29] FP7 project EcoGrid EU, EcoGrid EU: from implementation to demonstration: a large scale demonstration of a real-time market for demand side participation, 15 September 2017. [Online]. ,http://www.eu-ecogrid.net/images/Documents/150917_EcoGrid%20EU%20Implementation%20to%20Demonstration.pdf.. [30] ENTSO-E, CEDEC, EDSO, Eurelectric, GEODE, An integrated approach to active system management, [Online]. ,https://docstore.entsoe.eu/Documents/Publications/Position%20papers% 20and%20reports/TSO-DSO_ASM_2019_190416.pdf.. (accessed 20.05.2019). [31] ENTSO-E, International grid control cooperation - imbalance netting, [Online]. ,https://www. entsoe.eu/network_codes/eb/imbalance-netting/.. [32] V.T. Sæmundsson, M.M. Rezkalla, A. Zecchino, M. Marinelli, Aggregation of single-phase electric vehicles for frequency control provision based on unidirectional charging, in 52nd international universities power engineering conference IEEE, (2017). [33] FHP - flexible heat and power project, [Online]. ,http://fhp-h2020.eu.. [34] D. Divan, J. Sastry, Controllable network transformers, in 2008 IEEE power electronics specialists conference, Rhodes, (2008). [35] M. Rezkalla, S. Martinenas, M. Marinelli, A. Zecchino, E. Rikos, Implementation and validation of synthetic inertia support employing series produced electric vehicles, in 24th International Conference on Electricity Distribution CIRED, Glasgow, (2017). [36] M. Rezkalla, A. Zecchino, M. Martinenas, A.M. Prostejovsky, M. Marinelli, Comparison between synthetic inertia and fast frequency containment control on single phase EVs in a microgrid, Appl. Energy 210 (2018) 764 775. [37] H.G. Svendsen, Grid model reduction for large scale renewable energy integration analyses, Energy Procedia 80 (2015) 349 356. [38] REnnovates project, [Online]. ,http://rennovates.eu.. [39] Universal Smart Energy Framework (USEF), [Online]. ,https://www.usef.energy.. [40] S.S. Torbaghan, G. Suryanarayana, H. Höschle, R. d’Hulst, C. Caerts, F. Geth, et al., Ptimal flexibility dispatch problem using second-order cone relaxation of AC power flows, IEEE Trans. Power Syst. 35 (1) (2020) 98 108. [41] Z. Xu, The electricity market design for decentralized flexibility sources, [Online]. ,https://www. oxfordenergy.org/wpcms/wp-content/uploads/2019/08/The-electricity-market-design-for-decentralized-flexibility-sources-EL36.pdf.. [42] The European Commission, Directive (EU) 2019/944 of the European Parliament and of the Council of 5 June 2019 on common rules for the internal market for electricity and amending Directive 2012/ 27/EU, 5 June (2019). [Online]. ,https://eur-lex.europa.eu/legal-content/EN/TXT/?uri 5 uriserv:OJ. L_.2019.158.01.0125.01.ENG&toc 5 OJ:L:2019:158:TOC.. (accessed 5 January 2020). [43] Y. Parag, B.K. Sovacool, Electricity market design for the prosumer era, Nat. Energy 1 (4) (2016) 16 32. [44] What is peer-to-peer energy trading?, Infinite energy, 01 March 2020. [Online]. ,https://www. infiniteenergy.com.au/peer-to-peer-energy-trading/..

Architectures and concepts for smart decentralised energy systems

[45] J. Guerrero, A.C. Chapman, G. Verbic, Peer-to-peer energy trading: a case study considering network constraints, in Asia-Pacific solar research conference, Sydney, (2018). [46] Energy Technologies Institute, Energy systems architecture methodology: enabling multi-vector market design, (2017). [Online]. ,https://es.catapult.org.uk/wp-content/uploads/2017/12/SSH3Energy-Systems-Architecture-Methodology-Multivector-Market-Design.pdf.. [47] D.J. Vergados, I. Mamounakis, P. Makris, E. Varvarigos, Prosumer clustering into virtual microgrids for cost reduction in renewable energy trading markets, Sustain. Energy, Grids Netw. 7 (2016) 90 103. [48] ENTSO-E, CEDEC, EDSO, Eurelectric, GEODE, TSO-DSO data management report, [Online]. ,https://docstore.entsoe.eu/Documents/Publications/Position%20papers%20and%20reports/entsoe_TSODSO_DMR_web.pdf.. (accessed 20.05.2019). [49] Eurelectric, The value of the grid: why Europe’s distribution grids matter in decarbonising the power system, (2019) [Online]. ,https://cdn.eurelectric.org/media/3921/value-of-the-grid-final2019-030-0406-01-e-h-D1C80F0B.pdf.. (accessed 5.01.2020).

61

CHAPTER 3

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources Nilufar Neyestani

Centre for Power and Energy Systems, INESC TEC, Porto, Portugal

Abbreviations Acronyms AB CHP HC HS MED MES PEV PL SOC

auxiliary boiler combined heat and power home-charging stations heat storage multienergy demand multienergy system plug-in electric vehicle parking lot state of charge

Nomenclature Subscripts D e g h I r t ω

external dependency on the demand side electricity gas heat internal dependency reserve time interval uncertainty scenario

Superscripts ar/dep cha/dcha G2V in, out PL V2G

arrived/departed PEVs to/from PL or HC charging or discharging mode grid to vehicle input/output energy parking lot vehicle to grid

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00012-1

r 2021 Elsevier Inc. All rights reserved.

63

64

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Parameters and variables

C, c G, g n, N Q, q q_ r SOC, soc soc _ v W, w χ π φ γ ρ η e_ p I K c s

capacity of PEVs or PL (kW) gas energy number of PEVs heat energy heat storage level difference in two consecutive time intervals reserve state of charge (kWh) difference in SOC level in two consecutive time intervals continuous decision variable determining the share of each energy element from input energy carriers electrical power SOC to capacity ratio for each PEV (kWh/kW) price PEVs participation ratio in V2G mode charge/discharge rate probability efficiency (column vector) changes in stored carrier (column vector) input carrier (column vector) output carrier (column vector) surplus carrier coupling matrix storage coupling matrix

3.1 Introduction 3.1.1 Infrastructure and carrier dependency There are several titles attributed to the wholesome view towards energy systems. Multienergy systems (MES), Integrated Energy Systems (IES), and hybrid energy systems are some of the many terms that are widely used in the literature. The ground point is to move from a single energy carrier view towards considering multiple energy carriers coordinately. With the core objective that “Energy should flow freely across the Europe (EU)—without any technical or regulatory barriers [1]”, the pathway is reaching a fully-integrated internal energy market on the Eu-wide spectrum. Alongside this, the EU commission has put one of its main strategies in achieving fully-IES [2]. The basis of this strategy is paving the way towards sector coupling that means, linking the various energy carriers—electricity, heat, cold, gas, and liquid fuels—with each other and with the end-use sectors, such as buildings, transport or industry. The plan is to deploy various existing and emerging technologies, processes, and business models, such as Information and Communication Technology (ICT) and digitalisation, smart grids and meters, and flexibility markets. Several barriers and enablers have been found to achieve this goal. Although legislations and regulations are paving the way for the implementation of integrated energy concept, including the European Green Deal [3] and Clean Energy for all Europeans

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

package [4], from the technical point of view, further studies on the requirements for enabling full operation of integrated systems are needed. The core feature of the multicarrier paradigm of the system is the inherent interdependency among energy carriers and subsystems due to their correlated interactions. Relevant optimization, operation, and planning need to consider this aspect to be able to use the extended flexibility for enhancing the efficiency of the system or reducing the risk factor. According to the Intergovernmental Panel on Climate Change (IPCC), “energy carriers include electricity and heat as well as solid, liquid, and gaseous fuels. They occupy intermediate steps in the energy-supply chain between primary sources and end-use applications. An energy carrier is thus a transmitter of energy” [5]. Given this definition, the carrier dependency that is mentioned in this chapter mainly refers to the intertwined infrastructures that involve these carriers. The primary sector of one carrier is linked to another sector through specific components causing the injection and flow of one carrier into another sector. Therefore, the usage of carrier x in sector y depends on the prospects of sector x and vice versa. In a multienergy view of the system, these interactions of carriers in different sectors is the main transmitter and consequently the dependency spreads in the whole system. As the flow of carriers starts from the supply side and reaches the end-users, the mentioned carrier dependencies can happen on various levels of the system, starting from the demand side, then local system, integrated energy networks, and finally, on the upstream supply side.

3.1.2 Dependency categories The complexity of energy systems makes the study of dependencies difficult. Therefore, as the first step, these dependencies are divided into different categories based on their origins: (1) internal dependencies; and (2) external dependencies [6]. The internal dependencies refer to those types of dependencies that occur within the scope of a local network operator’s actions and are caused by the presence of energy converters on the local system level. This category is the most typical one and has been the focus of previous studies as well. The converters act as components of a smart local system that creates the conversion-based dependency between the input and output energy carriers. If each energy system should be operated individually, the decision of one operator on how to operate a converter could affect the operation of another carrier system. Likewise, when the multienergy view is considered, the system operator needs to optimize the usage/delivery of energy carriers based on this dependency. Other dependencies that do not get affected by the actions of network operators are categorized as external dependencies. It is a virtual view on the place of occurrence of the dependencies. External dependencies happen outside the decision circle of the energy system operator and are affected by how the end-users prefer to use the delivered energy carrier. Obviously, external dependency is also caused by the existing

65

66

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

converters on the demand side. Therefore, the consideration of these dependency categories is highly dependent on the architecture of the MES as well as the penetration level of distributed energy resources, converters, and smartness of available technology in the system, both on the local system level and demand side.

3.1.3 Objectives The path for the energy sector development foresees the diversification of energy utilization, as well as the rising of the electrification on the energy demand side. Therefore, the coupling between various types of energy processing, transformation, and transportation will be enhanced, and the interdependent characteristics of the comprehensive energy system will become evident. The complexity of the interdependency dimension is much higher than that of the single energy system, and the problem refers to interdisciplinary coupling among energy departments, technical and scientific aspects, system engineering, mathematical modeling, and computer science. The first step to achieve the benefits of MES is to determine and formulate the carrier dependencies on all layers. In this regard, the objective in this chapter is to comprehensively analyze the dependencies of energy carriers in a smart local network, namely gas, hydrogen, heating/cooling, and transportation. The analysis considers all the layers of the dependency, including the system components and intersectoral aspects, as well as mathematical modeling of MES interdependency. The goal of this chapter is to give a comprehensive view of the system coupling through analyzing the layer by layer dependencies.

3.2 Internal multicarrier dependency in a smart local system The study on the dependencies of system and planning accordingly is not a new perspective. The idea was first dominated when various extreme events interfered with the operation of civil infrastructure systems. Natural or manmade disasters have shown the vulnerability of infrastructures as well as their interdependency. Each of the systems, such as transportation, communication, energy, and water can have adverse impacts on each other when encountering an operational problem. Therefore, the main focus was on finding these interdependencies to avoid the negative impacts caused by another dependent system. The interdependencies according to [7] are categorized as: (1) functional, (2) physical, (3) budgetary, and (4) market and economic interdependency. The studies in [7] as well as [8,9] proposed approaches to understand the linkage between the infrastructures and to prevent emerging problems and increase resiliency. However, the new trend introduced by “Vision of Future Energy Networks” [10], has changed the previous views towards energy systems studies and put the focus on the synergies among various forms of energy. The study put into perspective the benefits of moving away from independently studying energy systems and

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

starting a model with multiple energy carriers. Moreover, it emphasized the benefits that can be obtained from the coupled vision of the system.

3.2.1 Components of a local energy systems The increased presence of distributed energy resources has equipped the local energy systems with various components that contribute to the carrier dependency of the system. In a multienergy environment, the components are mainly categorized as converters, storage devices, and interconnectors. According to [11], convertors are the components that take one energy carrier (e.g., electricity, gas, heat, etc.) and deliver another form of energy carrier. Convertors can have single or multiple input-output assemblies. Fig. 3.1 shows a typical local energy system with some of the distributed resources that can take the role of carrier convertor or interconnector. As shown in Fig. 3.1, the carrier flows among the system components and demonstrates the dependency that is caused by the flow of the carriers through each of these components. In the following, the details of this dependency for each of the components are described as well as the operational constraints that need to be considered. 1. Combined Heat and Power As can be seen in Fig. 3.1, the Combined Heat and Power (CHP) unit causes the linkage between input gas carriers and the electricity and heat carriers’ outputs. This coupling between the input/output carriers is bound to the conversion efficiency of the CHP unit, and the decision of the system operator on allocating the amount of Multi-energy demand

Smart local energy systems ES

EV fleet

HCS

PL

CHP

Gas WH AB

Heat HS

Figure 3.1 Local multienergy system.

Hot water

End user services

Electricity

67

68

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

the input energy carrier to each component. These considerations have to be reflected while modeling the conversion of the CHP unit and can be seen in Eq. (3.1) where the input energy carrier of pg;t will produce electricity and heat after the conversion in CHP CHP. However, the decision variable of vg;t , denoting the share of CHP from total input gas to the local system, also affects this dependency. # "   CHP CHP out   :ηe :ve;t vg;t Le;t
pg;t 5 ð3:1Þ CHP CHP Lh;t vg;t :ηh Moreover, the manufacturing characteristics of CHP unit affect the decision making of the system operator; therefore, indirectly affect the dependency within the system. The limits in the amount of electrical power or heat output of CHP units consequently need to be considered in the constraints of decision-making (Eqs. 3.23.3). Furthermore, as the CHP unit should be operated in certain heat to power ratio, (Eq. 3.4) serves as another constraint while adding the interdependence of the system and CHP. P CHP # pt CHP # P

CHP

H CHP # ht CHP # H

CHP

5 ht CHP =pCHP γ CHP t t

ð3:2Þ ð3:3Þ ð3:4Þ

2. Auxiliary boiler Auxiliary boilers are common components of MES as they provide the cogeneration option alongside the CHP units. They receive gas as input carriers and produce heat that causes the dependency between these two carriers. However, as usually auxiliary boilers operate with CHPs, the decision-making on what should be the share of input carrier for the auxiliary boiler in comparison to the required input of CHP and the gas demand of the consumers is another aspect of this dependency, as shown in (Eq. 3.5). Besides, the heat energy output of the auxiliary boiler has some limits in providing energy as in (Eq. 3.6). h i     boiler boiler :ηh ð3:5Þ
pg;t 5 Lh;t vg;t H Boiler # ht Boiler # H

Boiler

ð3:6Þ

3. Energy storage Storage units have proved to be among the effective distributed energy resources, bringing added value to the system operators. In a multienergy context,

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

heating and cooling have been one of the main carriers alongside electricity. Therefore, the heat and electric storage units have been a complementary component of the local energy systems alongside CHP and Combined Cooling Heat and Power (CCHP) units. The storage units, on their own, do not cause interdependencies. However, as they give the option of storing excess production at some points and using it at other periods, they affect the decision-making of the system operator on how to use the outputs from energy converters and consequently affecting the overall system dependency model. Moreover, the intertemporal dependency caused by the state of charge in storage affects the status of the system coupling at each time step. The status and intertemporal dependency of energy storage units are modeled by (Eqs. 3.7 and 3.8). The operational limits of storage units also need to be considered as shown in (Eqs. 3.9 and 3.10). E_ α;t 5 Eα;t 2 Eα;t21

eα 5

8 Storage < ηα;ch ; : 1=ηStorage α;dis ;

if Qα $ 0

ðCharge=StandbyÞ

if Qα , 0 ðDischargeÞ  storage  storage _h #r t h

H Storage # hstorage #H t

ð3:7Þ

ð3:8Þ

ð3:9Þ Storage

ð3:10Þ

4. Electric vehicle parking lots Electric vehicles are among the emerging technologies in the field of energy studies. They are affecting several energy sectors, namely electricity, and transportation. However, the role that they make in the electric system operation is more challenging. Several studies have investigated the aspects around the integration of electric vehicles in the network, addressing their charging programs, their services to the grid, the allocation of their charging stations, etc. The benefits of electric vehicles are mainly indicated by providing storage capacity and flexibility to the system. However, under the multienergy context, their role can change from a flexibility source to an energy carrier and sector coupler. The electric vehicles (EVs) carry the charging capacity and the state of charge that can act as a new carrier in the multicarrier system. With the increased penetration of EV charging technologies on different system levels (i.e., household charger, public chargers, public parking), the influence of EVs on the interdependencies will grow. For better comprehension, imagine an EV with access to home-charging equipment as well as

69

70

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

staying (i.e., charging) during working hours in a parking lot. The decision of the EV owner on charging the vehicle during night-time while parked at home or during the day is not as simple as a choice with price-based criteria. The reason emerges from the fact that with other entities such as technical and market aggregators coming into the business model of the flexibility deployment, saving the capacity of the EV battery and offering it to the grid could be a better solution rather than fueling the vehicle during cheaper hours. Given the incentives received from the aggregators or flexibility providers for participating in DR programs, scheduling the charging of EVs would become a multiagent and multiobjective decision-making process. Therefore, although EVs are generally considered as extra storage, with the example given above, it is clear that they affect the decision-making on different layers of the multienergy local system (i.e., the decision can be made either by home-energy manager, technical aggregator, or market aggregator), hence, enforcing the interdependency. Modeling this interdependency requires consideration of various factors, including the driving behavior of the vehicle owners. The mileage traveled by car is an important factor affecting the status of charge or the excess capacity available to the system operator. Moreover, if a parking lot is considered to be present in the local network, the arrival/departure pattern of the vehicles and their stay duration in the parking are among the affecting factors. The following shows the dependency modeling between the input/output carriers of a parking lot and the operational constraints that should be considered while modeling the dependencies caused by electric vehicles. It should be noted that the dependency modeling for the home-charging equipment is similar to the one of the parking lot. The difference is on the modeling of the parking lot, the aggregated values of the EVs using the parking should be calculated while in the home-charging one, it is not needed. In Eq. (3.11), the interdependency between input carriers (i.e., arrived vehicles’ capacity and state of charge) and the output carriers of a parking lot is shown. The interdependency to the electric grid happens with the charging and discharging of the vehicles in the parking lot that is included int the PL model with κPL ω;t . As shown in Eq. (3.12), the term κω;t denotes the total energy of the parking lot. The rest of the operational constraints of parking lot concerning its charge/discharge ratio and arrival/departure patterns are shown in Eqs. (3.13)(3.18). 3 2 PL κω;t 6 PL;ar 7  PL 6 SOCω;t 7 " PL;dep # 7 6 0 21 6 c PL;ar ηω;t 1 0 7 5 socω;t ω;t ð3:11Þ PL;dep 7 0 0 1 21 0 6 PL cω;t 7 6C _ 5 4 ω;t PL s_ocω;t PL;in PL;out κPL;in ω;t 5 wω;t 2 wω;t

ð3:12Þ

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

ηPL ω;t 5

8 < ηPL;cha ;

if Charge=Standby

: 1=ηPL;dcha ; if Discharge

ð3:13Þ

PL PL PL 5 socω;t 2 socω;t21 s_ocω;t

ð3:14Þ

PL PL PL;ar PL;dep _ PL C ω;t 5 Cω;t 2 Cω;t21 5 Cω;t 2 Cω;t

ð3:15Þ

PL;in # ΓPL nPL wω;t ω;t

ð3:16Þ

n o PL;out PL PL # min ΓPL nPL wω;t ω;t ; socω;t φ

ð3:17Þ

EV PL cω;t

ð3:18Þ

PL PL SOC EV cω;t # socω;t # SOC

3.2.2 Electricity—gas The cooperation and cooptimization of electricity and gas energy systems have been the most investigated aspect of the sector coupling paradigm. Being the most dominant energy carriers in the energy systems, the mutual effects of these two sectors were the subject of many studies. However, understanding the interdependency of these two systems requires comprehensive knowledge of these two sectors. The interdependencies between electricity and gas start from the supply chain in these two sectors and spread over other aspects, including the networks and the mutual operation of components. Due to the grid-bounded nature of natural gas energy carriers, the nodebased consumption for gas generation units is not a proper estimation. Instead, it should be considered through mathematical models considering the correlation between nodal consumption and pipeline exchange. Therefore, the early studies on this matter started, including the modeling of the gas network in the electricity sector studies. Proposed models often considered the gas network as a dual of power system. Hence, various studies disregarded the pressure drop of the gas network and used the linearized gas network model the same way as the power system DC load flow [12]. In such models, the exchange among nodes of the gas network is based on gas volume, and the pressure drop of gas pipelines is neglected. On the other hand, some studies have considered the effects of pressure drop. Therefore, in these models, the gas flow in the pipelines are considered dependent on the pressure drop [13]. Despite the similarities, electricity and gas networks have a major difference that lies in transient dynamics. Therefore, it is not a correct assumption to consider the gas

71

72

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

network as a power system dual. In power systems, the system dynamics is less than 1 minute long, while considering the gas system, several hours are needed to achieve the steady-state condition. Thus, in the modeling of this network, the maximum required demand in nodes should be considered based on the distances between nodes. This issue is particularly relevant in unit commitment studies, which has shorter time scales, as the steady-state condition in the power system might be achieved in an hour while the gas network is in its transient mode. Authors of [14] and [15] have presented a linear model for gas networks based on its dynamics in the unit commitment study. The time required to transfer gas from one node to another is modeled as a delay in the network’s mathematical model. In [16], stochastic programming has been used to model the uncertainties of the electric grid and their consequent spread to the gas networks. The interdependency of gas and electricity infrastructures also affects the long-term planning of these two infrastructures. Reference [17] proposes an integrated model for generation and transmission expansion planning of gas and electricity networks, and according to the results, the necessity of such integrated planning has been illustrated. Besides, reference [18] considers the same approach in a competitive energy environment. Authors of [19] have investigated the role of gas infrastructures on power system security. For this purpose, they have studied the role of combined cycle units and the lack of gas energy carrier in gas-based power plants. The results present a table for the output of gas-fired units in case of contingency in gas networks. After a contingency in gas network, the power system will lose some gas-fired units; therefore, an increase in local marginal price, an interruption in providing the demand, or congestion of the transmission lines may happen. Although the reference uses a simplistic security analysis and forced outage rates are used in the modeling, it concludes that the mitigation of such effects is highly dependent on precise modeling of the effects of gas infrastructure in power system security. Moreover, in [20], the reliability of gas and electric networks is studied and it is concluded that the simultaneous study of the reliability of gas and electric networks is necessary, and proposed a simultaneous solution for the maintenance of scheduling of gas and electric networks. Another aspect that took the most benefits from the interdependence of the electricity and gas sector is the energy market studies. Many recent studies were dedicated to investigating the potentials of cooptimized and coordinated operation of electricity and gas markets as a result of deploying the advantages of interdependencies between these two systems. Moreover, regarding the sector coupling and its economic growth, new market models for considering integrated market designs are necessary. The author of [21] maximizes the profit of generation units by considering joint contracts of gas and electricity. In this case, the contracts are considered mostly bilateral and based on the spot market. Moreover, the markets of gas and electricity in

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

Colombia and Mexico and a review of the role of the gas market on expansion planning of electric network have been introduced in references [22] and [23]. By considering the price of gas and maintenance costs, the role of gas contracts on maintenance scheduling of generation units was investigated in [24]. In [25], the competition of gas-based plants in the electricity and gas markets has been modeled based on the uncertainties of renewable energy resources and demand. In this regard, the market power of gas suppliers on the electricity market price has been investigated in [26]. In [27], an integrated model for electricity and natural gas markets is presented where the inherent flexibility of the gas system is used to facilitate the integration of renewable energy resources. Another example is the study in [28], where the clearing of the coordinated electricity and gas market is presented. The results of these studies show an increase in the effectiveness of the energy markets when they are being operated coordinately and use mutual flexibility.

3.2.3 Electricity—hydrogen While the interdependency of electricity and natural gas infrastructures has been the subject of many previous studies, the introduction of hydrogen into the equations has been the recent trend in the field of MES and sector coupling. This aspect has gained interest in the introduction of Power-to-Gas technologies (P2G) into the electricity system that refers to the process of converting excess production of renewable-based energy producers to gaseous energy carriers such as hydrogen and methane via water electrolysis [29]. This process provides the opportunity for long-term storage of electricity as well as an increased capacity for storing the excess production that would serve the best purpose of reaching a complete renewable-based production [30]. Adding P2G to the system will cause a chain of interdependency among several carriers. As shown in Fig. 3.2, the presence of P2G will cause the interdependency among electricity, gas, water carriers. However, in terms of modeling the impact of P2G on the total interdependency modeling, its role in a multienergy system is twofold: the convertor of electricity to gas and the storage gas (i.e., hydrogen storage or methane storage). A P2G solution is a promising approach for increasing efficiency in the electric systems, and it already has a certain level of maturity in its application. Assessing the effects of P2G units on the operation of integrated electricity and gas systems is similar to what discussed in the previous section regarding the interdependency

Figure 3.2 Interdependency of Power-to-Gas (P2G) technologies.

73

74

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

of electricity and gas networks. However, if the transmission of hydrogen is added to the system, the problem will gain new aspects. Transmission and distribution of hydrogen is an undergoing study that helps the increased role of hydrogen in the IES. There are three main options for the transportation of hydrogen: (1) compressed gas cylinders or liquid tankers, (2) hydrogen pipelines, and (3) blending with natural gas. Although transportation of hydrogen is considered to be easy, and even possible to be done in different forms (i.e., gas and liquid), it requires a high level of investment. Nowadays, the common approach for transportation of hydrogen is through lorries that on itself cause the dependency between the hydrogen sector and the transportation sector that is further analyzed in Section 2.5.

3.2.4 Electricity—gas—heating/cooling District heating and cooling is one of the main enhancements when it comes to moving from smart grids to smart cities. The coupling of these infrastructures not only brings a new level of flexibility due to the synergies between them but also increases the level of interdependency. Electricity, gas, and heating/cooling are the most common demand on the end-user side, specifically, the heating/cooling service that can be obtained interchangeably by each of these carriers. The provision of heating/cooling services is enabled by the installation of cogeneration units and has been in extensive use in coordination with renewable generation, especially in Europe. The components of the local energy system that act as the linkage among the carriers in the paradigm of district heating/cooling are CHP units, electric boilers, heat pumps, and circulation pumps [31] that as described in Section 2.1 represent the interdependencies. On the demand side, devices such as air conditioners and space heaters also cause dependency on using the carriers of electricity or heating/cooling. Similar to other coupled infrastructures, the operation of integrated electricity-gas-heating/cooling systems will be affected by the added flexibility as well as the constraints imposed by the interdependencies. Depending on the centralized or decentralized operational method, the approach towards incorporating the interdependencies will be different. In [32], the interaction mechanism of integrated electricity and heating system is analyzed based on the time-scale characteristics of both systems. It is studied that how a disturbance in one system can affect the operation in the other one and how the interdependencies as well as the control strategies in each of these infrastructures would affect the operation. The decentralized dispatch of electricity and heat systems using the market equilibrium and energy trading is presented in [33]. The energy trade for multilateral coupled gas-heat-power networks is presented in [34]. The basis of these studies is relying on the synergy and interdependency of the carriers. However, the results show that the elasticity of demand and the dependency of the carriers that occur on the demand side have

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

more impact on the trade decisions rather than the coupling dependency of the components. Therefore, the focus of the study was drawn towards new market frameworks for CHP markets where new DR programs could be perused [35].

3.2.5 Electricity—gas—hydrogen—transportation In previous sections, the coupling of electric infrastructure with gas and hydrogen was addressed individually. This section specifically addresses the interdependency among these carriers concerning the transportation sector. The interdependency of the power and transport sectors is converging with the emergence of electric vehicles. Several studies have addressed the modeling and analysis of the interdependency of electric transportation and the power sector [36]. The effects of this interdependency are mainly reflected in the urban planning problems, traffic flow studies as well as the power flow studies. The traffic flow of electric vehicles will affect the design of roads and highways [37] while also affecting the allocation of charging stations on the electrical network [38]. However, in a multienergy analysis of the system, the interdependencies of the components will be more impactful. The study in [39] showed the effect of home-charging electric vehicle devices and public parking lots on the dependency of multienergy demand. It shows that depending on the already existing components in a local multienergy system, the EV commute and the vehicle owners’ charging behavior will affect the operational decisions of the multienergy operator. Recent trends in this topic were dedicated to analyzing the multicarrier transportation sector, considering electricity, natural gas, and hydrogen as fuel options. The increased usage of these carriers, especially in the public transportation sector, emphasized the carrier dependencies in the transportation sector. However, as carrier converters do not exist in transportation in general, the types of dependencies are mainly due to interconnectors and storage units. When a vehicle is using petroleum as a fuel to transport hydrogen, or when it is electrically charged in one point, and the remainder of the charge in the battery is injected to the grid in another connection points, the interconnector of different energy carriers is the vehicle fleet. Therefore, the commute and vehicle behavioral patterns from the transportation sector will act as new factors affecting the infrastructure and sector interdependency. There are very few studies that have focused on this aspect of the study. Among them, [40] studied the traffic pattern of electric vehicles in different urban areas with different energy consumption types. The analysis shows that considering the commuting pattern in the optimization of the charging schedules will save energy costs for the system operator as well as providing extra flexibility by having the available capacity of the EV batteries to offer as ancillary services to the grid. Although having hydrogen as a fuel does not face as many problems that EVs face when it comes to charging and interactions with the grid, both of these fuels are facing

75

76

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

a lack of charging stations and fuel provision for the end-users. However, when such burdens are resolved, the benefits of the interdependencies among these carriers will be received not only by the involved sectors but also by achieving less CO2 emissions and better environmental measures.

3.3 External dependencies in a smart local system The dependencies of energy networks, infrastructures, and components within the energy sector have been addressed under the scope of internal dependencies. This section will address the dependencies that can occur outside the direct area of energy systems but still affect the overall dependencies of MES. Two types of external dependencies are identified in that sense: the dependency on the multienergy demand side and the dependency of the information/communication sector.

3.3.1 Multienergy demand As the systematic vision of the energy systems of the future is being subject to changes due to system coupling and multienergy aspects, the considerations of the demand side also need to evolve [41]. The end-users cannot still be considered as final consumers of the energy as being delivered to them, but rather the requesters of certain services that need carrier inputs to be activated. For instance, considering the residential consumers, what is needed by the household users is the proper functioning of their household electric or gas-fired devices, heating, and cooling of the space, heated water, and proper indoor air quality. Depending on the availability of the technology, each of these services can be provided by either electricity, gas, or heat carriers. Therefore, the ultimate goal of the user is to achieve its required service rather than a specific energy carrier. In such circumstances, we can assume that the available technology on the demand side can receive various energy carriers as an input and convert them to the required services. For better comprehension, assume that MES covers a residential urban area. In this case, some DGs such as diesel engines, CHP units and storage units exist in the local network. However, on the demand side, there exist devices that benefit from the multicarrier input technology. Therefore, this technology will bring the opportunity for both demand-side and system operators to take benefit from it. In this situation, whenever one carrier has a higher price compared to another one, the consumer may have the choice to select between two or more carriers to use as an input. On the other hand, the system operator will also be able to choose between various sources for supplying one certain service, which is beneficial during system emergencies or resource shortages, as well as high price intervals. Therefore, the situation will cause a dependency on the demand side. Since the estimation of the demand and the required input resource will be dependent on the customer’s choice of carrier, a new

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

dependency is shaped. As this dependency occurs on demand-side, it is different from those dependencies that are within the local network system due to its internal converters such as CHP units. As a result, it is called external dependency. This external dependency can be further employed in the operation of the multienergy system by introducing new DRe schemes. The general definition of the DR is the change in the consumption pattern of load in response to a factor such as prices, tariffs, or requirements of the grid. However, with the availability of external dependencies on the demand side, a new DR program can be defined as a carrier-based demand response where the definition is “the change in the conversion pattern” of the load. In a conventional DR program, whenever requested, the consumption is changed from one hour to another, or the level of consumption is reduced. However, in the carrier-based DR, the demand load is shifted from one energy carrier to another one while maintaining the final service for the end-user. Although activating such DR programs depends on the availability of the technology on the demand side, whenever it is being used it takes the optimal use of the interdependencies among the carriers and provides the system with an extra flexibility option.

3.3.2 Information/communication Communication and data flow have always had a significant role in the success of energy systems operations and coordination. In fact, communication and information infrastructures are the layer that all energy sectors have a dependency on it. However, recent trends in energy informatics and the usage of communication potential, have brought new paradigms in the energy sector as well. The topics such as decentralized operations, blockchain, and citizen-engagement have each brought new challenges and opportunities to the dependencies of the energy sector on the communication infrastructure. In the case of MES, the ability of one system to be able to communicate to the sources of flexibility in another system is of crucial importance for the coordinated operation of the system. Moreover, the management of various distributed energy resources that are available in different layers of the integrated system is only possible through a strong telecommunication infrastructure and information basis. One of the obvious examples of interdependency with communication is the management of electric vehicle parking lots. The flow of information starts when the EV owner arrives at a charging station, by submitting their expected departure state of charge and stay duration as well as recording their arrival time and the initial state of charge. These are the crucial information for the EV parking lot operator to manage the charging of vehicles. Without the availability of such information, the uncoordinated charging of EVs could have a negative impact on the electric grid. On the other hand, the main goal of the multienergy vision of the system, which is having the benefit of

77

78

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

extended flexibility, would not be possible without the communication layer. It is known that a failure in communication infrastructure largely affects the operation of the electric system. However, in the case of coordinated and cooptimised energy networks and markets, the effects of this cascading failure would be more harmful to the whole system.

3.4 Interdependency modeling 3.4.1 Coupling model of components and services Components and services shape the first layer of dependencies in MES, where the dependency between carriers and the services delivered to the multienergy demand is being modeled. As extensively explained in Section 2, incorporating the dependencies caused by the components of MES should follow the approach that is used for system modeling. In the following sections, the different approaches for system modeling are presented. The component dependency modeling can be adapted based on the modeling that is chosen for the system coupling modeling. Individual component modeling (as described in Section 2.1) would be interconnected to each other based on the assumed relations in a system and then used within the whole system model in an aggregated manner.

3.4.2 Coupling model of local energy systems There is a variety of existing literature addressing different modeling approaches for considering the interdependency of MES. Here is a review of the most common practices on local system level. 3.4.2.1 Energy hub method The energy hub method was first introduced by Geidl et al. in [42] and [43], where an energy-hub approach towards various components of the system was proposed. However, the basis of the energy hub method relies on the input-output model that was first introduced by [44], characterizing the interdependency between input and output in the economic system. The basis of this concept is the economic interaction of various sectors synthesis that has been adapted to the IES view of the energy systems. In the energy-hub approach, the MES is divided into two main parts: (1) energy hubs; and (2) interconnectors. Therefore, the input and output energy carriers are considered individually, and the energy hub has components with different functionalities of conversion and storage. To mathematically model such systems, a coupling matrix based on the models and the interactions of the components is established. The coupling matrix represents the interdependency of the output energy vectors to the input

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

carriers. The basic model was proposed in [45] and further extended in [46] and is described as follows. In the matrix modeling approach, the dependency between the input and output carriers of a MES is represented by the coupling matrix. The derivatives of the coupling matrix consist of the conversion models of the components existing in the system, each corresponding to the link between the input and output (i.e., Eq. (3.19)). However, with the categorization of the dependencies in a MES into internal and external dependencies, the modified input-output interaction and the coupling matrix would be as Eq. (3.20) ½λ 5 ½C
½p    λI CI ½λ 5 5
½p λD CD

ð3:19Þ



ð3:20Þ

In the energy-hub approach, the “hub” can have different scales. For example, a CHP or CCHP unit can be considered as a hub, while a network of various carriers, interconnectors, and components can be considered as an energy hub. The extension of energy hub systems follows the linear aggregation of components (refer Section 4.2.) that allows the integration of other components such as energy storage, electric vehicles, and DR programs along with other distributed energy resources. This aspect will contribute to one of the main advantages of this approach: simplicity. Following the synthesis of various systems with the matrix model and the input/output model provides a simple and relatively convenient form for modeling the interdependency of the MES. However, as in this approach, the relation of the elements of the matrix is linear, the nonlinear characteristics of carrier interdependency are not reflected in principle in the model. Moreover, the interdependencies in MES add to the complexity of the system operation, and following the full linear approach may not reflect this complexity. There are several studies that have considered the energy-hub approach as the core methodology for MES operation [4749]. Some of these studies specifically focused on the reliability analysis and security assessment (cascading failure) in operational conditions with systems equipped with storage units [5052]. However, there is still the need to improve the reflection of the risks of the interdependencies in the modeling. One of the solutions can be through interoperability modeling. With the recent trends in the interoperability increase among technologies and components, the vectorinteroperability aspect needs to be included in the model. Few studies started the modeling of interoperability in multienergy [53] and complex energy systems [54] to consider MES and its interdependencies as part of urban energy systems. Interoperability is the other side of the interdependency which gains more important when the coupling model should cover large-scale systems.

79

80

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Another aspect of the energy-hub approach is the physical energy input-output dependency. In some of the input-output based models for the MES [55,56], the interdependency between the input vectors and the output carriers can be studied as well as the input-output correlation, while the coupling matrix used in the energyhub approach only considers the input-output link. Although this approach provides a higher level of accuracy for interdependency modeling, the disadvantage of this approach is the need for the availability of data for physical dependency modeling. 3.4.2.2 Energy network method Although the energy-hub approach and the coupling matrix are the most common in the literature for the interdependency modeling, the energy network method is also widely used where the energy subsystems are considered as energy transfer lines, connecting on specific network nodes. These nodes are where the interdependency of the two networks occurs. This approach was the most common in studying the interdependency of electricity-fuel-transport energy systems where the electric charging stations and fuel stations would serve as the coupling points between networks (electric grid and highway/roads represent the networks). Moreover, from a thermodynamics point of view [57] and when studying the heating subsystem, the exergy principle gains more weight. Therefore, a generalized model of the energy networks can be developed, and only the equivalent energy transfer should be used in the equations. However, the problem with this method is that generalizing the system modeling affects the accuracy of the dependency modeling. One of the recent studies that tried to address this issue is [58] where the gridbased modeling of a MES is proposed considering several aspects of dependency, including the economic, energetic, and exergetic criteria.

3.4.3 Large-scale coupling When dealing with large-scale MES, the complexity of modeling the carrier dependency is enhanced as the mutual interactions of the networks and services are broadened. As shown in Fig. 3.3, the interdependency of various energy carriers of several energy subsystems and how this dependency can affect is depicted. On this scale, the energy system planning problem is as challenging as the operation where several entities need to consider the interdependencies for the long-term planning of the infrastructures of their networks. 3.4.3.1 Agent-based method Integrated or MES are complex adaptive systems, and the carrier dependency makes the decision-making process on a large-scale even more complex. One of the best approaches is the agent-based method that has been widely used in the literature. Considering the IES as “systems of the systems” which were proposed in [59],

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

Figure 3.3 Interdependencies in large-scale coupled systems.

examples of agent-based methods used for modeling the use of various carriers (electricity, heating-cooling, etc.) were presented. The study in [60] extends the operation of the low-voltage distribution network with other energy vectors using agent-based modeling. However, the majority of the studies using the multiagent approach focused on the decision-making architecture of the MES and how the dependency would affect decisions such as operation and market participation [61]. In [62], a multiagent structure is proposed for the optimization of energy systems with an energy-hub approach. The study in [63] proposes a multienergy intermediary agent for the trades on various layers of the integrated energy system. Using the agent-based method to simulate the behavior of decision-makers in MES and its subsystems considers the inherent interdependencies within the system, and therefore, provides a more effective system analysis through taking the correlation of discrete events into account. However, as a general problem of agent-based approaches, the quality of the study

81

82

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

depends on the modeling accuracy of the agents that can be a concern when dealing with the multienergy concept. The behavior of agents in different energy networks towards each other and the mutual dependencies of the carriers is still unprecedented when it comes to MES due to very recent concepts in this area. 3.4.3.2 Complex system method There are very few studies that have tried to address the whole complex system [64]. In this approach, to deal with the large-scale systems, the whole complex system is modeled using a network-based model. The main feature in this approach (either considering the independent or interconnected networks) is to model the whole system with both heterogeneous and homogenous nodes. It follows a multilayer network scheme similar to the energy network modeling approach (refer Section 4.2.2.), however, with large-scale infrastructure inclusion. In the case of MES, using a complex modeling approach requires simulating the mutual response of networks that have been the focus of existing studies. Therefore, in electric transmission and distribution systems as well as water-gas systems, the effects of different strategies under random failure and hurricane hazards need to be studied. These strategies generate parameters based on degree, spacing, and clustering coefficients. In addition, the cascade failure process of the integrated energy system from transient to steady-state performance is an important aspect of the study. To mitigate the cascading failures of IES, measures can be taken to add bypass, to enhance individual component performance, and to adjust the interface configuration. The main advantage of this approach is using the topological characteristics of MES while identifying the key MES component. This approach focusses on the modeling of the whole system with the purpose of improving the system robustness. However, this method is limited to random structural networks and random faults. In an integrated energy system, the components and the involved networks have different fault probability that has made it difficult to find the correlation between these components and networks of different energy subsystems.

3.5 A case study on interdependent MES model Previous sections have completely addressed various types of interdependencies in a MES, as well as several modeling approaches that exist in the literature for mathematically incorporating the dependencies in the decision-making problem of MES. In this section, a case study of interdependent MES is going to be presented. For this purpose, a structural view on the various layers of MES is presented, and it is shown how the interdependencies will affect the interaction on each layer and in-between layers. Fig. 3.4 illustrates various levels of an exemplary MES. As can be seen, a MES consists of several layers. On the first layer, the demand side stays where the presence of external

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

Figure 3.4 Different business interaction levels in a MES

dependencies of the multienergy demand (the first layer of dependency) also occurs. It should be noted that this dependency happens on this layer but will affect the upper layer of local energy systems. Moreover, with the implementation of peer to peer and local trades, the cross-effects of this dependency within the MED layer will increase. The second layer deals with the local energy systems where the interdependencies occur due to the presence of various MES components and distributed energy resources as described in previous sections. However, from an operational point of view, although the interdependencies occur within each LES and due to the available components, the interactions between local energy systems will be another cause for the interdependency on this level. The interdependencies among LES depend on the type of interactions that they might have and based on the operational decisions of the system operator. If one LES is equipped with CHP but not heat storage while another neighboring LES has excess heat storage capacity, the decision would be the usage of the excess capacity for the optimization objectives. However, it needs to consider the operational dependencies that this decision will impose on both local energy systems. In this case, the modeling of dependency will go beyond the modeling of components’ dependencies and has to take into account the system dependencies. The energy-hub approach, matrix modeling, and input-output modeling is the most common dependency modeling that is used in this layer. The third layer covers a larger scale MES where several local energy systems and multienergy demands are present. On this layer, MES components other than the ones from local energy systems could be present that creates the dependency between the input to MES and the input to local energy systems (LES). Moreover, this layer

83

84

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

provides several opportunities for new business cases and roles. One of these new roles is the multienergy aggregator, proposed by [65], offering various energy carriers in different energy markets. However, as described in the previous section, the interaction between the multienergy aggregators, conventional aggregators, different energy markets, and energy retailers will increase the level of complexity in this layer. Based on these new roles, the agent-based approach is the one which is used for modeling the dependencies of this layer as well as the energy-hub approach. The final layer which is shown in Fig. 3.4 is the high-level infrastructural and sectorrelated layer including various energy markets and energy networks. When analyzing the dependencies on this layer, the dependencies on the supply side and infrastructure is more dominant rather than the components and the services. However, these sectors have been collaborating with each other, already considering where they need to rely on another carrier. However, using the sector coupling concept and deploying the dependencies in an optimal centralized view requires incorporating the interdependencies and their benefits in the strategical approaches of enhancing sectors.

3.6 Conclusions This chapter presented the holistic view of the interdependencies among energy carriers in the local system with distributed energy resources. The dependencies occur due to the presence of energy converters and interconnectors in MES. Various distributed energy resources, as well as the modeling approaches of carrier dependencies that are caused by their presence in the system, were presented. It is clear that the dependencies are an important factor in MES’s studies as they can have impacts on various aspects of a MES, from operation to planning. This issue is not limited to local system operation but also affects the sector coupling and long-term planning. Several benefits can be obtained through the synergies of interdependent sectors. Although several modeling approaches have been investigated in the literature and studied in this chapter, the concern of system operators and decision-makers is the deployment of these models for taking the benefits of these interdependencies. In the analyzed case study, it was shown that the interdependency on various layers of the MES not only brings extra flexibility for the system operator to use in its operational decision-making process, but also opens new business opportunities for new roles and players in MES. These new players can address the interdependencies in a more beneficial way for the whole energy sector. With the sector coupling being among the most discussed subjects in the recent trends of studies, it can be seen that acknowledging the interdependencies and the associated complexity that comes with their benefits is the first step while the most crucial step is to model them with the most accurate possible method. This chapter provided the holistic modeling approaches and specific concerns that should be considered on multilateral interdependencies.

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

References [1] A fully-integrated internal energy market. Available at: ,https://ec.europa.eu/commission/priorities/energy-union-and-climate/fully-integrated-internal-energy-market_en.. [2] EU strategy on energy system integration. Available at: ,https://ec.europa.eu/energy/topics/ energy-system-integration/eu-strategy-energy-system-integration_en.. [3] A European green deal striving to be the first climate-neutral continent. Available at: ,https://ec. europa.eu/info/strategy/priorities-2019-2024/european-green-deal_en.. [4] Clean energy for all Europeans package. Available at: ,https://ec.europa.eu/energy/topics/energystrategy/clean-energy-all-europeans_en.. [5] IPCC Fourth assessment report: climate change: energy carriers. Available at: ,https://archive.ipcc. ch/publications_and_data/ar4/wg3/en/ch4s4-3-4.html.. [6] N. Neyestani, M. Yazdani-Damavandi, M. Shafie-khah, G. Chicco, J.P.S. Catalão, Stochastic modeling of multienergy carriers dependencies in smart local networks with distributed energy resources, IEEE Trans. Smart Grid 6 (4) (2015) 17481762. [7] P. Zhang, S. Peeta, A generalized modeling framework to analyze interdependencies among infrastructure systems, Transportation Res. Part. B: Methodol. 45 (3) (2011). [8] S.M. Rinaldi, J.P. Peerenboom, T.K. Kelly, Identifying, understanding, and analyzing critical infrastructure interdependencies, IEEE Control. Syst. 21 (6) (2001) 1125. [9] M. Geidl, G. Andersson, Optimal coupling of energy infrastructures, IEEE Lausanne Power Tech. (2007). [10] P. Favre-Perrod, A vision of future energy networks, in: IEEE Power Engineering Society Inaugural Conference and Exposition in Africa, Durban, (2005). [11] G. Chicco, P. Mancarella, Distributed multi-generation: a comprehensive view, Renew. Sustain. Energy Rev. 13 (2009) 535551. [12] T. Li, M. Eremia, M. Shahidehpour, Interdependency of natural gas network and power system security, IEEE Trans. Power Syst 23 (2008) 18171824. [13] C. Liu, M. Shahidehpour, Y. Fu, Z. Li, Security-constrained unit commitment with natural gas transmission constraints, IEEE Trans. Power Syst 24 (2009) 15231536. [14] M.Y. Damavandi, I. Kiaei, M.K. Sheikh-El-Eslami, H. Seifi, New approach to gas network modeling in unit commitment, Energy 36 (10) (2011) 62436250. [15] C. Liu, C. Lee, M. Shahidehpour, Look ahead robust scheduling of wind-thermal system with considering natural gas congestion, IEEE Trans. Power Syst 30 (2015) 544545. [16] M. Qadrdan, J. Wu, N. Jenkins, J. Ekanayake, Operating strategies for a GB integrated gas and electricity network considering the uncertainty in wind power forecasts, IEEE Trans. Sustain. Energy 5 (2014) 128138. [17] F. Barati, H. Seifi, M. Sadegh Sepasian, A. Nateghi, M. Shafie-khah, J. Catalao, Multi-period integrated framework of generation, transmission, and natural gas grid expansion planning for large-scale systems, Power Systems, IEEE Trans. on 30 (2015) 25272537. [18] C. Correa-Posada, P. Sanchez-Martin, Integrated power and natural gas model for energy adequacy in short-term operation, Power Systems, IEEE Trans. on 30 (2015) 33473355. [19] M. Shahidehpour, Y. Fu, T. Wiedman, Impact of natural gas infrastructure on electric power systems, Proc. IEEE 93 (2005) 10421056. [20] P. Fedora, Reliability review of North American gas/electric system interdependency, in: Proceedings of the 37th Annual Hawaii International Conference on System Sciences, 10, 2004. [21] H. Chen, R. Baldick, Optimizing short-term natural gas supply portfolio for electric utility companies, Power Systems, IEEE Trans. on 22 (2007) 232239. [22] J. Mejia, A. Brugman, Natural gas and electricity market issues in Colombia, in: Power Engineering Society General Meeting, IEEE, Vol. 2, pp. 19831985, 2005. [23] G. Arroyo, M. Madrigal, Growing interactions between Mexican gas markets and electricity system planning, in: Power Engineering Society General Meeting, IEEE, pp. 781783, 2005. [24] R. Chabar, M. Pereira, S. Granville, L. Barroso, N. Iliadis, Optimization of fuel contracts management and maintenance scheduling for thermal plants under price uncertainty, in: Power Systems Conference and Exposition, PSCE ’06. IEEE PES, pp. 923930, 2006.

85

86

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[25] P. Duenas, T. Leung, M. Gil, J. Reneses, Gas xelectricity coordination in competitive markets under renewable energy uncertainty, IEEE Trans. Power Syst 30 (2015) 123131. [26] S. Spiecker, Modeling market power by natural gas producers and its impact on the power system, IEEE Trans. Power Syst 28 (2013) 37373746. [27] C. Ordoudis, P. Pinson, J.M. Morales, An integrated market for electricity and natural gas systems with stochastic power producers, Eur. J. Operational Res. 272 (2) (2019). [28] R. Chen, J. Wang, H. Sun, Clearing and pricing for coordinated gas and electricity day-ahead markets considering wind power uncertainty, IEEE Trans. Power Syst. 33 (3) (2018) 24962508. [29] C. Wulf, J. Linssen, P. Zapp, Power-to-gas—concepts, demonstration, and prospects, in: C. Azzaro-Pantel (Ed.), Hydrogen Supply Chains, Academic Press, 2018, pp. 309345. [30] O. Stokke Burheim, Chapter 8hydrogen for energy storage, in: O. Stokke Burheim (Ed.), Engineering Energy Storage, Academic Press, 2017, pp. 147192. [31] X.I. Liu, J. Wu, N. Jenkins, A. Bagdanavicius, Combined analysis of electricity and heat networks, Appl. Energy 162 (2016) 12381250. [32] Z. Pan, Q. Guo, H. Sun, Interactions of district electricity and heating systems considering timescale characteristics based on quasi-steady multi-energy flow, Appl. Energy 167 (2016) 230243. [33] Y. Chen, W. Wei, F. Liu, E.E. Sauma, S. Mei, Energy trading and market equilibrium in integrated heat-power distribution systems, IEEE Trans. Smart Grid (2018). [34] Yue Chen, Wei Wei, Feng Liu, Shengwei Mei, A multi-lateral trading model for coupled gas-heatpower energy networks, Appl. Energy 200 (2017) 180191. [35] Y. Cao, W. Wei, L. Wu, S. Mei, M. Shahidehpour, Z. Li, Decentralized operation of interdependent power distribution network and district heating network: a market-driven approach, IEEE Trans. Smart Grid 10 (5) (2019) 53745385. [36] W. Wei, D. Wu, Q. Wu, M. Shafie-Khah, J.P.S. Catalão, Interdependence between transportation system and power distribution system: a comprehensive review on models and applications, J. Mod. Power Syst. Clean. Energy 7 (3) (2019) 433448. [37] N. Sathaye, S. Kelley, An approach for the optimal planning of electric vehicle infrastructure for highway corridors, Transp Res. Part. E: Logist. Transp Rev. 59 (2013) 1533. [38] G. Wang, Z. Xu, F. Wen, K.P. Wong, Traffic-constrained multiobjective planning of electricvehicle charging stations, IEEE Trans. Power Del. 28 (4) (2013) 23632372. [39] N. Neyestani, M.Y. Damavandi, T.D.P. Mendes, J.P.S. Catalão, G. Chicco, Effect of plug-in electric vehicles traffic behavior on multi-energy demand’s dependency, in: 2016 IEEE International Energy Conference (ENERGYCON), Leuven, (2016). [40] A. Hassan, S. Acharya, M. Chertkov, D. Deka, Y. Dvorkin, A hierarchical approach to multienergy demand response: from electricity to multienergy applications, Proc. IEEE 108 (9) (2020) 14571474. [41] M. Geidl, et al., Energy hubs for the future, IEEE Power Energy Mag. 5 (1) (2007) 2430. [42] M. Geidl, G. Andersson, Optimal coupling of energy infrastructures, in: Proceedings of IEEE Power Tech, Lausanne, Switzerland, (2007), pp. 13981403. [43] W. Leontief, Input-output economics, Oxford University Press, 1986. [44] G. Chicco, P. Mancarella, Matrix modeling of small-scale trigeneration systems and application to operational optimization, Energy 34 (3) (2009) 261273. [45] G. Chicco, P. Mancarella, Distributed multi-generation: A comprehensive view, Renew. Sustain. Energy Rev. 13 (3) (2009) 535551. [46] F. Adamek, M. Arnold, G. Andersson, On decisive storage parameters for minimizing energy supply costs in multicarrier energy systems, IEEE Trans. Sustain. Energy 5 (2014) 102109. [47] A. Parisio, C.D. Vecchio, A. Vaccaro, A robust optimization approach to energy hub management, Int. J. Electr. Power Energy Syst. 42 (1) (2012) 98104. [48] M. Bozchalui, S. Hashmi, H. Hassen, C. Canizares, K. Bhattacharya, Optimal operation of residential energy hubs in smart grids, IEEE Trans. Smart Grid 3 (2012) 17551766. [49] M. Almassalkhi, I. Hiskens, Cascade mitigation in energy hub networks, in: 50th IEEE Conference on Decision and Control and European Control Conference (CDC-ECC), (2011), pp. 21812188.

Modeling of multienergy carriers dependencies in smart local networks with distributed energy resources

[50] M. Almassalkhi, I. Hiskens, Impact of energy storage on cascade mitigation in multienergy systems, in: IEEE Power and Energy Society General Meeting, 2012, pp. 18. [51] A. Shahmohammadi, M. Moradi-Dalvand, H. Ghasemi, M. Ghazizadeh, Optimal design of multicarrier energy systems considering reliability constraints, IEEE Trans. Power Delivery 30 (2015) 878886. [52] S. Howell, Y. Rezgui, J. Hippolyte, B. Jayan, H. Li, Towards the next generation of smart grids: semantic and holonic multi-agent management of distributed energy resources, Renew. Sustain. Energy Rev. 77 (2017) 193214. [53] G. Santos, T. Pinto, Z. Vale, J.M. Corchado, Multi-agent semantic interoperability in complex energy systems simulation and decision support, in: 20th International Conference on Intelligent System Application to Power Systems (ISAP), 2019. [54] Z. Guevara, T. Domingos, The multi-factor energy input-output model, Energy Econ. (61)(2017) 261269. [55] R.E. Miller, P.D. Blair, Input-Output Analysis: Foundations and Extensions, Cambridge University Press, 2009. [56] A. Behzadi, E. Gholamian, P. Ahmadi, A. Habibollahzade, M. Ashjaee, Energy, exergy and exergoeconomic (3E) analyses and multi-objective optimization of a solar and geothermal based integrated energy system, Appl. Therm. Eng. 143 (2018) 10111022. [57] L. Kriechbaum, G. Scheiber, T. Kienberger, Grid-based multi-energy systems—modelling, assessment, open source modelling frameworks and challenges, Energy, Sustainability Soc. (2018). [58] S. Mittal, et al, A system-of-systems approach for integrated energy systems modeling and simulation, Presented at the Society for Modeling & Simulation International Summer Simulation Multi-Conference Chicago, IL, 2015, Available at: ,https://www.nrel.gov/docs/fy15osti/ 64045.pdf.. [59] J. María Gonzalez de Durana, O. Barambones, Agent based modeling of energy networks, Energy Convers. Manag. 82 (2014) 308319. [60] M. Moeini-Aghtaie, P. Dehghanian, M. Fotuhi-Firuzabad, A. Abbaspour, Multiagent genetic algorithm: an online probabilistic view on economic dispatch of energy hubs constrained by wind availability, IEEE Trans. Sustain. Energy 5 (2) (2014) 699708. [61] H. Lin, Q. Wang, Y. Wang, Y. Liu, N. Huang, R. Wennersten, et al., A multi-agent based optimization architecture for energy hub operation, Energy Procedia 142 (2017) 21582164. [62] M. Yazdani-Damavandi, N. Neyestani, G. Chicco, M. Shafie-khah, J.P.S. Catalão, Aggregation of distributed energy resources under the concept of multienergy players in local energy systems, IEEE Trans. Sustain. Energy 8 (4) (2017) 16791693. [63] J. Winkler, L. Dueñas-Osorio, R. Stein, D. Subramanian, Interface network models for complex infrastructure systems, J. Infrastruct. Syst. 17 (4) (2011) 138150. [64] Q. Wu, J. Zheng, Z. Jing, X. Zhou, Large-Scale Integrated Energy Systems, Planning and Operation, Springer, Singapore, 2019. [65] M. Yazdani-Damavandi, N. Neyestani, M. Shafie-khah, J. Contreras, J.P.S. Catalão, Strategic behavior of multi-energy players in electricity markets as aggregators of demand side resources using a Bi-level approach, IEEE Trans. Power Syst. 33 (1) (2018) 397411.

87

CHAPTER 4

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems Bing Yan1, Marialaura Di Somma2 and Giorgio Graditi2 1

Department of Electrical and Microelectronic Engineering, Rochester Institute of Technology, Rochester, NY, United States Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy 2

Abbreviations CCHP CHP DER DHW IES MILP SC SH

Combined Cooling Heat and Power Combined Heat and Power distributed energy resource domestic hot water integrated energy system mixed-integer linear programming space cooling space heating

Nomenclature Indices i j Parameters Acoll APV Bcin c Cbio Cgas Cgrid Cgridsell COPabs COPHP DRCCHP Ecin exbio

index of IES index of end-user solar collectors installed area (m2) PV installed area (m2) carbon intensity of natural gas (kgCO2kg) constant in Eq. (4.30) price of biomass (h/kg) price of natural gas (h/Nm3) time-of-day unit price of grid power (h/kWh) price of selling power to the grid (h/kWh) coefficient of performance of the absorption chiller coefficient of performance of the heat pump maximum ramp-down rate (kW) carbon intensity of the grid power (kgCO2/kWh) specifical chemical exergy of biomass (kWh/kg)

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00001-7

r 2021 Elsevier Inc. All rights reserved.

89

90

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DHW Exdem e Exdem SC Exdem SH Exdem exgas Exout Exsolar Fq Gcin Hdem DHW Hdem Hsolar I max ktech min ktech LHVbio LHVgas li,j PCTax Pdem PPV T0 Tout coll Treq URCCHP βj,i Δt εgen ηAB ηBatC ηBatD ηcoll ηcomb,abs ηe,GT ηHN ηHR,abs ηICE th ηICE ηPV ηsto μGT ς bio ς gas ω Variables Bboil CAB

exergy rate associated to thermal demand for DHW (kW) exergy rate associated to electricity demand (kW) exergy rate associated to thermal demand for SC (kW) exergy rate associated to thermal demand for SH (kW) specifical chemical exergy of natural gas (kWh/Nm3) total exergy output from the IES (kJ) solar exergy input rate (kW) Carnot factor carbon intensity of natural gas (kgCO2/Nm3) heating demand (kW) thermal demand for DHW at time t (kW) heat rate provided by solar collectors (kW) solar irradiance (kW) maximum generation level for the technology (kW) minimum generation level for the technology (kW) lower heat value of biomass (kWh/kg) lower heat value of natural gas (kWh/Nm3) distance between IESs and users carbon tax on CO2 emissions (h/kgCO2) electric demand (kW) power provided by PV (kW) reference temperature (K) temperature of the heat transfer fluid leaving the collectors (K) required temperature for thermal demand (K) maximum ramp-up rate (kW) heat loss factor when transferring heat from IES i to user j length of the time interval (h) exergy efficiency of generation plants efficiency of the gas-fired boiler charging efficiency of the battery discharging efficiency of the battery collector efficiency combustor’s efficiency in the absorption chiller electrical efficiency of the gas turbine generator efficiency of the heating network waste heat recovery efficiency of the absorption chiller electrical efficiency of CHP thermal efficiency of CHP electric efficiency of PV efficiency of the thermal storage heat loss rate of the gas turbine generator exergy factor of biomass exergy factor of natural gas weight in objective function mass flow rate of biomass (kg/h) heating demand provided by the auxiliary boiler for cooling demand through the absorption chiller (kW)

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

Cabs CarbonTax CCCHP Cdi Cex CICE,ex Cost Env EnvAB EnvICE Exbio Exe Exgas Exin Fobj GAB Gabs GDHW boil GSH boil Gbuy GGT GICE HAB HAB,H Hbio DHW HCCHP HHP HICE,ex Hsto Hsto,in Hsto,out ktech PBat PBatC PBatD PCCHP Pgridsell PHP PICE PPG QGT,ex xtech ξDHW ξDHW ξSC ψ

cooling rate provided by the absorption chiller (kW) carbon tax (h) total cooling rate provided by the absorption chiller (kW) cooling rate provided by the absorption chiller powered by natural gas (kW) cooling rate provided by the absorption chiller powered by exhaust gas (kW) heat rate recovered from CHP for cooling demand (kW) total energy cost (h) total CO2 emissions (kgCO2) CO2 emissions related to the auxiliary boiler (kgCO2) CO2 emissions related to CHP (kgCO2) exergy input rate of biomass (kW) exergy rate of grid power (kW) exergy input rate of natural gas (kW) total exergy input to IES (kJ) objective function volumetric flow rate consumed by the gas-fired auxiliary boiler volumetric flow rate consumed by the absorption chiller (Nm3/h) volumetric flow rate consumed by the heat recovery boiler for DHW (Nm3/h) volumetric flow rate consumed by the heat recovery boiler for SH (Nm3/h) total gas consumed by CCHP (Nm3/h) volumetric flow rate consumed by the gas turbine generator (Nm3/h) volumetric flow rate consumed by CHP (Nm3/h) heat rate provided by the auxiliary boiler (kW) heat rate provided by the auxiliary boiler for heating demand (kW) heat rate provided by the biomass boiler (kW) total heat rate provided from CCHP for DHW heat rate provided by the heat pump (kW) heat rate recovered from CHP for heating demand (kW) thermal energy stored (kWh) charging heat rate to the thermal storage (kW) discharging heat rate to the thermal storage (kW) generation level of the technology (kW) state of charge of the battery (kW) power charged to the battery (kW) power discharged from the battery (kW) power provided by CCHP (kW) power sold back to the grid (kW) power required by the heat pump (kW) power provided by CHP (kW) grid power (kW) heat rate recovered from exhaust gas in the gas turbine (kW) on/off status of the technology (binary variable) fraction of exhaust gas for DHW fraction of exhaust gas for SH fraction of exhaust gas for SC overall exergy efficiency of the IES

91

92

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

4.1 Importance of multiobjective operation optimization for shortand long-run sustainability of local integrated energy systems In past few decades, depletion of fossil resources and global warming have fostered worldwide actions to improve sustainability of the energy supply. The ongoing energy transition defines new opportunities for distributed energy resources (DER) integration at local level to achieve decarbonization of the energy system as a whole, thanks to their numerous economic and environmental benefits, allowing the possibility of integrating renewables as well as exploiting synergies among energy carriers for satisfying the users’ energy needs in a sustainable way [1
3]. Indeed, compared to traditional centralized energy supply, decentralized local energy systems with multiple DER enhance users’ self-sufficiency and foster energy supply sustainability, bringing various benefits to the final users such as reduction in energy costs, while also supporting the overall energy and environmental objectives [4
8]. In such a context, local integrated energy systems (IES) represent a valid option to integrate different energy carriers at local level via a variety of local generation of electricity, heat and cooling, and energy storage in different forms and optimized energy flow management. As a result of an integrated approach, these systems are able to satisfy the users’ multiple types of energy demands in a sustainable way through exploiting synergies among the various energy technologies and carriers, by also promoting cross-sector integration [9,10]. To reach the expected potentials of local IES and enable their wide deployment, coordinated operation of generation and conversion technologies together with storage units is crucial and also brings significant challenges both in modeling and solution strategies [11]. The optimal energy management of DER in local IES is a complex task because of the rapid change of user demand, the limited operation flexibility of certain technologies and the presence of renewables, and the priority is the necessity to balance supply and demand at all times. Another crucial aspect is related to the coexistence of multiple objectives when addressing an operation optimization problem for a local IES. Indeed, a single-objective approach is usually practical from the point of view of operators, whose greatest interest is the economic factor representative of the short-run sustainability of the energy supply system. In fact, through an economic optimization, the operators of a local IES can obtain key information about the operation strategies to apply for the various generation, conversion and storage units to minimize the daily energy costs. However, generally different stakeholders ideally participate in the management of local IES. Hence, objectives can be from different perspectives, such as operators, or the civil society that is ideally represented by the regulator [12]. Some of these operation objectives are thus naturally conflicting. For instance, the interest of the society in sustainable energy supply systems, and with low environmental impacts, might conflict with the economic interest of the operators of local IES. Consequently, there is not a single management solution, which can satisfy

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

all the stakeholders, and even if in the short-run the energy cost minimization brings the main benefits, the economic analysis alone is not sufficient to guarantee the longrun sustainability of such systems. Instead, a multiobjective approach helps to identify the compromise solutions, which benefit the different stakeholders involved in the decision-making process. Moreover, multiobjective approaches applied to the operation optimization of local IES can provide important information about the correlation between the DER integration benefits and impacts at a local level [13]. From a highlevel point of view, multiobjective analysis can thus help promote policies and incentives to encourage the local IES deployment, which ensure to exploit their benefits and minimize the associated negative impacts. For an optimization problem, when there is only one objective function, the “best solution” is one-dimensional and this corresponds to a single optimal solution. Conversely, an optimization problem with multiple objectives does not have a single optimal solution, while it has a set of optimal solutions. Therefore the multidimensional concept of dominance is needed to identify whether one solution is better than others or not. In detail, in a multiobjective optimization problem, a solution a is said to dominate solution b if the following two conditions are satisfied [14]: • In all objectives, a is no worse than b; • In at least one objective, a is better than b. In this case, it is said that b is dominated by a. In other words, a is nondominated by b. A dominated solution is also known as a suboptimal solution. The solution of a multiobjective optimization problem is characterized by the Pareto set, that is, the set of nondominated solutions. In terms of their objective functions, the Pareto set refers to the Pareto frontier. In the presence of three dimensions, the Pareto frontier is a surface, whereas when there are two objectives, it becomes a trade-off curve, consisting of the range of optimal choices available for decision makers. For a solution belonging to the Pareto frontier, no improvement in one objective is possible without harming the other objective. The information contained in the Pareto frontier elucidates compromise solutions among the stakeholders, or trade-offs among incommensurable objectives. In the literature, several efforts have been done for the economic optimization studies of DER in local IES, with reference to a certain type of technology, for example, Combined Heat and Power (CHP) or Combined Cooling Heat and Power (CCHP) systems [15
25], or for multiple DER technologies including renewables and storage. Among others, in [26] a mixed-integer linear programming (MILP) model was developed for the economic operation optimization of a local IES with multiple DER technologies with the aim to minimize the cost of electricity and natural gas. The branch-and-cut method was used to solve the problem, and results demonstrate that significant economic savings can be reached through the integrated scheduling and control of the various energy supply sources. In [27], an optimization framework was developed for the optimized operation management of a microgrid

93

94

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

with photovoltaic (PV), wind, fuel cell, micro-turbine and energy storage to minimize the energy cost. In [28], a MILP model was formulated to find the optimal operation strategies of an eco-community with multiple DER technologies to minimize the total daily energy cost. When considering also the environmental aspects, a multiobjective model was established in [29] to find the optimized operation strategies of a local IES by combining energy cost minimization with CO2 emission minimization, and the trade-off curve was obtained via a compromise programming method. The optimal management of a hybrid microgrid with renewables, back-up micro-turbine, fuel cell and battery was addressed in [30] by considering operating cost and CO2 emissions as objectives. A stochastic multiobjective optimization framework was developed in [31] to find the optimal day-ahead operation strategies of a local IES by taking into account energy costs and CO2 emissions, and the Pareto frontier was identified with the weighted-sum method. With specific reference to sustainability aspects, according to the Annex 49—ECBCS—Low Exergy Systems for High Performance Buildings and Communities [32], application of exergy principles in the context of energy supply systems can achieve rational use of energy resources through consideration of the different energy quality levels of energy resources and those of users’ energy demands. Indeed, the energy demands in buildings for heating and cooling purposes, which account for more than one third of the final worldwide energy consumption [32], are generally met by fossil fuel-based systems through combustion, which cause greenhouse gas emissions. While a lot has already been achieved especially regarding the increase of renewables penetration levels in power systems, there are still large potentials in the heating and cooling sector for satisfying the building’s energy needs. In such a context, evaluations of energy use in buildings that are mostly based on quantitative assessments through the First Law of Thermodynamics, do not take into account energy quality degradation that happens when high-quality energy resources, for example, electricity or fossil fuels, are used to meet low-quality thermal demands. Derived from the Second Law of Thermodynamics, exergy is a measure of the energy quality. It represents the maximum amount of work that can be obtained from an energy flow as it reaches the equilibrium with a reference environment [32
37], and can be viewed as the potential of a given energy amount. Different from energy, exergy is not subjected to conservation (except for reversible processes). The concept of exergy was introduced in the building sector in recent years as demonstrated by [35
38]. Energy demands in buildings are characterized by different energy quality levels. As the required temperatures for heating and cooling of indoor spaces are low, the exergy analysis shows that the quality of the energy needed to satisfy these thermal demands is also low (Carnot factor Fq  7%). The production of domestic hot water (DHW) requires temperature levels slightly higher than those for heating and air conditioning, and the quality of the energy needed to satisfy this demand is also higher (Carnot factor Fq  15%). For lighting and electrical appliances, the

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

highest quality of energy is needed (Carnot factor Fq  100%). By the matching of the energy quality levels of supply and demand, the exergy analysis promotes the coverage of low-quality thermal demands by low exergy sources, for example, solar thermal or waste heat from power generation, and electricity demands by high exergy sources, for example, fossil fuels and electricity, as shown in Fig. 4.1. In this way, sustainability of energy supply is improved by rational use of the energy resources, thereby reducing the fossil fuels consumption with the related CO2 emissions. In fact, the total CO2 emissions can be substantially reduced as a result of the reduced usage of high-quality energy resources to meet thermal demands characterized by low energy quality levels. In addition, from an economic perspective, high price stability can be expected thanks to the use of local renewables, or surplus heat recovered from power generation processes for thermal purposes. A consequent advantage is also characterized by the lower dependency on foreign fuel supplies. The above mentioned benefits can be transformed into longrun sustainability of energy supply at local level. As mentioned earlier, in the context of local IES, different energy resources, including renewable ones can be integrated, and low-temperature waste heat from power generation processes can be recovered for thermal purposes to meet the users’ multiple types of energy demands. In these applications, local IES represent an opportunity to show the benefits of exergy analysis in the improvement of energy supply sustainability at local level. This chapter focuses on the multiobjective operation optimization of DER to achieve the short- and the long-run sustainability of local IES. A multiobjective optimization framework is presented to attain the optimized operation strategies of the DER in the local IES to obtain rational use of energy resources, while satisfying timevarying user demands. This goal is achieved by considering both short- and long-run

Figure 4.1 Energy supply with sources at different energy quality levels for a typical building with demand at different energy quality levels.

95

96

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

priorities represented by the minimization of the energy costs and sustainability of the energy supply, respectively. The achievement of this latter is discussed through exergy-based assessments, as well as environmental impact aspects. The approach developed allows to identify the best possible trade-offs between the short- and longrun objectives, by giving the possibility to operators to choose the operation strategies for DER technologies from the Pareto frontier based on their priorities. Moreover, the operation optimization of DER is also discussed in the context of local energy communities characterized by the presence of multiple interconnected IESs. By exploiting synergies among interconnected IESs sharing electricity and thermal energy within an energy community, the potential benefits for both energy cost and CO2 emission reduction will be investigated. In this case, economic and environmental aspects are taken into account with CO2 emissions quantified through the carbon tax. With the general mathematical models established, the optimization frameworks developed in this chapter are flexible and scalable for potential adaptation to real contexts also considering the wide variety of generation, conversion and storage technologies modeled as DER. They can thus represent powerful tools to provide support to decision- and policy-makers in the quantification of the benefits derived by local IESs and the optimized management of local energy resources to foster efficient and rational use of available energy.

4.2 Multiobjective optimization for the operation of a local integrated energy system To achieve rational use of energy resources by considering both short- and long-run priorities, a multiobjective optimization framework is developed to identify optimized operation strategies of a local IES while satisfying time-varying user demands. The mathematical formulations and the solution methodologies are presented below.

4.2.1 Description of the local integrated energy system under study and mathematical formulation There are multiple generation and conversion technologies and storage units, and different types of end-user demands, that is, electricity, space heating (SH), space cooling (SC), and DHW, in the local IES under consideration as shown in Fig. 4.2. In detail, the generation technologies including the solar thermal plant, gas turbine, and biomass boiler convert a set of primary input energy carriers, that is, solar energy, natural gas and biomass into electricity and heat. Conversion technologies including the heat recovery boilers, absorption chiller and heat pump convert heat and electricity for SH and cooling purposes. Thermal storage units store energy for SH and SC and DHW purposes. The possible paths of energy flows from various energy resources with different energy quality levels through energy generation,

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

Figure 4.2 Scheme of the local integrated energy system for the operation optimization problem.

conversion and storage processes to meet energy demands with different energy quality which are also shown in the figure. A multiobjective MILP problem is formulated below, with the modeling of energy technologies, thermal storage units and energy balance and the formulation of the economic and the exergetic objectives. As an alternative of the exergetic objective for sustainability, the environmental objective in terms of minimization of CO2 emissions is also presented. 4.2.1.1 Modeling of DER in the local integrated energy system In this section, linear models of the biomass boiler, solar thermal plant, CCHP, and heat pump are presented. Constant conversion efficiencies are considered. A common constraint for most of these energy generation and conversion technologies (except for solar thermal plant) is the capacity constraint, formulated below: max xtech ðtÞkmin tech # ktech ðtÞ # xtech ðtÞktech ; ’t

ð4:1Þ

meaning that if the technology tech is in use at time t (i.e., the on/off binary decision variable xtech(t) is 1), its generation level ktech(t) (the decision variable) has to be within the min max minimum limit ktech , and the maximum limit (i.e., capacity) ktech , and 0 if the technology is not in use. Other constraints of individual technologies are described below. The biomass boiler is for DHW demand. To provide the amount of heat Hboil(t), the mass flow rate Bboil(t) required by the biomass boiler is given by: Bboil ðtÞ 5 Hboil ðtÞ=ðηboil;bio LHVbio Þ; ’t

ð4:2Þ

where ηboil,bio is the combustion efficiency, and LHVbio is the biomass lower heat value.

97

98

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

The solar thermal plant is also for DHW. The produced heat rate Hsolar(t) is obtained as: Hsolar ðtÞ 5 ηcoll Acoll IðtÞ; ’t

ð4:3Þ

where Acoll is the collector area which is assumed to be known, ηcoll is the efficiency, and I(t) is the hourly solar irradiance. In the CCHP system under consideration, there are: (1) a gas turbine generator for electricity demand; (2) two heat recovery boilers for SH and DWH demands by using exhaust gas; (3) and an absorption chiller to satisfy the SC demand by using exhaust gas [39]. The layout is sketched inside the bold red lines in Fig. 4.1. In addition, SH and SC demands can be directly satisfied by supplementary natural gas combustion in the boilers and the absorption chiller, respectively. The constraints for CCHP are established below. Ramp rate constraints limit the changes in the power generation between two successive time steps to be within ramp-up URCCHP and ramp-down DRCCHP and [40]: 2DRCCHP # PCCHP ðtÞ 2 PCCHP ðt 2 ΔtÞ # URCCHP ; ’t

ð4:4Þ

where PCCHP(t) and PCCHP(t
Δt) are the power provided at time t and (t
Δt), respectively, and Δt is the length of the time step. To provide power PCCHP(t), the natural gas volumetric flow rate GGT(t) in the gas turbine is formulated as: GGT ðtÞ 5 PCCHP ðtÞ=ðηe;GT LHVgas Þ; ’t

ð4:5Þ

where ηe,GT is the electric efficiency of the gas turbine, and LHVgas is the natural gas lower heat value. The heat rate QGT,ex(t) recovered from exhaust gas in the turbine is formulated as: QGT ;ex ðtÞ 5 PCCHP ðtÞð1 2 ηe;GT 2 μGT Þ=ηe;GT ; ’t

ð4:6Þ

where μGT is the heat loss rate of the gas turbine. The exhaust gas is split into three parts among heat recovery boilers and the absorption chiller for DHW and SC, and SC, respectively. Considering the absorption chiller as an example, the cooling rate Cex(t) delivered by the chiller via exhaust gas is: Cex ðtÞ 5 QGT ;ex ðtÞ ξ SC ðtÞ ηHR;abs COPabs ; ’t

ð4:7Þ

where ηHR,abs is the recovery efficiency of waste heat, COPabs is the coefficient of performance, and ξSC (t) is the percentage of heat supplied to the chiller, a continuous decision variable.

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

As mentioned earlier, SC demand can be also directly satisfied by natural gas combustion in the chiller. To directly provide cooling rate Cdi(t), the natural gas volumetric flow rate Gabs(t) in the chiller is: Gabs ðtÞ 5 Cdi ðtÞ=ðCOPabs ηcomb;abs LHVgas Þ; ’t

ð4:8Þ

where ηcomb,abs is the efficiency of the combustor. Therefore, the total cooling rate CCCHP (t) provided by the absorption chiller is the sum of the cooling rates obtained from exhaust gas and directly provided by natural gas combustion: CCCHP ðtÞ 5 Cex ðtÞ 1 Cdi ðtÞ; ’t

ð4:9Þ

Modeling of heating by CCHP for DHW and SH demands is similar to that described above. The sum of gas turbine exhaust fractions for DHW ξDHW(t) and SH ξ SH(t) in heat recovery boilers, and for SC ξSC (t) in the absorption chiller has to be 1: ξSC ðtÞ 1 ξDHW ðtÞ 1 ξSH ðtÞ 5 1; ’t

ð4:10Þ

The overall natural gas volumetric flow rate GCCHP(t) consumed by CCHP is: DHW SH GCCHP ðtÞ 5 GGT ðtÞ 1 Gabs ðtÞ 1 Gboil ðtÞ 1 Gboil ðtÞ; ’t

ð4:11Þ

The reversible heat pump has two operating modes, that is, the heating mode for SH demand and the cooling mode for SC demand. In the heating mode, to produce heating rate HHP(t), the required power PHP(t) is given by: PHP ðtÞ 5 HHP ðtÞ=COPHP ; ’t

ð4:12Þ

where COPHP is the coefficient of performance of the heating mode. Modeling of the cooling mode is similar to that described above. For the DHW tank, the energy stored Hsto(t) at time t can be expressed as follows: in out ðtÞ 2 Hsto ðtÞÞΔt; ’t Hsto ðtÞ 5 Hsto ðt 2 ΔtÞηsto 1 ðHsto

ð4:13Þ

where ηsto is the efficiency, and Hstoin(t) and Hstoout(t) are the heat rates brought in and taken out by the flow-in and flow-out water (continuous decision variables), respectively. Modeling of thermal storage systems for SH and SC is similar to that described above. 4.2.1.2 Modeling of energy balances The sum of demand Pdem(t) and the power required by the heat pump PHP(t) is equal to the sum of the power delivered by CCHP PCCHP(t), and bought from the grid PPG(t), that is, Pdem ðtÞ 1 PHP ðtÞ 5 PCCHP ðtÞ 1 PPG ðtÞ; ’t

ð4:14Þ

99

100

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DHW The DHW demand Hdem (t) is equal to the sum of the heat delivered by CCHP thermal plant Hsolar(t), the biomass boiler Hbio(t), and storage is,

DHW HCCHP (t), the solar Out HSto (t)
HStoIn(t), that

DHW DHW out in ðtÞ 5 HCCHP ðtÞ 1 Hsolar ðtÞ 1 Hbio ðtÞ 1 Hsto ðtÞ 2 Hsto ðtÞ; ’t: Hdem

ð4:15Þ

Energy balances for SH and SC are expressed in a similar way. 4.2.1.3 Economic objective The economic objective is to minimize the total energy cost Cost. This total cost has three components: costs of grid power, natural gas, and biomass, and is formulated as follows: X Cost 5 ðCgrid ðtÞPPG ðtÞ 1 Cgas Gbuy ðtÞ 1 Cbio Bbuy ðtÞÞΔt ð4:16Þ t

where Cgrid(t) is the time-of-day grid power price, and Cgas and Cbio are the constant natural gas and biomass prices, respectively. Here the volumetric flow rates of natural gas bought Gbuy(t) and biomass bought Bbuy(t) are equal to the total energy consumption of CCHP and the biomass boiler, respectively. 4.2.1.4 Exergetic objective The exergetic objective is to maximize the overall exergy efficiency of the IES under consideration. This efficiency is defined as the ratio of the total exergy required to meet the given energy demands to the total primary exergy input to the system. As mentioned earlier, energy demands of buildings are characterized by different energy quality levels. For electrical demand, the highest quality of energy is needed as electricity is fully convertible e into useful work in theory. The exergy rate Exdem (t) required is evaluated as [41]: Exedem ðtÞ 5 Pdem ðtÞ; ’t

ð4:17Þ

As for thermal demands, the temperature required for the demand under consideration mostly determines the amount of exergy—the higher the temperature, the higher the exergy. DHW The exergy rate Exdem (t) required to meet the DHW demand is evaluated as [41]: DHW ExDHW dem ðtÞ 5 Fq ðtÞHdem ðtÞ; ’t

ð4:18Þ

with the Carnot factor Fq(t) defined as, Fq ðtÞ 5 1 2 T0 ðtÞ=Treq ; ’t

ð4:19Þ

which depends on both the reference and required temperatures T0(t) and Treq. Here reference temperatures are captured by the hourly ambient temperatures [42]. The exergy required to meet SH and SC demands can be evaluated in a similar way.

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

Total exergy output Exout from the IES corresponds to the total required exergy to meet the energy demands, that is: X SH SC Exout 5 ðExedem ðtÞ 1 ExDHW ð4:20Þ dem ðtÞ 1 Exdem ðtÞ 1 Exdem ðtÞÞΔt t

In terms of input energy carriers, they are also characterized by different energy quality levels. In detail, the input energy carriers to the local IES under consideration involve grid power, natural gas, biomass and solar as discussed in the following. Here, the exergy rate of grid power is formulated as [43,44]: Exe ðtÞ 5 P PG ðtÞ=εgen ; ’t

ð4:21Þ

where the exergy efficiency εgen of grid power is based on the technologies used in the power plants. The exergy input rate of natural gas Exgas(t) is the product of the overall natural gas volumetric flow rate required by CCHP and the specific chemical exergy of natural gas1 exgas, that is, Exgas ðtÞ 5 exgas GCCHP ðtÞ; ’t

ð4:22Þ

with exgas evaluated by the product of exergy factor ς gas and lower heat value: exgas 5 ς gas LHVgas

ð4:23Þ

The exergy factor for natural gas is usually 1.04% 6 0.5% [45]. Similar to natural gas, the exergy input rate of biomass Exbio(t) is the product of the biomass volumetric mass flow rate required by the boiler and the specific chemical exergy of biomass exbio, that is, Exbio ðtÞ 5 exbio Bboil ðtÞ; ’t

ð4:24Þ

with exbio evaluated by the product of exergy factor ς bio and lower heat value: exbio 5 ς bio LHVbio

ð4:25Þ

The exergy factor for wood is usually in the range of 1.15
1.30 [45]. The solar exergy input rate Exsolar(t) is calculated at the output of solar collectors [46,47], that is, out Exsolar ðtÞ 5 Hsolar ðtÞð1 2 T0 ðtÞ=Tcoll Þ; ’t

1

ð4:26Þ

The specific chemical exergy of natural gas is the maximum amount of work that can be obtained from the substance when it reaches the chemical equilibrium with the reference environment at the constant temperature and pressure [45].

101

102

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning out where Tcoll is the temperature of the heat transfer fluid leaving collectors, which is assumed as known. The total primary exergy input to the IES Exin is the sum of exergy rates of all primary energy carriers, that is: X Exin 5 ðExe ðtÞ 1 Exgas ðtÞ 1 Exbio ðtÞ 1 Exsolar ðtÞÞΔt ð4:27Þ t

The overall exergy efficiency ψ is defined as the ratio of the total exergy output Exout to the total primary exergy input Exin, that is, ψ 5 Exout =Exin

ð4:28Þ

As mentioned earlier, given energy demands, the total exergy output Exout is known, so that the overall exergy efficiency in Eq. (4.28) can be maximized by minimizing the exergy input to the IES system. The exergetic objective is thus formulated as the total primary exergy input Exin in Eq. (4.27) to be minimized. 4.2.1.5 Environmental objective As an alternative of the exergetic objective for sustainability, the environmental objective is also presented. It is to minimize the total environmental impacts of the IES in terms of CO2 emissions from the power grid and the fossil fuels consumed. The CO2 emissions related to the use of grid power are calculated as the product of the carbon intensity of power grid Ecin and the total amount of grid power PPG(t). The power grid carbon intensity represents the amount of CO2 emissions per unit of power generated, depending on the fuel mix. The CO2 emissions caused by natural gas consumption are calculated as the product of the carbon intensity of the fuel Gcin and the total amount of fuel consumed Gbuy(t) [48], whereas the CO2 emissions due to biomass combustion in the biomass boiler are calculated as the product of the carbon intensity of the fuel Bcin and the total amount of biomass consumed Bbuy(t) [48]. Therefore, the total CO2 emissions Env to be minimized are expressed as follows, X Env 5 ðEcin PPG ðtÞ 1 Gcin Gbuy ðtÞ 1 Bcin Bbuy ðtÞÞΔt ð4:29Þ t

4.2.2 Solution methodologies With the economic objective function defined in Eq. (4.16) and the exergetic objective function defined in Eq. (4.27) (or the environmental objective in terms of CO2 emissions defined in Eq. (4.29)), the IES operation optimization problem has two objective functions. To solve this multiple-objective problem with the economic and

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

exergetic objectives, a single-objective function Fobj is considered as a weighted sum of total energy cost Cost and total primary exergy input Exin: Fobj 5 cωCost 1 ð1 2 ωÞExin

ð4:30Þ

where ω is the weight for cost, and constant c is chosen so that cCost and Exin are in the same order of magnitude. When ω is 1, the minimum energy cost and maximum exergy input are obtained. When ω is 0, the minimum exergy input (also the maximum exergy efficiency) and maximum energy cost are obtained. Then constant c is calculated as the ratio of the maximum exergy input obtained by energy cost minimization to the maximum energy cost obtained by exergy input minimization. With constant c, the Pareto frontier consisting of the best possible trade-off solutions between the two objectives are obtained by solving the problems with different values of ω varying between 0 and 1. The problem formulated above is linear and contains both discrete and continuous variables. The branch-and-cut method, a powerful method for MILP problems, is thus used. It solves the integer relaxation problem first. If the solution is feasible to the original MILP problem, it is optimal. If not, valid cuts are added to cut off regions outside the convex hull (the smallest convex set that contains all feasible solutions [49]). If the convex hull can be obtained by cuts, the problem is directly solved. If not, the method replies on time-consuming branching operations. The stopping criteria are usually the pre-set stop time or the relative mixed-integer gap (relative difference between the objectives of the current integer solution and optimal relaxed solution) [50].

4.3 Case study: eco-exergetic operation optimization of a local integrated energy system for a large hotel in Beijing In the case study, a large hotel of 30,000 m2 in Beijing is considered for the ecoexergetic operation optimization problem. A typical winter day in January and a typical summer day in July are considered, and 1 hour is considered as a time step. The multiobjective optimization model is solved by using IBM ILOG CPLEX Optimization Studio [50] on a PC with 2.90 GHz Intel Core(TM) i7 CPU and 16 GB RAM.

4.3.1 Input data The required input data consist of hourly building energy demand profiles, prices of primary energy carriers and their exergy factors, and technical data of DER technologies in the local IES.

103

104

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 4.3 Hourly energy demand profiles for the hotel in Beijing in (A) winter day in January and (B) summer day in July.

The hourly energy demand profiles for electricity, SH, SC and DHW are shown in Fig. 4.3, and they are defined based on [51]. The exergy of thermal demands is calculated by considering the hourly ambient temperatures [52] of the winter and summer days as reference temperatures. For SH, SC and DHW demands, the temperatures required are assumed as 293.15 K, 299.15 K and 333.15 K, respectively [53]. Regarding the energy prices, the time-varying grid power price shown in Fig. 4.3 is based on the energy market prices in Beijing [54], whereas the natural gas price is assumed as 0.38 $/Nm3 [54]. The wood pellet is assumed as biomass fuel with a price of 70 $/t [55]. The exergy efficiency of the power plant is set to 0.32 [56,57]. The natural gas and biomass exergy factors are set to 1.04 and 1.16, respectively [45]. In calculating the heat rate from solar collectors, for each hour of the representative winter day, the solar irradiance is calculated as the average of the hourly mean values of that in the corresponding hours in all January days, and the same approach is used for the representative summer day in July. For calculation of the solar exergy input rate, the temperature of the heat transfer fluid leaving collectors is set to 353.15 K. The technical information of the DER technologies in the local IES are shown in Table 4.1.

4.3.2 Case study results The optimization problem can be solved within 10 seconds with a relative mixedinteger gap lower than 0.1%. The Pareto frontiers consisting of the best possible tradeoffs between the economic and exergetic objectives (i.e., the short- and long-run objectives) are presented in Fig. 4.4 for the winter and summer days. In the winter day, the point a is obtained under the economic optimization with a minimum cost of 3340 $, whereas the total exergy input reaches a maximum of 100,210 kJ. Conversely,

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

Table 4.1 Technical information of DER technologies in the local integrated energy system. Generation technology

Gas turbine Biomass boiler Solar thermal Conversion technology

Size (MW)

1.25 1.0 0.41

Efficiency Electrical

Thermal

0.24

0.68 0.80 0.40

Size (MW)

Efficiency Heating mode

Cooling mode

COP 5 3.0 ηboil 5 0.90 ηHR,boil 5 0.74

COP 5 3.2

Heat pump Heat recovery boiler

2 3 3.0 1.0

Absorption chiller

6.0

Storage technology

Capacity (MWh)

Storage loss fraction

DHW storage SH storage SC storage

0.5 2.3 3.5

φsto 5 0.10 φsto 5 0.10 φsto 5 0.10

COPabs 5 1.2 ηabs 5 0.85 ηHR,abs 5 0.70

DHW, Domestic hot water; SH, space heating; SC, space cooling.

Figure 4.4 Pareto frontiers obtained for (A) winter day and (B) summer day.

the point b is obtained under the exergetic optimization with a minimum exergy input of 92,549 kJ, whereas the energy cost reaches a maximum of 3549 $. The internal points are obtained by dividing the weight interval into 100 equally-spaced points. The Pareto frontier for the summer case is obtained in a similar way, and points a0 and b0 are obtained under the economic and exergetic optimizations, respectively.

105

106

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 4.5 Optimized operation strategies of the DER in the local integrated energy system for electricity at trade-off points c (winter) and c’ (summer).

The operators of the system can choose the operation strategies to apply for the DER technologies in the Pareto frontier based on their economic or sustainability priority. Points c in the winter day is selected to show the optimized operation of the system when a higher weight is assigned to the economic objective, whereas point c0 in the summer day is selected to show the optimized operation of the system when a higher weight is assigned to the exergetic objective. The optimized operation strategies for electricity are compared at point c and c 0 in Fig. 4.5. For point c, the grid power is used in correspondence of low prices, from 0:00 to 5:00 and it is mostly used to power the heat pump. Conversely, as for higher grid prices, for example. at 15:00, 16:00, 18:00 and 19:00, the CCHP is used to satisfy the electricity demand and to power the heat pump. For point c 0 , the operations strategies are not dependent on the grid price variation as expected, considering the higher weight of the exergetic objective. For instance, from 0:00 to 5:00 although the grid price is low, the CCHP is used instead of grid power, thereby contributing to reduce the total exergy input to the local IES. The optimized operation strategies for DHW obtained for point c and c 0 are shown in Fig. 4.6. The difference in the operation strategies obtained under different weight values for the economic and exergetic objectives is evident. In the winter day, for point c, the biomass boiler is used to satisfy the demand while not the gas-fired boiler, due to the lower energy price of biomass. Conversely, in summer day for point c 0 under a higher weight for the exergetic objective, the contrary occurs and the biomass boiler is never used. This result is explained by the fact that biomass is a high-quality renewable energy resource and should not be used to satisfy low-quality energy demands since this does not promote efficient use of the energy resource potential. Finally, the optimized operation strategies for SH and SC obtained at point c and c 0 are shown in Fig. 4.7. The presented results are similar to those shown in Fig. 4.5 for electricity, since the operation strategies in the winter day are more sensitive to the grid price variation, whereas this sensitivity is much lower in the summer day.

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

Figure 4.6 Optimized operation strategies of the DER in the local integrated energy system for DHW at trade-off points c (winter) and c’ (summer).

Figure 4.7 Optimized operation strategies of DER for SH and SC at trade-off points c (winter) and c’ (summer).

In order to show the benefits achieved through local IES for both economic and exergetic objectives thanks to the multiobjective optimization framework established, a conventional energy supply system is also considered, for energy cost and exergy input comparison purposes. In the system the grid power is used to meet the total electrical load given by the sum of (1) the electricity demand; (2) the electricity required by an electric heater with 100% energy efficiency to satisfy the demand; and (3) the electricity required by an electric boiler with 98% energy efficiency to satisfy the DHW demand. In this case, the total energy cost amounts to 12,728 $, which is 3.8 times of the cost for the local IES (3340 $) obtained with the economic optimization. The total exergy input is 277,291 kJ, which is nearly three times of the exergy input for the local IES (92,548 kJ) obtained with the exergetic optimization. This result highlights that through the optimized operation of DER in local IES, both short- and long-run sustainability of the energy supply can be ensured.

107

108

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

4.4 Operation optimization of multiple integrated energy systems in a local energy community In the past few years, local energy communities have drawn increasing attention as an efficient prototype to explore local energy sources for local multiple types of energy demands. A local energy community can consist of multiple IESs which are interconnected by local grids and heating networks. To ensure short-and long-run sustainability, these multiple IESs need to be coordinated cost-effectively and environmental-friendly. In the following, a mathematical model is established for daily operation optimization of multiple IES in a local energy community with an aim to minimize the net daily energy cost and the CO2 emission cost through carbon tax. As an illustrative example, a local energy community in United States is considered, which consists of multiple IESs for different types of utility customers, including both commercial (Super Market, Strip Mall and Office Building) and residential (Midrise Apartment) sectors. In the numerical testing, a typical winter day in January is chosen, and 1 hour is considered as a time step.

4.4.1 Description of the local energy community under study and mathematical formulation The local energy community under consideration involves multiple IESs as shown in Fig. 4.8. These energy systems have their own energy devices, such as CHP, PV, absorption chiller, boiler, heat pump, battery and thermal energy storage systems. They can also share power provided by CHPs with each other via the local grids, and share the thermal energy recovered from CHPs via the heating networks. Heat transfer losses across IESs are assumed as a function of the distance of

Figure 4.8 Scheme of the local energy community with the multiple IES.

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

Figure 4.9 Energy flows intra and inter IES in the local energy community under consideration.

two energy systems. The distances among IESs are assumed known, while the distance between IES i and its associated end-user j (i.e., i 5 j) is assumed null so that heat losses within IESs are ignored. For each end-user in the IES: (1) electricity demand can be satisfied by its own CHP, PV and battery, CHPs of other IESs, and grid power; (2) heating demand can be satisfied by its own CHP, boiler, heat pump, and thermal storage, and CHPs in other IESs; and (3) cooling demand can be satisfied by its own heat pump, chiller, and thermal storage. The chiller can use thermal energy provided by the CHP and boiler in the same IES, and by CHPs in other IESs. Also when grid-connected, IESs can sell excess power generated from their CHPs back to the utility grid. These detailed energy flows among the multiple IES are shown in Fig. 4.9. For optimized operation of multiple IES in the local energy community, a MILP problem is formulated below. Modeling of energy technologies/storage systems and energy balance are presented in Sections 5.1.1 and 5.1.2, respectively. The objective is discussed in Section 5.1.3. 4.4.1.1 Modeling of DER in the local energy community In this section, linear models of the auxiliary boiler, PV, CHP system, absorption chiller, and battery in individual IES are presented. The modeling of the heat pump

109

110

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

and thermal storages are the same as in Section 3.1. A common constraint for most of these energy technologies (except for PV) is the capacity constraint described in Eq. (4.1). Other constraints of individual technologies are described below. The auxiliary gas-fired boiler (AB) is for thermal demand. To provide heat rate Hi,AB(t) (a continuous decision variable) in IES i, the volumetric flow rate Gi,AB(t) is modeled as: Gi;AB ðtÞ 5 Hi;AB ðtÞ=ðηi;AB LHVgas Þ; ’i; ’t

ð4:31Þ

where ηi,AB is the AB efficiency. The amount of heat is split into two parts to meet heating demand Hi,ABH(t), and to meet cooling demand Ci,AB(t) via the absorption chiller: Hi;AB ðtÞ 5 Hi;AB;H ðtÞ 1 Ci;AB ðtÞ; ’i; ’t

ð4:32Þ

Because of the gas combustion in the boiler, the amount of CO2 Envi,AB(t) is given by: Envi;AB ðtÞ 5 Gi;AB ðtÞLHVgas Gcin ; ’i; ’t

ð4:33Þ

PV systems are for electricity demand. The power Pi,PV(t) provided by PV of IES i is formulated as [58,59]: Pi;PV ðtÞ 5 Ai;PV ηi;PV IðtÞ; ’i; ’t

ð4:34Þ

where Ai,PV is the total area, and ηi,PV is the electrical efficiency. In CHP, an internal combustion engine is for electricity demand, and the exhaust heat from power generation is for thermal demand. Constraints are presented as follows. To provide power Pi,ICE(t) (a continuous decision variable) in IES i, the natural gas volumetric flow rate Gi,ICE(t) required by the internal combustion engine is given by: Gi;ICE ðtÞ 5 Pi;ICE ðtÞ=ðηi;ICE LHVgas Þ; ’i; ’t

ð4:35Þ

where ηi,ICE is the electrical efficiency. Different from CCHP mentioned in Section 3.1.1.3, the CHP is considered as lumped without heat recovery boilers for simplicity. The heat Hi,ICE,ex(t) in the exhaust gas recovered from the combustion engine is expressed as: Hi;ICE;ex ðtÞ 5 Pi;ICE ðtÞηthi;ICE =ηi;ICE ; ’i; ’t where ηthi;ICE is the thermal efficiency.

ð4:36Þ

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

The recovered heat is split into two parts to satisfy heating demand Hi,ICE,ex(t), to satisfy cooling demand Ci,ICE,ex(t) through the absorption chiller, and to share with other IESs via the heating networks for heat Hi,j,ICE,ex(t) and cooling Ci,j,ICE,ex(t) (all continuous decision variables). X Hi;ICE;ex ðtÞ 1 Ci;ICE;ex ðtÞ 1 ðHi;j;ICE;ex ðtÞ 1 Ci;j;ICE;ex ðtÞÞ 5 Hi;ICE;ex ðtÞ; ’i; ’t ð4:37Þ j:j6¼i Because of the gas combustion in the engine, the amount of CO2 Envi,ICE(t) is formulated as: Envi;ICE ðtÞ 5 Gi;ICE ðtÞLHVGas Gcin ; ’i; ’t

ð4:38Þ

The absorption chiller is for cooling demand. It is powered by the thermal energy provided by CHP, boiler, and the other IESs. The cooling Ci,abs(t) provided by the chiller is formulated as: ! X Ci;abs ðtÞ 5 Ci;ICE;ex ðtÞ 1 Ci;AB ðtÞ 1 Cj;i;ICE;ex ðtÞUηj;i;HN COPi;abs ; ’i; ’t ð4:39Þ j:j6¼i where ηj,i,HN is the efficiency of the heating networks from IES j to i. Since heat losses between IESs are assumed proportional to the distance between the two energy systems, ηj,i,HN is evaluated as: ηj;i;HN 5 1 2 β j;i lj;i ; ’i; ’j

ð4:40Þ

where β j,i is the heat loss factor when transferring heat from IES j to i, and lj,i is the distance from IES j to i. Battery state dynamics are considered with charging and discharging efficiencies ηi, BatC and ηi,BatD, respectively. With the state of charge at time t denoted as Pi,Bat(t), the standard state of charge equation on is modeled as, Pi;Bat ðt 1 1Þ 5 Pi;Bat ðtÞ 1 ηi;BatC Pi;BatC ðtÞ 2 Pi;BatD ðtÞ=ηi;BatD ; Pi;BatC ðtÞ # Pi;PV ðtÞ; ’i; ’t ð4:41Þ In the above, decision variables are Pi,BatC(t) and Pi,BatD(t), power charged to and discharged from the battery, respectively. In addition, the battery cannot be charged and discharged at the same time. 4.4.1.2 Modeling of energy balances The sum of the power demand Pi,dem(t), and the power required by the heat pump Pi,HP(t), charged to the battery Pi,BatC(t), sent to other IESs Pi,j(t) (a continuous

111

112

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

decision variable), and sold back to the utility grid Pi,gridsell(t) (a continuous decision variable), is equal to the sum of the power delivered by CHP Pi,ICE(t), PV systems Pi,PV(t) and discharged from battery Pi,BatD(t), obtained from the other IESs Pj,i(t) (a continuous decision variable), and bought from the utility grid PPG(t) (a continuous decision variable), i.e., X ð4:42Þ Pi;dem ðtÞ 1 Pi;HP ðtÞ 1 Pi;BatD ðtÞ 1 Pi;gridsell ðtÞ 1 ðPi;j ðtÞÞ j:j6¼i 5 Pi;ICE ðtÞ 1 Pi;PV ðtÞ 1 Pi;BatD ðtÞ 1 Pi;PG ðtÞ X X 1 ðPj;i ðtÞÞ; ðPi;j ðtÞÞ # Pi;ICE ðtÞ; ’i; ’t j:j6¼i j:j6¼i The heat demand Hi,dem(t) is equal to the sum of the heat delivered by CHP Hi, StoOut (t) - HiStoIn(t), and from the ICE,ex(t), the auxiliary boiler Hi,AB(t), the storage Hi other IESs, that is, X Hi;dem ðtÞ 5 Hi;ICE;ex ðtÞ 1 Hi;AB ðtÞ 1 ηHN;j;i Hj;i;ICE;ex ðtÞ 1 Histo;out ðtÞ 2 Histo;in ðtÞ; ’i; ’t j:j6¼i ð4:43Þ The cooling energy balance can be expressed in a similar way. 4.4.1.3 Objective function The objective is to minimize the total daily cost of operation, that is, the sum of the net energy cost and the CO2 emission cost. Net energy cost Cost involves three components, the cost of natural gas, the cost of grid power, and the revenue of selling power back to the utility grid, that is, XX Cost 5 ðCgas ðGi;ICE ðtÞ 1 Gi;AB ðtÞÞ 1 Cgrid ðtÞPi;PG ðtÞ 2 Ci;gridsell ðtÞPi;gridsell ðtÞÞΔt i

t

ð4:44Þ where Ci,gridsell(t) is the unit price for selling power to the grid for IES i at time t. The cost of CO2 emissions because of gas combustion in the CHP combustion engine and the auxiliary boiler is quantified by carbon tax CarbonTax [60], that is, XX CarbonTax 5 P CTax ðEnvi;ICE ðtÞ 1 Envi;AB ðtÞÞΔt ð4:45Þ i

t

where PCTax is the carbon tax on CO2 emissions. For grid power, its CarbonTax is already reflected in the grid price, and thus not involved here. The overall objective function is Cost 1 CarbonTax.

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

4.4.2 Case study: eco-environmental optimization of a local energy community in the United States The above optimization problem is solved by using IBM ILOG CPLEX Optimization Studio on the same PC mentioned in Section 3. In this case study, a local US energy community located is considered. The energy community under investigation consists of multiple IESs satisfying the energy needs of a set of users belonging to commercial and residential sectors, namely strip mall, supermarket, a cluster of office buildings, and midrise apartment. The numerical testing is carried out for a typical winter day in January, and 1 hour is considered as a time step. The input data are presented in Section 5.2.1. The results of the optimization problem under different operation modes of multiple IESs are discussed in Section 5.2.2, whereas in Section 5.2.3, optimized results for one IES are presented in detail. 4.4.2.1 Input data The input data of the optimization problem consist of the heating network structure, hourly energy demands, energy prices, hourly solar irradiance profiles, carbon intensity, and technical information of DER technologies as discussed in the following. The IES for super market, strip mall, small office, and midrise apartment are defined as IES1, IES2, IES3 and IES 4, respectively. The heating networks are assumed to be a one directional rectangle (i.e., lj,i 6¼ li,j). The distances from IES1 to IES2, from IES2 to IES3, from IES3 to IES4, and from IES4 to IES1 are assumed as 50, 100, 50, and 100 (meters), respectively. The heat loss factor is assumed as 4 3 1025 according to [61]. The hourly electricity and heating demand profiles of the end-users are shown in Fig. 4.10 (A) and (B), respectively [62]. Regarding energy prices, the time-of-day grid price and the natural gas price are chosen based on the current US market. In detail, grid prices ($/kWh) are based on Energy rate demand (kW)

Energy rate demand (kW)

300 250 200 150 100 50 0 1:00 4:00 7:00 10:00 13:00 16:00 19:00 22:00 Time SuperMarket

Strip mall

Office building

(A)

Midrise apartment

500 450 400 350 300 250 200 150 100 50 0 1:00 4:00 7:00 10:00 13:00 16:00 19:00 22:00 Time SuperMarket

Strip mall

Office building

Midrise apartment

(B)

Figure 4.10 (A) Hourly profiles for power demand in the winter day. (B) Hourly profiles for heating demand in the winter day.

113

114

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

EverSource tariff Rate 27 (commercial) and 7 (residential) [63] (because the demand charge is not considered, the grid price is set as 2.5 times of the sum of the remaining charges; and the monthly distribution service charge is evenly distributed to every day). The price of the excess electricity sold back to the utility grid is assumed as 48% of the grid price at the same hour. For the natural gas, the unit price ($/Nm3) is based on data from Energy Information Administration [64]. The hourly solar irradiance profiles, which are needed to calculate the PV power, are defined through local meteorological information [52]. In detail, the hourly solar irradiance profiles for the representative January winter day are obtained by using the average hourly values of solar irradiance of all the days in January. The natural gas carbon intensity is assumed as 0.202 kg/kWh (the carbon tax for year 2015) [60]. The technological information of DER technologies belonging to the four IESs (for super market, strip mall, office building, and midrise apartment) are summarized in Table 4.2. 4.4.2.2 Case study results In order to investigate the effectiveness of the optimization model for the economic and environmental performance of local energy communities, various operation modes of the multiple IESs are analyzed: • Case a (current case): the multiple IESs can share electricity and thermal energy among each other, and can also sell excess electricity back to the utility grid. • Case b: the multiple IESs do not share energy, but can sell excess electricity back to the utility grid. • Case c: the multiple IESs can share energy, but are connected to the grid. • Case d: the multiple IESs do not share energy, and are not connected to the grid. The optimized results obtained for the cases above are reported in Table 4.3 in terms of energy costs, carbon tax and revenues, and net energy costs. The current case gives the minimum total net energy cost. For case b represented by IESs which cannot share energy but can sell excess electricity back to the utility grid, the total net energy cost increases by 1.45% but the revenue of selling electricity back to the grid decreases by 47.4%. This highlights that through the exploitation of interplay among energy carriers and related technologies belonging to the different IESs, a larger amount of electricity can be sold back to the utility grid thereby allowing the reduction of the total net energy cost. The optimized performance of the energy community shows a worsening when the IESs work in the islanded mode. As compared to case a, the net energy cost of case c increases by 8.44%, while case d shows the worst performance among the four cases with a cost increase of 10.53%. Fig. 4.11 shows the optimized IES operation strategies for electricity in the energy community for the various cases.

Table 4.2 Sizes and efficiencies of DER technologies in the four IESs. Size (kW)-(m2)-(kWh)

Efficiency

IES1

IES2

IES3

IES4

IES1

IES2

IES3

IES4

Auxiliary boiler PV panels: m2 Heat pump CHP

160 1600 430 425

65 590 175 155

60 1370 170 355

40 430 60 115

Battery (kWh)

200

50

180

30

Thermal storage (kWh)

400

200

300

100

0.90 0.14 3.5 ηICE 5 0.34 ηICEth 5 0.48 ηBatC 5 0.75 ηBatD 5 0.75 0.90

0.90 0.10 3.5 ηICE 5 0.30 ηICEth 5 0.50 ηBatC 5 0.80 ηBatD 5 0.80 0.90

0.90 0.10 3.5 ηICE 5 0.30 ηICEth 5 0.50 ηBatC 5 0.80 ηBatD 5 0.80 0.90

0.90 0.10 3.5 ηICE 5 0.30 ηICEth 5 0.50 ηBatC 5 0.80 ηBatD 5 0.80 0.90

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 4.3 Optimized results under different operation modes of the multiple integrated energy systems. Case

Case Case Case Case

a b c d

Energy cost ($)

Carbon tax ($)

Revenue ($)

Total net energy cost ($)

3380.17 3365.45 3525.09 3593.17

115.39 114.07 110.4 112.53

137.93 72.55 0 0

3357.57 3406.87 3635.49 3705.70

12000 Case a

8000 Electricity (kWh)

116

4000

Case b Case c Case d

0 –4000 –8000 –12000

Figure 4.11 Optimized IES operation strategies for electricity in the energy community for the various cases in the winter day.

It can be noticed that electricity sold back to the utility grid in case a is slightly higher than that in case b, thereby increasing the revenue. Cases a and b have the same total electrical demand, and the amount of electricity provided by PVs are almost the same, as well as that provided by batteries. As compared to the case b, the increase of grid power in case a is lower than the increase of the electricity from the CHP, thanks to the sharing of electricity among IESs. This operation strategy brings to the minimization of the net energy cost of the local energy community. When the local energy community is not connected to the utility grid, the total electrical demand for both cases c and d is lower than that of cases a and b. This result can be explained by the lower usage of the heat pumps for thermal demand. Another key result is that CHPs are the only technology used to meet the electrical demand of the local energy community beyond PV systems in both cases c and d. This restricts the benefits of utilizing grid power, which is more convenient than CHPs in the hours of the day characterized by low prices. The latter result explains the energy cost increase for cases c and d as compared to those of cases a and b. Fig. 4.12 shows the IES optimized operation strategies for heat in the energy community for the various cases. For case a, the lowest amount of thermal energy from CHPs is self-used in the IESs. This is due to the sharing of electricity and thermal energy among the multiple IESs and

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

Thermal enegry (kWh)

15000 Case a

10000

Case b Case c

5000

Case d

0 –5000 –10000

Figure 4.12 Optimized IES operation strategies for heat in the energy community for the various cases in the winter day. Table 4.4 Cases a and c: Exchange of thermal energy among IESs and users. “x” show there is energy exchange between the two entities. From

IES1

IES2

IES3

IES4

Self-use x -

Self-use -

Self-use x

Self-use

Self-use x x x

Self-use x

x x Self-use x

Self-use

To user

Case a 1. 2. 3. 4.

Supermarket Strip mall Office building Midrise apartment

Case c 1. 2. 3. 4.

Supermarket Strip mall Office building Midrise apartment

the usage of heat pumps powered by electricity from the grid. For case b, without energy sharing among IESs, the thermal energy from CHP for self-use increases, with the consequent increase in the thermal storage usage. When the local energy community is not connected to the utility grid, the usage of heat pumps significantly reduces for case d, and it is zero for case c. This also explains the increase of the energy cost, because the heat pump is a convenient technology from the economic point of view because of the high energy conversion efficiency. Instead, for case d, without energy sharing among the IESs, the large amount of the self-use thermal energy allows to meet the heating demand excluding the need to integrate other heat technologies. The presence of exchange of thermal energy among IESs and end-users for cases a and c are presented in Table 4.4. Note that self-use means that the thermal energy usage of an end-user from its own IES. For case a, the thermal energy exchange occurs between

117

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

120

0.5

100

0.4

80

0.3

60 0.2

40

0.1

20 0

0 1

3

5

7

9

11

13

15

17

19

21

Energy price ($/kWh)

Power

IES1 and the strip mall, and between IES3 and the midrise apartment, due to the shorter distances, which allows to reduce the heat losses occurring in pipelines. For case c, without grid power, heat pumps are never used to meet the heating demand as discussed earlier, which leads to a large amount of thermal energy share among IESs, and corresponding a large amount of heat losses in pipelines due to the longer distances. For the illustration purpose, the optimized operation strategies for electricity of IES2 in the current case a are presented in Fig. 4.13. Results of the other IESs in the local energy community are similar. The power balance and the optimized CHP operation for the strip mall are shown in Fig. 4.13 (A) and (B), respectively.

23

Hour of the day Grid power Net PV Grid price_buy

CHP self-use Bat

From other IES Total load

(A) 50 Electricity sold

To other DESs

CHP self-use

40

Power (kW)

118

30 20 10 0 1

3

5

7

9

11

13

15

17

19

21

23

Hour of the day

(B) Figure 4.13 ( (A) IES2: Power balance in the winter day. (B) IES2: Optimized CHP operation strategies in the winter day.

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

Heat rate (kW)

200 150 100 50 0 1

3

5

7

9

11

13

15

17

19

21

23

Hour of the day Net CHP self-use

AB

HP

Other IES

TES

Demand

Figure 4.14 IES2: Heat energy balance in the winter day.

In Fig. 4.13(A), it can be seen that the grid power is used to satisfy the electricity demand in correspondence of low grid prices, for instance from hours 1 to 12, and from hours 22 to 24, thereby allowing the energy cost minimization for IES2. Conversely, in correspondence of high grid prices, for instance from hours 13 to 21, the demand is mostly covered by the storage and other IESs in the energy community. In Fig. 4.13(B), it can be seen that the power provided by the CHP is only used for self-use within the local energy community, and it is never sold back to the utility grid. The heat energy balance for the strip mall is presented in Fig. 4.14. It can be seen that the thermal energy in the storage charged by CHP is used to meet the peak demand of hour 7. When the grid price is low, the heat pump is generally used until hour 12. The demand at hour 13 to the end of the day is completely satisfied by the thermal energy from other IESs and in particular by IES1 as also shown in Table 4.3.

4.5 Conclusions and key findings This chapter presents a multiobjective operation optimization framework for local IESs with multiple DER and energy carriers. The goal is to optimize the operation strategies of the DER by taking into account both economic and sustainability priorities. The latter is discussed through exergy and environmental analysis. In detail, the exergy analysis promotes the matching of the energy quality of input energy carriers to the IES and users’ energy demands, thereby fostering the usage of low-quality waste heat to satisfy thermal demands and reducing the waste usage of high-quality energy resources such as fossil fuels and electricity with consequent reduction of related CO2 emissions. To solve the multiobjective optimization problem, a weighted-sum method

119

120

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

is used to obtain the Pareto frontier consisting of the range of optimized choices available for decision makers. The general mathematical modeling of a wide variety of heat and power technologies and the solution methodology developed makes the optimization models scalable and adaptable to real contexts thereby representing a powerful tool to support decision-makers in the management of local energy resources taking into account short- and long-run priorities. Indeed, generally different stakeholders ideally participate in the management of local IESs, and objectives can be defined from different perspectives, being naturally conflicting. Consequently, there is no single optimal solution that can be beneficial to all the stakeholders, and even if in the short-run the energy cost minimization brings the main benefits, the economic analysis alone is not sufficient to guarantee the long-run sustainability of such systems, and a multiobjective approach may be required. The operation optimization of DER is also discussed in the context of local energy communities characterized by the presence of multiple interconnected IESs. In this case, economic and environmental aspects are taken into account with CO2 emissions quantified through the carbon tax. The effectiveness of the optimization frameworks established is tested through different case studies. A large hotel in Beijing is chosen for the eco-exergetic optimization of a local IES, and results demonstrate that a strong reduction of energy costs and primary exergy input can be achieved as compared to conventional energy supply systems through the optimization model developed. Moreover, results also demonstrate that the primary exergy input minimization promotes an efficient energy supply system so that all energy resources, including renewable ones, are used in efficiently. In detail, the CCHP system, heat pumps and solar thermal play a crucial role from an exergy perspective. Conversely, the biomass boiler is found to be a good solution for energy costs, due to the low fuel price, but a solution to be avoided for the exergetic objective, since high-quality biomass should not be used to meet the low-quality thermal demand. The model developed for the eco-environmental optimization of multiple IESs is tested for a local energy community in the United States, consisting of a set of users belonging to both commercial and residential sectors, including strip mall, supermarket, office buildings, and midrise apartment. Results show that by exploiting the synergies among interconnected IESs sharing electricity and thermal energy within the energy community, the benefits for energy cost and CO2 emission reduction are evident in comparison to what occurs under the other operation modes where there are no interconnections among IESs.

References [1] A. Kari, S. Arto, Distributed energy generation and sustainable development, Renew. Sustain. Energy Rev. 10 (2006) 539
558. [2] M.F. Akorede, H. Hizam, E. Poresmaeil, Distributed energy resources and benefits to the environment, Renew. Sustain. Energy Rev. 14 (2010) 724
734.

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

[3] G. Pepermans, J. Driesen, D. Haesoldoncklx, R. Belmans, W. D'haeseleer, Distributed generation: definition, benefits and issues, Energy Policy 33 (2005) 787
798. [4] B.P. Koirala, E. Koliou, J. Friege, R.A. Hakvoort, P.M. Herder, Energetic communities for community energy: a review of key issues and trends shaping integrated community energy systems, Renew. Sustain. Energy Rev. 56 (2016) 722
744. [5] J.C. Rogers, E.A. Simmons, I. Convery, A. Weatherall, Public perceptions of opportunities for community-based renewable energy projects, Energy Policy 36 (11) (2008) 4217
4226. [6] C. Rae, F. Bradley, Energy autonomy in sustainable communities—a review of key issues, Renew. Sustain. Energy Rev. 16 (9) (2012) 6497
6506. [7] J. Garcia-Guarin, D. Rodriguez, D. Alvarez, S. Rivera, C. Cortes, A. Guzman, et al., Smart microgrids operation considering a variable neighborhood search: the differential evolutionary particle swarm optimization algorithm, Energies 12 (16) (2019) 31
49. [8] R. Roberto, et al., A multi-objective optimization analysis to assess the potential economic and environmental benefits of distributed storage in district heating networks: a case study, Int. J. Sustain. Energy Plan. Manag. (2019) 5
19. [9] Foiadelli F., et al. Optimal design of DER for economic/environmental sustainability of local energy communities, in: 2018 IEEE International Conference on Environment and Electrical Engineering and 2018 IEEE Industrial and Commercial Power Systems Europe. [10] B. Yan, M. Di Somma, G. Graditi, P.B. Luh, Markovian-based stochastic operation optimization of multiple distributed energy systems with renewables in a local energy community, Electr. Power Syst. Res. 186 (2020) 106364. [11] Z. Zhou, J. Zhang, P. Liu, Z. Li, M.C. Georgiadis, E.N. Pistikopoulos, A two-stage stochastic programming model for the optimal design of distributed energy systems, Appl. Energy 103 (2013) 135
144. [12] A. Alarcon-Rodriguez, G. Ault, S. Galloway, Multi-objective planning of distributed energy resources: a review of the state-of-the-art, Renew. Sustain. Energy Rev. 14 (5) (2010) 1353
1366. [13] G. Celli, E. Ghiani, S. Mocci, F. Pilo, A multi-objective evolutionary algorithm for the sizing and siting of distributed generation, IEEE Trans. Power Syst. (2005) 20. [14] K. Deb, Multi-objective optimization using evolutionary algorithms, John Wiley and Sons, 2001. ISBN 047187339X. [15] O.A. Shaneb, P.C. Taylor, G. Coates, Optimal online operation of residential mCHP systems using linear programming, Energy Build. 44 (2012) 17
25. [16] X.Q. Kong, R.Z. Wang, Y. Li, X.H. Huang, Optimal operation of a microcombined cooling, heating and power system driven by a gas engine, Energy Convers. Manag. 50 (3) (2009) 530
538. [17] Z. Wu, W. Gu, R. Wang, X. Yuan, W. Liu, Economic optimal schedule of CHP microgrid system using chance constrained programming and particle swarm optimization, in: Proceedings of 2011 IEEE Power and Energy Society General Meeting, pp. 1
11. [18] A.D. Hawkes, M.A. Leach, Cost-effective operating strategy for residential microcombined heat and power, Energy 32 (5) (2007) 711
723. [19] M. Houwing, A.N. Ajah, P.W. Heijnen, I. Bouwmans, P.M. Herder, Uncertainties in the design and operation of distributed energy resources: the case of micro-CHP systems, Energy 33 (10) (2008) 1518
1536. [20] M. Motevasel, A.R. Seifi, T. Niknam, Multi-objective energy management of CHP (combined heat and power)-based micro-grid, Energy 51 (2013) 123
136. [21] T. Wakui, R. Yokoyama, K. Shimizu, Suitable operational strategy for power interchange operation using multiple residential SOFC (solid oxide fuel cell) cogeneration systems, Energy 35 (2) (2010) 740
750. [22] J. Wang, Y. Jing, C. Zhang, Optimization of capacity and operation for CCHP system by genetic algorithm, Appl. Energy 87 (2010) 1325
1335. [23] P. Roque Díaz, Y.R. Benito, J.A.R. Parise, Thermoeconomic assessment of a multi-engine, multiheat-pump CCHP (combined cooling, heating and power generation) system a case study, Energy 35 (2010) 3540
3550. [24] G. Chicco, P. Mancarella, Matrix modelling of small-scale trigeneration systems and application to operational optimization, Energy 34 (2009) 261
273.

121

122

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[25] J. Ortiga, J.C. Bruno, A. Coronas, Selection of typical days for the characterisation of energy demand in cogeneration and trigeneration optimisation models for buildings, Energy Convers. Manag. 52 (2011) 1934
1942. [26] X. Guan, Z. Xu, Q. Jia, Energy-efficient buildings facilitated by microgrid, IEEE Trans Smart Grid 1 (2011) 466
473. [27] S. Mohammadi, S. Soleymani, B. Mozafari, Scenario-based stochastic operation management of microgrid including wind, photovoltaic, micro-turbine, fuel cell and energy storage devices, Int. J. Electr. Power Energy Syst. 54 (2014) 525
535. [28] B. Yan, P.B. Luh, B. Sun, C. Song, C. Dong, Z. Gan, et al., Energy-efficient management of ecocommunities, in: Proceedings of IEEE CASE, Madison, WI, 2013. [29] R. Hongbo, Z. Weisheng, N. Ken’Ichi, G. Weijun, W. Qiong, Multi-objective optimization for the operation of distributed energy systems considering economic and environmental problems, Appl. Energy 87 (2010) 3642
3651. [30] A.A. Moghaddam, A. Seifi, T. Niknam, M.R.A. Pahlavani, Multi-objective operation management of a renewable MG (micro-grid) with back-up micro-turbine/fuel cell/battery hybrid power source, Energy 36 (11) (2011) 6490
6507. [31] M. Di Somma, G. Graditi, E. Heydarian-Forushani, M. Shafie-Khah, P. Siano, Stochastic optimal scheduling of distributed energy resources with renewables considering economic and environmental aspects, Renew. Energy 116 (2018) 272
287. [32] ECBCS—Annex 49—Low exergy systems for high performance buildings and communities. 2010. Available: , http://www.ecbcs.org/annexes/annex49.htm . . [33] J. Szargut, International progress in second law analysis, Energy 5 (1980) 709
718. [34] M.J. Moran, Availability Analysis: A Guide to Efficient Energy Use, revised ed., ASME, New York, 1990. [35] F. Björk, Ș. Kilki¸s, M. Molinari, Energy quality management and low energy architecture, in: ASES National Solar Conference, NC, 2011. [36] Ș. Kilki¸s, A rational exergy management model for curbing building CO2 emission, ASHRAE 113 (2007) 113
123. [37] M. Molinari, Exergy and Parametric Analysis: Methods and Concepts for a Sustainable Built Environment, Royale Institute of Technology, Stockolm, 2012. [38] M. Molinari, Exergy Analysis in Buildings: A Complementary Approach to Energy Analysis, Stockolm, KTH- Stockolm, Sweden, 2009. [39] X.Q. Kong, R.Z. Wang, X.H. Huang, Energy optimization for a CCHP system with available gas turbines, Appl. Therm. Eng. 25 (2005) 377
391. [40] R. Aiying, L. Risto, An effective heuristic for combined heat-and-power production planning with power ramp constraints, Appl. Energy 84 (2007) 307
325. [41] H. Lu, K. Alanne, I. Martinac, Energy quality management for building clusters and districts (BCDs) though multi-objective optimization, Energy Convers. Manag. 79 (2014) 525
533. [42] A. Angelotti, P. Caputo, The exergy approach for the evaluation of heating and cooling technologies, first results comparing steady state and dynamic simulations, Proceedings of the 2nd PALENC and 28th AIVC Conference, vol. 1, Crete, 2007. [43] L.M. Ramirez-Elondo, G.C. Paap, R. Ammerlaan, R.R. Negenborn, R. Toonssen, On the energy, exergy and cost optimization of multi-energy-carrier power systems, Int. J. Exergy 13 (2013) 364
385. [44] T. Krause, F. Kienzle, S. Art, G. Andersson, Maximizing exergy efficiency in multi-carrier energy systems, in: Proceedings of IEEE Power and Energy Society General Meeting, Minneapolis, MN, 25
29, 2010. [45] Y.J. Kotas, The Exergy Method for Thermal Plant Analysis, reprint ed., Krieger, Malabar, FL, 1995. [46] H. Torìo, A. Angelotti, D. Schmidt, Exergy analysis of renewable energy-based climatisation systems for buildings: A critical view, Energy Build. 41 (2009) 248
271. [47] H. Torìo, D. Schmidt, Framework for analysis of solar energy systems in the built environment from an ex H, Schmidt D. Framework for analysis of solar energy systems in the built environment from an exergy perspective, Renew. Energy 35 (2010) 2689
2697.

Multiobjective operation optimization of DER for short- and long-run sustainability of local integrated energy systems

[48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64]

Educogen, The European Educational Tool on Cogeneration, 2nd Ed., December 2001. D.P. Bertsekas, Nonlinear Programming, third ed., Athena Scientific, 2016. IBM ILOG CPLEX V 12.1, User’s Manual. Z. Zhou, P. Liu, Z. Li, W. Ni, An engineering approach to the optimal design of distributed energy systems in China, Appl. Therm. Eng. 53 (2013) 387
396. ASHRAE International weather files for energy calculations (IWEC weather files). Users Manual and CD-ROM, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, GA, 2001. ANSI/ASHRAE Standard 55, Thermal environmental conditions for human occupancy; 2013; Available from: ,http://www.techstreet.com/products/1868610.. Beijing Municipal Commission of DevelopCurrent Prices for Public Commodities. J. Chau, T. Sowlati, S. Sokhansanj, F. Preto, S. Melin, S. Bi, Techno-economic analysis of wood biomass boilers for the greenhouse industry, Appl. Energy 86 (2009) 364
371. S.C. Kaushik, V.S. Reddy, S.K. Tyagi, Energy and exergy analyses of thermal power plants: a review, Renew. Sustain. Energy Rev. 15 (2011) 1857
1872. Forum
Oxford Institute for Energy Studies, Issue 95, 2014. Available from: ,http://www. oxfordenergy.org/wpcms/wp-content/uploads/2014/04/OEF-95.pdf.. E.D. Mehleri, H. Sarimveis, N.C. Markatos, L.G. Papageorgiou, A mathematical programming approach for optimal design of distributed energy systems at the neighbourhood level, Energy 44 (2012) 96
104. C. Weber, N. Shah, Optimisation based design of a district energy system for an eco-town in the United Kingdom, Energy 36 (2) (2011) 1292
1308. Options and Considerations For a Federal Carbon Tax, Available at: ,http://www.c2es.org/ publications/options-considerations-federal-carbon-tax.. S. Obara, Equipment arrangement planning of a fuel cell energy network optimised for cost minimisation, Renew. Energy 32 (3) (2007) 382
406. Data base for state and local energy profiles. Available at: ,https://www.energy.gov/eere/buildings/ building-technologies-office.. ,https://www.eversource.com/content/ct-c/residential/my-account/billing-payments/about-yourbill/rates-and-tariffs/electric-tariffs-rules.. ,https://www.eia.gov/dnav/ng/ng_pri_sum_dcu_SCT_m.htm..

123

CHAPTER 5

Impact of neighborhood energy trading and renewable energy communities on the operation and planning of distribution networks Alberto Borghetti1, Camilo Orozco Corredor1, Carlo Alberto Nucci1, Ali Arefi2, Javid Maleki Delarestaghi2, Marialaura Di Somma3 and Giorgio Graditi3 1

Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy College of Science, Health, Engineering and Education, Murdoch University, Perth, WA, Australia Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy 2 3

Abbreviations Acronyms relevant to the local energy community model: LEC stands for local/citizen/renewable energy community ADMM corresponds to alternating direction method of multipliers EMS denotes the energy management system BES unit corresponds to battery energy storage unit PV systems corresponds to photovoltaic systems LV and MV corresponds to low voltage and medium voltage, respectively OF denotes the objective function Acronyms relevant to the distribution network planning model: NS corresponds to network solutions NNS corresponds to nonnetwork solutions DR corresponds to demand response ESS corresponds to energy storage system DG corresponds to distributed generation NET stands for neighborhood energy trading MSDEP corresponds to multistage distribution expansion planning DER corresponds to distributed energy resources PDF corresponds to the probability distribution function PSO corresponds to particle swarm optimization NPV is the net present value POE corresponds to probability of exceedance DSNS stands for demand supplied by NSs SD corresponds to standard deviation LDC stands for load duration curve GMM stands for Gaussian mixture model ETS stands for event tree analysis Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00007-8

r 2021 Elsevier Inc. All rights reserved.

125

126

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

FTA CDF SAIDI SAIFI D ESS/DG VCR MPSO GA RMC VR

stands for fault tree analysis corresponds to cumulative distribution function corresponds to the system average interruption duration index corresponds to the system average interruption frequency index is the maximum effective load corresponds to the nonnetwork solution based on DG or ESS corresponds to the value of customer reliability stands for modified PSO corresponds to genetic algorithm corresponds to the risk-managed cost stands for voltage regulator

Nomenclature Symbols relevant to the distributed optimization model of the energy community: t Pbuy is the power bought from the external grid by the prosumer i in time interval t Grid i t PSell is the power sold to the external grid by the prosumer i in time interval t Grid i πtbuy are the prices of energy bought from the external utility grid πtsell are the prices of energy sold to the external utility grid Δt time step of the daily optimization horizon t Pbuy power bought by prosumer i from prosumer j in time interval t i;j t Psell power sold by prosumer i to prosumer j in time interval t i;j λti are the Lagrangian multipliers for prosumer i in time interval t t PPV profile of the photovoltaic power generation for prosumer i i t PLoad profile of the power demand for prosumer i i t represents an estimation of the losses in the branch b of the internal network Lb;i due to the transactions involving prosumer i uti is a binary variable used to avoid simultaneous purchases and sales by prosumer i max Psell limits the power sold by prosumer i in time interval t i max Pbuy limits the power bought by prosumer i in time interval t i t Pdis discharging power of the battery of prosumer i in time interval t i t Pch charging power of the battery of prosumer i in time interval t i t EBES is the energy stored in the battery of prosumer i at time t i max EBES is the battery capacity of prosumer i i ηch i and ηdis i the efficiencies during charge and discharge of the battery, respectively max PBES maximum power for the charging and discharging processes of the battery i min EBES minimum level of state of charge allowed for the batteries during the day i utBES i binary variable used to avoid the simultaneous charging and discharging of the batteries. Rb is the resistance of branch b in the internal network of the energy community Vn is the line-to-line rated voltage value t Fb;i is the three-phase power flow in branch b, only due to the transaction that involves prosumer i AGrid and A are a 2-D matrix and a 3-D array, respectively, that describe the position of each branch b with respect to the buses where the prosumers are connected t t Lbuy , L , are the losses in branch b attributed to the power bought by prosumer i from Grid b;i sell Grid b;i t and Lbuy the utility grid, to the power sold by prosumer i to the utility grid, and to the b;j;i power sold by prosumer i to j, respectively

Impact of neighborhood energy trading and renewable energy communities

ηtbuy Grid i , ηtsell and ηtbuy i;j

are efficiency parameters assigned to each transaction between prosumer i and the grid (i.e., when i buys from the external grid or sells to the utility grid) or between prosumer i and j (i.e., prosumer j sold to i), respectively Parameters used for the implementation of the ADMM algorithm: ρ is the penalization parameter m is a scale factor for the penalized term rit are the values of the primal residual sνi is a vector with the dual residual elements v current iteration of the ADMM algorithm t are the values in the previous iteration of power bought and sold by prosumer i P^ buy Grid i and t from and to the utility grid, respectively P^ sell Grid i ^P tbuy i;j and P^ tsell i;j are the values in the previous iteration of power exchange between prosumers i and j (i.e., bought and sold respectively) ε is the tolerance defined to achieve the convergence of the ADMM algorithm Symbols and acronyms relevant to the distribution network planning model: P ðAÞ is the probability of event A FD is the cumulative density function of D pi is the probability of D lying within the i-th load level ava HDR is the available duration of the demand response kVAava is the available kVA of demand response DR μHDR denotes the mean of the available duration of the demand response ava ava σHDR corresponds to the standard deviation of the available duration of the demand response μkVAava is the mean of the available kVA of demand response DR σkVAava is the standard deviation of the available kVA of demand response DR min max HDR and HDR represent the minimum and maximum duration of demand response, respectively max kVAmin represent the minimum and maximum kVA of demand response, respectively DR and kVADR kVAy;i corresponds to the required kVA of NNSs’ contribution in planning stage y at req i-th load level at a specific node y Ceu is the cost function of end-users y INCeu is the investment cost incurred by end-users y OPCeu is the operating cost incurred by end-users y; j INCeu; PV is the investment cost of new PV units installed by j-th end-user in planning stage y y; j INCeu; is the investment cost of ESS units installed by j-th end-user in planning stage y ESS y; j OPCeu is the operating cost incurred by j-th end-user in planning stage y prijy , plijy1 , and plijy2 are the purchased energy from the utility, purchased energy in the NET, and sold energy in the NET by j-th end-user in planning stage y, respectively Riy denotes the retail price in planning stage y at i-th load level nciy corresponds to the network charge in planning stage y at i-th level of load of end-user DCjy is the degradation cost of the j-th end-user’s ESS unit in planning stage y y OPCNET is the NPV of operating cost incurred by end-users and the utility associated with the NET in planning stage y y CNET is total cost incurred by end-users and the utility associated with the NET in planning stage y y;i CNNS; is the operating cost of k-th type of NNS at i-th level of load in planning stage y k kWhy;i k kVAy;i k

Grid i ,

is the maximum continuous kWh that is delivered by k-th type of NNS at i-th level of load in planning stage y is the peak of power that is delivered by k-th type of NNS at i-th level of load in planning stage y

127

128

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning y;i CNNS y CProb y CProb;NNS y CRMC y CProb;NS y Crel H viy Iiy n y;i CNNS;DR y;i CNNS;ESS=DG

m y CProb;NS y CNS;fix y CNS;var y CO&M y CO&M ;NS y Csalvage y y ELoss and PLoss y y CELoss and CPLoss y SAIDI and SAIFI y y y CSAIDI and CSAIFI Δvi ΔIi CΔvi CΔIi

corresponds to the total cost of procuring NNSs given the required kWhy;i and kVAy;i at i-th level of load in planning stage y corresponds to the probabilistic NPV of the total cost of planning in planning stage y denotes the probabilistic NPV of the cost of NNSs in planning stage y y denotes the risk-managed cost in planning stage y, which is equal to CProb;NNS denotes the probabilistic NPV of the total cost of NSs in planning stage y corresponds to the reliability cost in planning stage y is the horizon year of the network planning is the voltage magnitude at i-th node in planning stage y is the current magnitude of i-th branch in planning stage y is the number of nodes denotes the costing parameter of the DR scheme for NNS at i-th level of load in planning stage y denotes the costing parameter of a ESS/DG scheme for NNS at i-th level of load in planning stage y is the number of load levels is the probabilistic cost of NSs is the NPV of the fixed investment cost of NSs in planning stage y is the NPVs of the variable investment cost of NSs in planning stage y is the NPV of operation and maintenance (O&M) cost in planning stage y is the fixed O&M cost of the NSs in planning stage y y is the salvage value of CNS;var are total expected energy loss and power loss in planning stage y are the unit cost of energy loss and power loss in planning stage y, respectively denote the expected SAIDI and SAIFI in planning stage y, respectively denote the VCR for SAIDI per customer-minute and SAIFI per failure-customer in planning stage y is the voltage magnitude violation at i-th bus is the current magnitude violations at i-th branch is the cost function associated with Δvi is the cost function associated with ΔIi

5.1 Introduction With the current energy transition pushed by social, environmental, and economical factors, the electrical system is moving towards a distributed scheme based on renewable resources, in which the customers play a new active role through self-generation, that is acting as prosumers. This chapter deals with the scenario in which direct energy transactions between prosumers located within a so-called local energy community (LEC) are allowed in addition to the energy transactions with the external energy provider. In this framework, this chapter is aimed to study the impact of such an energy trading scheme on the operation and planning of distribution networks. The participants in the energy community can be residential, small commercial or industrial sites connected to the same distribution network. Each participant can, in general,

Impact of neighborhood energy trading and renewable energy communities

consume or produce electricity in different time periods, that is can be considered as a prosumer. In general, each prosumer may be equipped with local generation units (photovoltaic panels in this chapter) and battery energy storage (BES) units, which help supplying its loads. The regulatory challenges and opportunities for such entities are analyzed in for example [1], which also refers to the recent legal framework called “Clean Energy for all Europeans” approved by the European Union (see e.g., [2,3]). The economic justification for the LEC is mainly due to the difference between the prices of the energy supplied by the external energy provider and the energy sold by the LEC to the utility grid. This difference can be significant for various reasons, for example due to the costs of the ancillary services. In the literature, there are several studies regarding real implementations of the LEC concept: one of the most popular is the Brooklyn microgrid project [4]. The chapter is divided in two main parts: the first part is devoted to the optimal scheduling of dispatchable energy resources available inside the energy communities, the second is devoted to the presentation of a planning scheme for utilities that explicitly consider the implementation of neighboring trading schemes between end-users. The structure of the chapter is the following. Section 5.2 reviews the main characteristics of a distributed approach for the day-ahead scheduling of the LEC and the implemented distributed optimization problem. Section 5.3 presents some results for different test cases of such an operation problem. Section 5.4 is devoted to the presentation of the distribution network planning model considering neighborhood energy trading. Section 5.5 presents the application of the planning tool to case of test systems. Section 5.6 concludes the chapter.

5.2 A distributed approach for the day-ahead scheduling of the LEC The community concept implies the implementation of an energy management system (EMS) to achieve the prosumers’ common goals and the optimal operation of the installed energy resources [5]. According to several approaches presented in the literature (e.g., [69], and references therein), a day-ahead scheduling procedure is convenient to minimize the energy procurement costs of the LEC. In [10,11], a distributed approach based on the alternating direction method of multipliers (ADMM) has been presented that guarantees each prosumer has an advantage in the participation in the LEC with respect to the case in which it can exchange energy only with the external provider. We focus here on the EMS function that provides the day-ahead scheduling of the BES units, under the assumption that all the generation units of the LEC are photovoltaic (PV) systems. The LEC definition considered in this chapter is characterized by being local and cooperative: all the prosumers are connected to the same low voltage (LV) distribution network and they collaborate without any competitive strategy for the common goal of minimizing the costs due to the exchanges with the utility grid. These characteristics makes the studied

129

130

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

framework quite different from those dealt with in, for example [6,12], and references therein, which consider competitive markets where the various operators act with different objectives in a noncooperative scheme. The community concept can be more general, particularly if there is the possibility for a prosumer to join another local trading platform, or if the prosumer energy comes only from renewables and it converted in different energy carriers, namely electricity, thermal energy and cooling, a possibility that is not considered in this chapter. The scheduling function can be structured for instance as a centralized optimization problem, in which a control unit collects and keeps updated all the characteristics of the prosumers’ equipment and all the load and PV production forecasts. Compared to a centralized approach, the use of a distributed approach, like the ADMM, limits the information that every prosumer needs to communicate. We focus here on the scheduling function structured as a distributed optimization algorithm, which aims at minimizing the energy procurement cost of the LEC considering the power loss in the internal network. The internal network losses are allocated to each energy transaction between two prosumers or between a prosumer and the utility grid. A billing procedure is proposed using the metering systems typically installed in a LEC. The main inputs of the decisions of each prosumer are the forecast of the local photovoltaic production and the load. The prosumers’ decisions are coordinated by the procedure that iteratively updates the multipliers. The procedure for the update of multipliers only requires the knowledge of the energy exchange between prosumers as shown in [11]. Furthermore, a distributed procedure is more appropriate when the implementation of new transaction methods based, for example on blockchain [13,14], or, more generally, on distributed ledger technologies is required. The LEC considered in this chapter has an internal LV distribution network, which is connected to a point of common coupling, through an MV/LV transformer, to the external utility grid. In the considered scenario, each prosumer uses the available energy resources in cooperation with the other prosumers to minimize the energy procurement cost of the entire LEC. The operation of such collective requires the implementation of an EMS for the optimal exploitation of the available resources. Fig. 5.1 illustrates the scheme of the LEC. The grid meter Mg, positioned at the point of common coupling with the external utility grid, is bidirectional and measures the energy exchanged in each time interval. For the implementation of the distributed optimization approach, each prosumer i is equipped with a local bidirectional meter Mi measuring the energy that the specific prosumer exchanges with the internal network in each time interval. Given the collaborative characteristic of the LEC, a prosumer cannot act as producer and as consumer in the same time interval. The day-ahead scheduling provides a plan of the optimal use of the LEC energy resources during the following day and calculates the prices of the energy transactions

Impact of neighborhood energy trading and renewable energy communities

Figure 5.1 Model of the local energy community.

between prosumers. It is assumed that the prices of exchanges with the utility grid are predefined, although in general they vary according to the time of day. The electricity billing procedure can be described as follows: 1. In each time interval, if the LEC buys energy from the utility grid (measured by Mg), the relevant cost is allocated to each consumer i (i.e., to each prosumer who consumes energy in excess of the local generation in that time interval) proportionally to the ratio between its consumption measured by Mi and the total consumption in the LEC, that is the sum of the measured energies by all the prosumers acting as consumers. 2. If the LEC sells energy to the utility grid (measured by Mg), the corresponding revenue is allocated to each producer j (i.e., to each prosumer that produces energy in excess of the local load in that time interval) proportionally to the contribution of j to the total LEC production, that is to the ratio between the energy measured by Mj and the sum of the measurements of all the prosumers acting as producers. 3. Each consumer i is also charged for the energy bought from the producers of the LEC, that is for the difference between the measurement of Mi and the energy

131

132

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

allocated to consumer i in step 1. The corresponding revenue of producer j is estimated proportionally to the contribution of j to the total LEC production as in step 2. The day-ahead scheduling procedure calculates the energy prices of each producer j in each time interval.

5.2.1 Distributed optimization model formulation The set of participants in the LEC is denoted as Ω 5 {1, 2, . . . , N}, T 5 {1, 2, . . . , tend} corresponds to the set of intervals t in the time horizon (a day), and B 5 {1, 2, . . ., bend} denotes the set of branches of the internal network of the community. The objective function is the minimization of the energy procurement cost of the LEC (5.1) given by the cost associated to the exchanges of electricity with the external energy provider during the day. The profiles of the prices of energy bought and sold from and to the external utility grid (in h/kWh), that is πtbuy and πtsell , respectively, are assumed deterministic for the next day. X  t t t OF 5 πtbuy Pbuy ð5:1Þ Grid i 2 πsell Psell Grid i Δt tAT iAΩ t where Pbuy Grid i is the power bought from the external grid by the prosumer i in time t interval t, and Psell Grid i is the power sold to the external grid by the prosumer i in time t t interval t. The quantities Pbuy Grid i Δt and Psell Grid i Δt are the energy bought from and sold to the utility grid, respectively, in a time step Δt. For the numerical tests included in this chapter the powers are expressed in kW and Δt is equal to 0.25 h. In the ADMM algorithm, OF is decomposed in local subproblems, one for each prosumer i, by means of the Lagrangian decomposition. The objective function of each subproblem is 2 t 3 t t t πbuy Pbuy Grid i Δt 2 πsell Psell Grid i Δt 1 X6 X X 7 t t t t t7 6 OFi 5 min λ P Δt 2 λ P Δt 1 ‘ ð5:2Þ sell i;j i5 j buy i; j i 4 t t Pbuy jAΩ jAΩ Grid i ; Psell Grid i tAT j6¼i

j6¼i

t t Pbuy i;j ; Psell i;j

where 3 2X  t 2 X t 2 t t ^ ^ P buy j;i 2Psell i;j 1 Pbuy i;j 2 P sell j;i 5 ‘t 5 m  ρ  4 i

jAΩ j6¼i

jAΩ j6¼i

ð5:3Þ

Eq. (5.2) is given by the summation of three terms: (1) costs and revenues associated to exchanges of energy with the external utility grid, considering the respective prices; (2) cost and revenues for the exchanges of i with the other prosumers, where

Impact of neighborhood energy trading and renewable energy communities

λti and λtj are the Lagrangian multipliers of the equilibrium between power sold and bought in each internal transaction; (3) finally, the squared norm of the imbalance of each energy transaction between prosumer i and every other prosumer j. The constraints of the implemented model are the following: X t t t PPV 1 P 1 P 1 i dis i buy Grid i t t t t ð5:4Þ jAΩ Pbuy i;j 5 PLoad i 1 Pch i 1 Psell Grid i j6¼i X 1 1X t t L tAT ; iAΩ jAΩ Psell i;j 1 2 bAB b;i j6¼i (

t Pbuy

Grid i

t t 5 0 and Pbuy i;j 5 0 if ui 5 0

uti Af1; 0g

t Psell

Grid i

t t 5 0 and Psell i;j 5 0 if ui 5 1

i and jAΩ

t 0 # Pbuy

Grid i

max t # Pbuy i 0 # Psell

Grid i

max # Psell i tAT ; iAΩ

t max t max 0 # Pbuy i;j # Pbuy i 0 # Psell i;j # Psell i tAT ;

 t  t t21 t EBES i 5 EBES i 1 Pch i ηch i 2 Pdis i =ηdis i Δt 

t Pch i 5 0 if t Pdis i 5 0 if

t max 0 # Pdis i # PBES i

ð5:7Þ

iAΩ tAT ; t . 1

ð5:8Þ

iAΩ

utBES i 5 0 utBES i Af1; 0g utBES i 5 1 iAΩ t max 0 # Pch i # PBES i

min t max EBES i # EBES i # EBES i

ð5:6Þ

i and jAΩ

 t51  t51 max t51 EBES i 5 EBES i 1 Pch i ηch i 2 Pdis i =ηdis i Δt tend max EBES i 5 EBES i iAΩ 

ð5:5Þ

tAT ; iAΩ

tAT ; iAΩ

ð5:9Þ

ð5:10Þ ð5:11Þ ð5:12Þ

Constraint (5.4) represents the power balance for the i-th prosumer in time interval t: t t where parameters PPV i and PLoad i are the forecasts of PV generation and load demand t t (in kW), respectively; nonnegative variables Pch i and Pdis i are the charging and discharging t power of the BES (in kW), respectively; Psell i;j is the power sold by prosumer i to prosumer t t j, and Pbuy i;j is the power bought by prosumer i from prosumer j; Lb;i represents an

133

134

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

estimation of the losses in branch b due to the transactions involving prosumer i. Since each transaction is between two prosumers, only half of the power loss is attributed to each prosumer. The omission of the concurrent presence of the transactions of all the prosumers is an approximation justified by the lack of counter-flows due to the assumed noncompetitive behavior of the prosumers in the LEC. t Lb;i in (5.4) is defined by the following constraints t Lb;i 5

Rb  t 2 F 3Vn2 b;i

tAT ;

bAB; iAΩ

t t t 5 AGrid b;i Pbuy AGrid b;i Psell Fb;i Grid i 2X Grid i X t t 1 Ab;i;j Pbuy 2 A P b;i;j sell i;j i;j jAΩ

tAT ; bAB; iAΩ

ð5:13Þ

ð5:14Þ

jAΩ

In (5.13), Rb is the resistance of branch b, Vn is the line-to-line rated voltage value, t and Fb;i is the three-phase power flow in branch b, due to the transaction that involves i, considered to be positive when directed from the substation to the end of the feeder. Constraint (5.13) assumes rms bus voltage values equal to the rated value, a balanced LV network, and neglects reactive power flows. In (5.14), the position of each branch with respect to the buses where the prosumers are connected are described by 2-D matrix AGrid and 3-D array A, assuming a radial configuration: • AGrid b,i is the b,i element of matrix AGrid. It is equal to 0 if branch b cannot be crossed by a power flow due to the transaction between prosumer i and the external network, whilst it is 1 otherwise. • Ab.i,j is the b,i,j element of array A. It is equal to 1 if branch b is crossed in the assumed positive direction by the power flow created when i buys from j, it is 2 1 if branch b is crossed in the negative direction by the flow created when i buys from j, and it is 0 if branch b is not crossed by the flow created when i buys from j. Indicator constraints (5.5), with binary variable uti , are used to avoid simultaneous purchases and sales by the same prosumer. In each time interval t, purchases and sales are limited by constraints (5.6) and max t t max max (5.7), where Psell i is the largest value between 0 and PPV i 2 PLoad i 1 PBES i ; Pbuy i is the t t max max largest value between 0 and PLoad i 2 PPV i 1 PBES i ; PBES i is the maximum power output of the BES of prosumer i. The state of energy (SoE) of the i-th BES unit is defined by (5.8) and (5.9), which t represent a simple energy balance model, where EBES i is the SoE at time t (in kWh), max EBES i is the capacity, ηch i and ηdis i are positive numbers lower than 1 that represent the efficiencies during charge and discharge, respectively. In (5.9) we assume that BES units are fully charged at the beginning and at the end of the day. Indicator constraints

Impact of neighborhood energy trading and renewable energy communities

(5.10), with binary variable utBES i , prevent simultaneous charging and discharging of the batteries. Constraint (5.11) limits the discharging and charging power within the max min maximum value PBES i . The SoE is bounded between minimum level EBES i and maximax mum EBES i one by constraint (5.12). At the beginning of the procedure, Lagrange multipliers λti , penalization parameter ρ, and scale factor m are initialized. Then, at each iteration ν, local subproblem (5.2) is solved by each one of the prosumers i considering set of constraints (5.45.14). t t The prosumers communicate to each other Pbuy i;j and Psell i;j obtained at the end of their own optimization problem. Then, each prosumer i updates Lagrangian multipliers λti (i.e., the prices associated to the internal energy exchanges in the LEC) based on the imbalance between their local variables and the values commut t nicated by the others prosumers, denoted by a hat in (5.3): P^ buy Grid i and P^ sell Grid i are the values in the previous iteration of power bought and sold by prosumer t t i from and to the utility grid, respectively; P^ buy i;j and P^ sell i;j are the values in the previous iteration of power exchange between prosumers i and j (i.e., bought and sold respectively). The imbalances are equal to primal residual rit . Furthermore, the convergence of the ADMM procedure is improved by adding the following constraints, starting from the second iteration, as they provide a coordination between the sales and purchase decisions of prosumer i with respect to those of the other prosumers: X t t ^t Psell ð5:15Þ P^ buy j;k tAT ; j and kAΩ i;j # P buy Grid j 1 iAΩ k6¼i t ^t Pbuy i;j # P sell

Grid j 1

X

t P^ sell j;k

tAT ; j and kAΩ

ð5:16Þ

iAΩ k6¼i

In order to accelerate the convergence of the ADMM procedure, the value of penalization parameter ρ and scale factor m are adjusted at each iteration according to Fig. 5.2 that shows the iterative procedure that implements the ADMM procedure, where : :2 is the Euclidian norm and sνk is the |T|-dimensional vector of the dual residual elements. The value of parameter m needs to be adjusted to accelerate the convergence. For the test cases considered in this chapter, the initial value of m, equal to 1  1025, is multiplied by 5 and by 1.5 when the maximum value of the total mismatch P r t 5 k jrkt j becomes less than 20 kW and 1 kW, respectively, and further multiplied by 5 when maxðjrkt jÞ , 100 W.

135

136

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 5.2 Implemented ADMM algorithm with parameter update.

Once the procedure converges, ‘ti tends to zero and the value OF for the community is equal to the summations of the objectives of the prosumers: X OF 5 OFi ð5:17Þ iAΩ

As described in [15], the convergence of ADMM to a global optimal point is not guaranteed when it is applied to nonconvex problems. However, it will possibly have better convergence than other local optimization methods and it has been successfully applied to large-scale mixed integer problems as shown in, for example [7]. In the models considered in this chapter, the binary variables are used only in indicator constraints (5.5) and (5.10). These constraints do not affect the optimal value of the OF, but are useful for finding the solution among those with the optimal value of OF which can be more easily applied, that is which avoids the occurrence of prosumers who buy and sell energy without benefit (i.e., at the same price). Furthermore, these constraints together with (5.15) and (5.16) make the ADMM convergence significantly faster.

Impact of neighborhood energy trading and renewable energy communities

The procedure is iteratively repeated until the absolute values of all residuals rit are less than a small tolerance ε, which is assumed to be equal to 10 W in the numerical tests of this chapter. In [10] an additional procedure that improves the estimation of the power loss in the internal network has been proposed. For this purpose, efficiencies parameters of each transaction (between prosumer i and the grid or between prosumers i and j) have been included in the energy balance of i, as follows: X t t t t t PPV Pbuy i;j i 1 Pdis i 1 ηbuy Grid i Pbuy Grid i 1 ð5:18Þ jAΩ j6¼i

t t 5 PLoad i 1 Pch i 1

t Psell Grid i ηtsell Grid i

1

t X Psell i;j jAΩ j6¼i

tAT ; iAΩ

ηtbuy j;i

where ηtbuy Grid i , ηtsell Grid i , and ηtbuy i;j are efficiency parameters assigned to each transaction between prosumer i and the grid (i.e., when i buys from the external grid or sells to the utility grid) or between prosumer i and j (i.e., prosumer j sold to i), respectively. These efficiencies are given by: P t bAB Lbuy Grid b;i t ηbuy Grid i 5 1 2 tAT ð5:19Þ t Pbuy Grid i ηtsell

ηtbuy i;j

Grid i 5

5

t Psell

t Pbuy i;j 1

t Psell PGrid i t 1 Grid i bAB Lsell t Pbuy P i;j bAB

t Lbuy b;i;j

tAT

ð5:20Þ

Grid b;i

tAT ; jAΩ; j 6¼ i

ð5:21Þ

t t t where Lbuy Grid b;i , Lsell Grid b;i , and Lbuy b;j;i are the losses in branch b attributed to the power bought by prosumer i from the utility grid, to the power sold by prosumer i to the utility grid, and to the power sold by prosumer i to prosumer j, respectively. The corresponding losses are calculated at the end of an distributed optimization defined by the formulation (5.25.16). The η calculation can be carried out by a central coordinator that knows the topology and parameters of the network and collects the power t t t exchanges of the first stage (i.e., the values Pbuy i;j , Pbuy Grid i , Psell Grid i from each prosumer i). This calculation can be also done by each prosumer in a distributed manner (assuming that each prosumer knows resistance values of the branches and matrices AGrid b;i , Ab;i;j ) provided that each prosumer i communicates its power exchanges to every other prosumer of the LEC. In such approach, the optimization is then repeated considering constraint (5.18).

137

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

5.3 Implementation and numerical tests The model of the LEC has been implemented in the AIMMS Developer modeling environment [16] and tested by using the Cplex V12.8 solver [17] on a 2-GHz Intel-i7 computer with 8 GB of RAM, running 64-bit Windows 10. The MIQP (mixed integer quadratic programming) solver is used for the ADMM model. All the calculations refer to a time window of one day, divided into 96 periods of 15 minutes each. The test system is composed of two LV feeders. Each feeder consists of five lines, each with resistance Rb 5 189 mΩ. Five prosumers are connected to each feeder: prosumers 15 to a feeder and prosumers 610 to the other as shown in Fig. 5.1. Each prosumer may be equipped with a PV system and a load. The total daily consumption of the LEC is 313 kWh and the corresponding PV production is 231 kWh (73.8% of the load). The load profiles adopted for each prosumer are shown in Fig. 5.3. For all the PV units we assumed the same profile of the ratio between power output and panel surface shown in Fig. 5.4. The area of the PV panel of each prosumer is given in Table 5.1. Fig. 5.4 also shows the price profile of the energy bought from the utility grid πtbuy . We assume that the price of the energy sold by the LEC to the utility grid max (i.e., πtsell ) is half of πtbuy . Sizes EBES of the BES units are shown in Table 5.2 and the max max corresponding PBES values are assumed to be equal to the ratio EBES =Δt. The total capacity of the BES units is 30 kWh (13% of the daily PV production). The total power flow at the connection of the LEC with the utility grid and the power profiles from each prosumer when it exchanges energy with the others that is 35 load 1

30

load 2 load 3

25 Power profile (kW)

138

load 4 load 5

20

load 6 load 7

15

load 8 load 9

10

load 10

5 0 0

6

Figure 5.3 Load profile for each prosumer.

12 Time (hour)

18

24

Impact of neighborhood energy trading and renewable energy communities

0.4

0.4

0.3

0.3

Energy price

0.2

0.2

0.1

0.1

0

Grid price (€/kWh)

PV power profile ( kW/m2)

PV power profile

0 0

6

12 Time (hour)

18

24

Figure 5.4 Profile of the PV production and grid purchase price. Table 5.1 PV panel surface for each prosumer.

Prosumer Area (m2)

1 32

2 14

3 21

4 32

5 28

6 14

7 42

8 32

9 14

10 42

3 4

4 2

5 3

6 1

7 2

8 2

9 2

10 6

Table 5.2 Sizes of the BES units.

Prosumer Size (kWh)

1 5

2 3

sells and buys, are presented in Fig. 5.5, 5.6, and 5.7 respectively. The total OF value of (5.17) obtained by the ADMM procedure is h18.12. Fig. 5.8 shows the detail of the SoE profiles of each BES unit, whilst Fig. 5.9 provides the profiles of the total energy contained in the BES units of the LEC. Fig. 5.10 shows the energy prices λti of each prosumer i, the dotted lines correspond to the prices of the energy bought from and sold to the utility grid (i.e., πtbuy and πtsell ), while the black dots represent the transaction prices of the various prosumers when they sell energy to any other prosumer of the LEC. The comparison between Fig. 5.10 and 5.5 shows that the prices of the internal transactions are not equal to πtbuy and πtsell only during the time interval (starting just after 6 am) when there is no power exchange with the utility grid. Table 5.3 compares the individual energy procurement costs of each prosumers, taking into account both the exchanges with the external grid and the internal exchanges and the prices of Fig. 5.10. Additionally, the table shows the corresponding

139

Figure 5.5 Power flow exchanged with the utility grid (positive if consumed by the LEC).

Figure 5.6 Power flows from every prosumer when it sells to the others (excluding the utility grid).

Impact of neighborhood energy trading and renewable energy communities

Figure 5.7 Power flows from every prosumer when it buys from the others (excluding the utility grid).

values obtained by preventing the transactions between prosumers. The total energy procurement cost of the LEC is around 16% less than the corresponding cost without internal transaction among the prosumers.

5.3.1 Scalability of the distributed approach In order to perform an analysis of the scalability of the implemented distributed approach, three different configurations have been considered: • Scenario 1: 2 feeders with 5 prosumers each. The characteristics of the 10 prosumers correspond to the model considered in the previous numerical tests and illustrated in the Fig. 5.1; • Scenario 2: 1 feeder with 10 prosumers with the same characteristics of those considered in Scenario 1 and illustrated in the Fig. 5.11;

141

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 5.8 Battery SoE for each prosumer.

30 Total energy in the BES units (kWh)

142

20

10

total energy stored 0 0

6

12 Time (hour)

18

24

Figure 5.9 Total energy in the batteries of the LEC obtained by the ADMM approach.

Impact of neighborhood energy trading and renewable energy communities

0.4

Energy price (€/kWh)

prosumer

π_buy

π_sell

0.3

0.2

0.1

0 0

6

12

18

24

Time (hour)

Figure 5.10 Energy prices of selling prosumers. Table 5.3 Energy procurement cost in h (negative values indicate revenues) for each prosumer. Prosumer

1

2

3

4

5

6

7

8

9

10

ADMM Without internal exchanges

5.24 0.08 0.96 20.98 20.65 20.21 14.80 1.63 20.47 22.29 5.46 0.28 1.10 20.82 20.46 20.15 16.38 1.71 20.30 21.67

• Scenario 3: 2 feeders with 10 prosumers each as shown in Fig. 5.11. The 10 prosumers connected to the first feeder are the same considered in the Scenario 1 and Scenario 2. The additional 10 prosumers of the second feeder are characterized by the load profiles show in Fig. 5.12, by the PV panel surfaces shown in Table 5.4, and the sizes of the BES units shown in Table 5.5. For all the configurations, the profile of the ratio between power output and panel surface, and the price profile of the energy bought from the utility grid are the one shown in Fig. 5.4. Transactions are allowed also between prosumers connected to different feeders. For each scenario, the corresponding distributed model of the LEC has been solved based on the application of the ADMM procedure. Table 5.6 compares the computational effort (number of iterations and CPU time) for each scenario, where the optimization problems of the prosumers are solved in sequence, without considering delays or limitations in the communication channels. As expected, the computational effort decreases if a longer Δt is adopted. For example, if Δt 5 30 min, the distributed model of scenario 1 requires 125 s. If Δt 5 1 h, the solution of the same model requires 70 s.

143

Figure 5.11 Additional configurations considered in the analysis: (A) 1 feeder with 10 prosumers; (B) 2 feeders with 10 prosumers each. Each prosumer is equipped with PV generation, local, and battery storage system. 40 load 11 35

load 12 load 13

Power profile (kW)

30

load 14 load 15

25

load 16 20

load 17 load 18

15

load 19 load 20

10 5 0 0

6

12 Time (hour)

18

24

Figure 5.12 Load profiles of the 10 prosumers connected to the second feeder of Scenario 3.

Impact of neighborhood energy trading and renewable energy communities

Table 5.4 PV panel surface for each prosumer of the second feeder of Scenario 3.

Prosumer Area (m2)

11 36

12 20

13 22

14 25

15 34

16 35

17 15

18 13

19 22

20 33

17 2

18 1

19 3

20 4

Table 5.5 Sizes of the BES units of the second feeder of Scenario 3.

Prosumer Size (kWh)

11 4

12 3

13 3

14 3

15 5

16 6

Table 5.6 Comparison of the computational effort for the three scenarios considered in the scalability analysis. Scenario

Total prosumers

Iterations

Solution time (s)

1 2 3

10 10 20

46 21 20

300 900 3000

To illustrate the convergence behavior of the ADMM procedure, Fig. 5.13 shows the augmented OF according to (5.2), the OF value of (5.17), and corresponding average value of the primal residuals rkt (denoted by R) at each iteration, for the considered scenarios.

5.3.2 Scenario considering uncertainties on the energy generation and consumption In order to consider the uncertainty associated with both the generation from renewable sources and the energy consumption [18], introduces an extension of the ADMM that adapts the distributed procedure to a multistage approach as the one studied by [19]. Stochastic optimization approaches are widely adopted to solve this kind of problems (e.g., [2022]). In [19], the day-ahead scheduling of a single site is adapted to the actual intra-day operating conditions and the optimization problem is formulated as a multistage decision problem modeled by a scenario tree. We consider the LEC illustrated in Fig. 5.14 in which the characteristics of the prosumers are the same of those of the feeder 1 in Scenario 1. In order to better adapt the day-ahead solution to the actual intra-day operating conditions, the optimization problem is formulated as a multistage decision problem in which the battery output set points are determined at the beginning of the day and subsequently readjusted at the middle of the day.

145

200 Augmented OF

OF

R 160

35

120

80 20 40

5

0 1

6

11

16

21

(A)

26

31

36

41

46

Iterations

200

Objective function value (€)

Augmented OF

OF

R

160

35 120

80 20 40

0

5 1

(B)

Avearage value of residuals (W)

50

5

9

13

17

21

Iterations

Objective function value (€)

Augmented OF

OF

R 160

70

120

60 80 50

40

40

Avearage value of residuals (W)

200

80

0 1

(C)

Mean value of residuals (W)

Objective function value (€)

50

4

7

10 13 Iterations

16

19

Figure 5.13 ADMM convergence—augmented OF, OF value corresponding to the exchanges with the utility grid, average value R of primal residuals at each iteration for: (A) Scenario 1, (B) Scenario 2, and (C) Scenario 3.

Impact of neighborhood energy trading and renewable energy communities

Figure 5.14 Scenario with one feeder and five prosumers.

The specific characteristics of this method are the following: • it considers the uncertainties of the forecasts of load and PV generation, represented by a scenario tree model that combines the different scenarios of the various prosumers; • it includes a routine that merges the scenario tree of each prosumer in a common tree for the operation of the entire LEC; • it includes a decision-making procedure that updates the BES units scheduling and the LEC internal transactions according to the intra-day operating conditions; First, the procedure generates the scenario tree for each prosumer i. Then, the routine implemented merge the scenario tree of each prosumer in a common scenario tree for the operation of the entire LEC. Finally, the obtained scenario tree is used both for the day-ahead scheduling and for the intra-day decision-making procedure. For a set of 45 generated scenarios (i.e., operating conditions during the intra-day), Fig. 5.15 shows the comparison between the OF values calculated by using the intraday decision-making procedure (multistage solution) and those given by the day-ahead scheduling that takes into account only the forecast profiles (forecast-based). The multistage scheduling provides better results with respect to a forecast-based solution. Fig. 5.15 also shows the OF values of the deterministic solutions, in which the profiles employed by the ADMM correspond to the current PV generation and loads during the intra-day. Table 5.7 reports the percentage increase in the OF value with respect to the deterministic solution of the combinations in Fig. 5.15, by applying the multistage and forecast-based solution, respectively. The average increase and the higher deviation (i.e., the maximum value) confirm the advantage of the implemented multistage day-ahead scheduling over the forecastbased solution.

147

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

deterministic solution

multistage solution

forecast-based

6

Objective function (€)

148

5

4

3

2 1

5

9

13

17

21

25

29

33

37

41

45

45 new scenarios

Figure 5.15 Comparison between the values of the objective function for 45 new combinations. Table 5.7 Percentage increase in the OF with respect to the deterministic solution. Comparison

Average (%)

Min. difference (%)

Max. difference (%)

Deterministic—multistage Deterministic—forecast-based

6 9

2 2

20 28

5.4 Distribution network planning model considering nonnetwork solutions and neighborhood energy trading The current practice in the utilities in tackling the increasing adoption of renewable resources is to increase the capacity of the network which has often led to overinvestments. There is a need for new planning tools that enable network planners to exploit the risk associated with uncertainties when making decisions on network reinforcement. Such a planning tool will allow the planners to achieve the most efficient investment plans for the reliable operation of the grid at the minimum cost. Traditionally, planners make decisions about how much the network should be upgraded to meet the demand using only the network solutions (NSs), which are the investment decisions of transformers, line cables, voltage regulators, etc. However, nonnetwork solutions (NNSs) have recently gained increasing interests regarding their promising capability in addressing network congestions. The NNSs refer to short-term or permanent options that can be employed to manage the risks, that is demand response (DR), energy storage system (ESS) units, distributed generation (DG), and end-user engagement in neighborhood energy trading (NET) schemes. Finding an affordable and optimal plan by taking both the NSs and NNSs into account is a key

Impact of neighborhood energy trading and renewable energy communities

driver of this study. In this regard, a multistage distribution expansion planning (MSDEP) model is developed to find the optimal network upgrading level using both NSs and NNSs. The major complexities in MSDEP problems in real-sized networks are: • large number of variables, • dynamic nature of problem in case of multistage planning, and • more uncertainties compared to passive distribution networks due to DERs. To handle the large problem size, heuristic approaches have been adopted to overcome the challenges of large problem sizes [2331]. To overcome the dynamic nature of the MSDEP problems, different decomposition techniques have been used to break the original dynamic multistage MSDEP problem up into multiple smaller single-year MSDEP problems which are much easier to solve [24,25,3134]. Various uncertainty handling methods have been suggested for dealing with the uncertainty of load and distributed energy resources (DERs) generation by using the probability distribution functions (PDFs) of the uncertain parameters such as the probabilistic methods which build the PDFs using the available historical probabilistic data [27] or assuming that the PDFs are known from the knowledge of experts [31,3541], or probabilisticpossibilistic approach [42]. In [23], an MSDEP method is presented to find the least cost plan considering the investment, operating, and reliability costs and the alternatives to installing new transformers, cable, and tie lines using a heuristic-based tabu search optimization approach. The application of dispatchable DG units in reducing the cost of grid loss and reliability is exploited in [28] through a modified particle swarm optimization (PSO) based MSDEP. The optimal investment decisions regarding the batteries and voltage control devices are determined in [3] through solving an MSDEP via an efficient forwardbackward approach. This work is extended in [24] to incorporate the network reconfiguration. In [33], the MSDEP considers the investment in DG units and line reinforcement. A recursive forwardbackward algorithm in solving MSDEP is presented in [34]. However, the uncertainty of load and DERs generation is considered in these models. An integrated model of demand side management with NET in a microgrid is proposed in [43] to minimize the total cost in the microgrid, including the cost of purchases from the upstream grid and the end-users’ cost. A recent review of the concept of NET and different approaches to implement that is provided in [44]. There are very few works in the literature that consider the risk treatment [26,35]. The risks in [35] are treated in a short term investment plan that determines what type of investments in a short-term (up to three years) will accommodate the forecasted demand. The “reduced risk” short term plan obtained by performing a scenario-based approach can provide long-term savings. In [26], the net present value (NPV) is calculated using discounted cash flow and the variability of NPV found by Monte Carlo simulations is considered as a measure of risk, and the risks are only dealt with DG units. This approach suffers from large problem size and can only be implemented in

149

150

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

relatively small networks, as no decomposition is deployed and the Monte Carlo simulations for large networks are very time-consuming. Below, a forwardbackward approach is described in which the uncertainties of load demand, renewable generation, and DR are modeled using the concept of probability of exceedance (POE). The model considers energy trading between customers and utility and amongst customers through DR and NNSs to manage the risks. The model determines if it is beneficial for the utility to treat the risks that are related to the uncertainties of load and renewable generation by procuring NNSs with or without NET and defer the investments in the network. Besides that, the model provides a solution to treat the risks at each planning stage.

5.4.1 Concept of risk-managed planning The current practice of distribution network planning is to have a large capacity margin to be able to cope with the uncertainties. This traditional planning approach will result in higher network capacities, as shown by the “¢” line in Fig. 5.16. This conservative planning approach does not provide an efficient solution for future active distribution networks mainly due to the underutilized assets in the grid which are barely needed. The novel risk-managed planning framework proposed in this section aims at finding the optimal level of demand (shown in Fig. 5.16 as “’” line) in each planning stage at which the network needs to be upgraded (named DSNS, stands for demand supplied by NSs). If the demand exceeds this level, that is DSNS, the associated risk is treated by NNSs. This optimal value of DSNS is determined in our model by exploiting the trade-off between savings due to avoided investment in NSs and the cost of procuring temporary NNSs. In this regard, the DSNS is a decision variable in our proposed model whose optimal value will be determined by the solution of optimization. As shown in Fig. 5.16, risk-managed planning requires a lower network capacity

Figure 5.16 Risk-managed planning concept.

Impact of neighborhood energy trading and renewable energy communities

when compared to traditional planning approach, which means the total cost of network expansion is reduced due to deferring the network upgrade. For example, a network capacity of 3MVA, that is 2 3 1MVA and 2 3 0.5MVA distribution transformers, is required to be installed in 2016 following the traditional planning approach. However, the proposed risk-managed approach suggests deferring this 3MVA network augmentation to 2019, 2020, 2022, and 2023. As seen, each 1MVA distribution transformer is installed in 2019 and 2020 and each 0.5MVA distribution transformer is planned to be installed in 2022 and 2023. During this period, 2016 to 2023, any load above the capacity of the network is supplied by NNSs. In this example, NNSs cover about 1 MVA and 1.5 MVA of the load in 2018 and 2022, respectively.

5.4.2 Concept of planning with neighborhood energy trading From the end-user’s perspective, NET provides a mechanism for prosumers to sell their excess energy at a higher rate compared to current feed-in tariffs or buy cheaper electricity from other end-users. However, inefficiency may occur in cases where NET participants do not consider the investment decisions of other prosumers in their own energy-related decisions, leading to an over/underinvestment in their premises. This means that it may be more convenient for some NET participants not to invest in their own PV units or ESSs because they can fulfill their electricity demand by using the excess generation and storage facilities of their neighbors. However, there might be cases in which more investment in PV units or ESSs is profitable because of the high rate of return in the LEC that implements a NET scheme. Fig. 5.17 summarizes the concept of the effects of NET in network planning and justify the advantage of planning for both end-users and grid utility investments: (A) refers to the case of separate investment planning for utility and users that results in replacing new line sections and upgrading the substation transformer; (B) shows separate investment planning with NET which implies relatively lower investment in poles and wires but higher investment in PV units and ESSs; (C) illustrates the expected effects of planning for both the utility and prosumers that results in avoiding investment in poles and wires and increases the investments in PV units and ESSs; and (D) compares the costs relating to each planning approach.

5.4.3 Modeling of the uncertainties In this study, the PV generation is considered as active power disregarding, for simplicity, reactive capabilities [27]. The historical demand data of one year are considered as the mean values of forecasts with a standard deviation (SD) for each time interval, that is half an hour, which is defined as the uncertainty level. The next step is to generate several profiles (e.g., 50) for demand and PV generation at each node using the Gaussian PDF at each time interval and the difference between the demand and PV

151

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

(A)

Green line sections and transformers are upgraded

(B) 8

(M$)

152

4

0

(C)

individual planning without NET individual planning with NET joint planning with NET

utility investment

customer investment

total electrification cost

utility customer total cost of investment investment electrification (D)

Figure 5.17 Illustration of the effect of NET in distribution network planning.

generation profiles is defined as an effective time series load profile. Then, the effective load duration curves (LDCs) are generated by arranging these effective time series load profiles from higher to lower values. Then, load levels are defined to facilitate the modeling of these effective LDCs. Fig. 5.18 shows 50 effective LDCs at a node with an uncertainty level of 5% for both the demand and PV generation forecast divided into 5 load levels. According to [45], the SD of a variable is obtained by σ 5 ε 3 μ/300, where σ, ε, and μ are the SD, the percentage error, and the mean value of that variable, respectively. The effective LDCs for each node can be modeled by a multivariate Gaussian mixture model (GMM) of each load point [27]. However, as the PDF of each load level in an effective LDC is similar to the equivalent Gaussian PDF [46], thus it suffices to model each load level of the effective LDC by a single Gaussian PDF [47]. The reactive power is obtained using power factors that are fixed for high and low demand, namely, 0.88 and 0.82 [48], respectively. The effective LDCs at different nodes are treated using the same time series to maintain the correlation between load points. A comprehensive review of risk assessment techniques can be found in Annex A and B of ANSI Z690.32011 standard [49]. The main categories of these techniques are [50]: rule based such as the structured what-if technique, probabilistic based, and judgment based. The risk assessment technique in this study belongs to the category of probabilistic based assessments which evaluates the risk of uncertainties using some other tools [51]. Generally, the risk related to uncertainties is evaluated as its consequence weighted by its probability as [52]: Risk value 5 Probability 3 Impact. Two common approaches to calculate the probabilities are event tree analysis (ETA) and fault tree analysis (FTA) which, respectively, follow bottom-up and top-down logic [51]. In this section, a similar approach to FTA is utilized to find the cumulative distribution functions (CDF) of

Impact of neighborhood energy trading and renewable energy communities

Figure 5.18 Effective LDCs at a node for the same uncertainty level of 5% for both demand and PV forecast with determined five load levels.

occurrence of uncertain parameters over the planning period for the top node of the network. Besides, the risk associated with equipment failure is included in the proposed model by incorporating the reliability cost (cost of SAIFI and SAIDI) in the objective function. The reliability indices are calculated by an approach like ETA. Moreover, the expected unit cost for treating the consequence of each event is considered as the impact of that event. The probability of exceedance (POE) is retrieved from the CDF of effective load profiles at each node as the model of peak load uncertainty. Fig. 5.19 shows the CDF of effective load profiles over ten years with an average growth of 2% for a node. It should be noted that the term POEx represents the level of effective load that has x% probability of being exceeded by the maximum effective load recorded in any year based on the historical demand data of the current year. Hence, regarding the probability of occurrence of maximum effective load (D) the following holds: P ðD . POEx Þ 5 1 2 FD ðPOEx Þ 5 x%

ð5:22Þ

where P(A) is the probability of event A and FD is the CDF of D. Therefore, the probability of D lying within the i-th load level, namely, pi, is expressed as: pi 5 PðPOExi , D , POExi11 Þ 5 FD ðPOExi Þ 2 FD ðPOExi11 Þ 5 xi11 % 2 xi %

ð5:23Þ

These load levels are examined in the proposed planning framework to obtain a cost-effective solution to treat the risks of overloading due to demand and renewable uncertainty.

153

154

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 5.19 The CDF of effective load over 10 years of planning for a node.

5.4.4 Modeling of nonnetwork solutions Ideally, individual NNSs (such as DR, ESS/DG) are among the operational decision variables in the model, but this will substantially increase the size of the problem in real size networks. Therefore, a compact lumped model is proposed in this section to represent all the NNSs by only one decision variable in the MSDEP model. As previously mentioned, the level of demand that is supplied by the NSs is represented by DSNS. Therefore, the difference between total demand and DSNS is the required power to be serviced by NNSs. The compact lumped model will take this power as input and gives the nominal power and energy of each NNS along with their corresponding operating hours and costs. Since the duration of peak shaving depends on the level of demand to be shaved, the operating schedule of NNSs will depend on the load profile as shown in Fig. 5.20. Both the power (kVA) ratings and energy ratings (kWh) or duration of NNSs are obtained to find a feasible yet cost-effective operating schedule. Some NNSs perform better in supplying a peaky demand shape with high output power for a short period, while others are appropriate for supplying a lower level of power for longer periods. Therefore, the characteristics of NNSs (kVA rating, kWh rating, and cost) and the load shape can considerably impact the output of the proposed model. The compact lumped NNS model has two phases: 1) preprocessing the NNSs response to different levels of peak demand reduction, 2) calculating the kVA and kWh of each NNS. The preprocessing phase at each node has the following steps: firstly, a “peak reduction level (kVA),” for example 5 and 20 kVA as seen in Fig. 5.20 is selected, then the periods of d1 for 5 kVA and d2 and d3 for 20 kVA peak reduction in Fig. 5.20 are determined. Finally, the maximum duration of the period as the “maximum continuous hours” of NNS is assigned.

Impact of neighborhood energy trading and renewable energy communities

Figure 5.20 Different durations associated with different peak load reduction.

The maximum continuous hours of NNS deployment for different feasible peak reduction levels (blue line) are shown in Fig. 5.21 which is retrieved from an effective time series load profile at a node. In the second phase of the lumped model for NNS, the required power and energy ratings of the NNSs will be determined. We assume that procuring DR is cheaper than temporary ESS/DG [53]. Therefore, the strategy in the second phase is to use the DR as much as possible and cover the excess duration (hours) and/or excess demand by temporary ESS/DG. In this section, DR is treated as end-users’ demand deferral which is limited by a maximum demand and maximum hours (red lines in Fig. 5.21). DR uncertainties modeling requires advanced and complicated economic and social analyses [54]. We use a truncated Gaussian distribution to model the uncertainty of DR [29]. Therefore, the available ava kVA ðkVAava DR Þ and duration ðHDR Þ of DR have the following PDFs:   ava min ava max BGaussian μHDR # HDR # HDR HDR HDR ava ; σ H ava ; DR   ð5:24Þ min ava max ava ; kVAava ; σ # kVA # kVA kVA kVA DR BGaussian μkVAava DR DR DR DR DR where xBy denotes that x is distributed as y; μHDR ava and σ H ava denote the mean and SD DR ava of the duration of DR; μkVAava and σ are the mean and SD of kVA of DR; kVA DR DR max min max HDR , HDR , kVAmin , and kVA represent the minimum and maximum of duration DR DR and kVA of DR, respectively. Therefore, given the required NNSs contribution in planning stage y at i-th load level at a specific node (for simplicity, the node index is not written), that is kVAy;i req , the operating schedule of NNSs is calculated based on the formulation in [55].

155

156

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 5.21 Maximum continuous hours of NNS deployment versus peak load shaving.

5.4.5 Modeling of NET The customers can invest on ESS and PV to reduce their total cost of energy by managing purchasing/selling energy within NET. The cost function of end-users in a distribution network with NET scheme is: y y y Ceu 5 INCeu 1 OPCeu X  y;j y;j y INCeu 5 INCeu; PV 1 INCeu; ESS jAN y OPCeu 5

X jAN

y; j OPCeu 5

X jAN

DCjy 1

X m

  Riy prijy 1 plijy1 2 plijy2 nciy



ð5:25Þ

i51

y y y where INCeu , OPCeu , and Ceu are the NPV of investment, operating, and the total y; j y; j cost incurred by end-users in planning stage y, respectively; INCeu; PV and INCeu; ESS denote the NPV of investment cost of new PV and ESS units installed by j-th endy; j user in planning stage y, respectively; OPCeu is the NPV of operating cost incurred y y1 by j-th end-user in planning stage y; prij , plij , and plijy2 are the purchased energy from the utility, purchased energy in the NET, and sold energy in the NET by j-th enduser in planning stage y, respectively; Riy and nciy denote the retail price and network charge at i-th level of load of end-user in planning stage y, respectively; and DCjy is the degradation cost of the ESS unit of by j-th end-user in planning stage y. The NET participants will pay the utility by a network charge for using the network infrastructure which enables them to trade energy. The cost terms in (5.25) is comprised of the cost of energy purchases from the utility, the cost of network charge,

Impact of neighborhood energy trading and renewable energy communities

and the degradation cost of ESSs. Regarding the energy purchases by end-users from the utility and the network charge, the total cost for end-users equals the total revenue for the utility. In this context, the incurred cost by end-users resulting from energy purchases from the utility and the network charge does not have an impact on the total cost borne by the utility and end-users associated with the NET. As a result, total cost incurred by end-users and the utility associated with the NET is modeled as: y y y CNET 5 INC eu 1 OPCNET X y y OPCNET 5 DCj

ð5:26Þ

jAN y y and CNET are the NPV of operating and the total cost incurred by where OPCNET end-users and the utility associated with the NET in planning stage y, respectively.

5.4.6 Costing of NNSs The operating cost of k-th type of NNS at i-th level of load in planning stage y, that y;i is CNNS; k , is expressed in a general form as:  y;i y;i  y;i y;i CNNS;k kWhy;i ð5:27Þ k ; kVAk 5 A1;k 1 A2;k 3 kWhk 1 A3;k 3 kVAk y;i where kWhy;i k and kVAk are the maximum continuous kWh and the peak of power that is delivered by k-th type of NNS at i-th level of load in planning stage y. The three constants in (5.27), that is A1;n , A2;n , and A3;n are defined in Table 5.8. Therefore, the total cost of procuring NNSs given the required kWhy;i and kVAy;i at i-th level of load in planning stage y is calculated as:   y;i  y;i y;i y;i  CNNS kWhy;i ; kVAy;i 5 CNNS;DR kWhDR ; kVADR   ð5:28Þ y;i y;i kWhy;i 1 CNNS;ESS=DG ESS=DG ; kVAESS=DG

5.4.7 Planning problem formulation In the model, NNSs and NSs along with the NET are employed to service the projected demand at the lowest cost considering the uncertainties of load forecast and renewable Table 5.8 The definition of costing parameters for NNSs. Type of NNS

A1

A2 ($/kWh)

A3 ($/kW)

Temporary ESS/DG

preparation cost

DR

availability cost [56]

battery/delivered energy cost deferred energy cost [57]

inverter/DG purchased cost deferred demand cost [57,58]

157

158

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

generation subjected to the physical constraints of the distribution network. The reliability cost is also included in the objective function. Therefore, the risk-managed planning problem is modeled as: min

H X

y

CProb 5

y51

subject to

H X

y

CProb;NNS 1

y51

H X

y

CProb;NS 1

y51

H X y51

y

Crel 1

H X

ð5:29Þ

y

CNET

y51

 y    v  # 1:05 $ 0:95 P 0:95 # i  P Iiy  # 1:1 $ 0:95 ; i. . . n; y 5 1. . .H

ð5:30Þ

y y y y y where CProb , CProb;NNS , CProb;NS , Crel , and CNET denote the probabilistic NPV of the total cost of planning, the cost of NNSs, the total cost of NSs, reliability cost, and the total cost incurred by the utility and end-users associated with NET in planning stage y, respectively; H is the horizon year of the network planning; viy and Iiy are the voltage magnitude at i-th node and the current magnitude of i-th branch in planning stage y, respectively; n is the number of nodes, and P(x) is the probability of event x. As the load and renewable uncertainties are mainly handled by NNSs, the expected value of NPV of NNSs at each planning stage is defined as the risk-managed y cost CRMC and is given by: H H m X X X y y y;i CProb;NNS 5 CRMC 5 CNNS ðkWhy;i ; kVAy;i Þ 3 pi y51

5

y51 H X m X

i51

ð5:31Þ

y;i CNNS ðkWhy;i ; kVAy;i Þ 3 ðxi11 % 2 xi %Þ

y51 i51

where m is the number of load levels. The different load levels are usually provided by utilities or market operators [59]. The probabilistic cost of NSs is modeled as follows: H X

y CProb;NS 5

y51

H X m X y51 i51

5

H X y51

y

y;i CNS 3 pi

y CNS

5

H n X

y CNS;fix

y 1 CNS;var

y 1 CO&M

y 2 Csalvage

o

ð5:32Þ

y51

y

where CNS;fix and CNS;var are NPVs of the fixed and variable investment cost of NSs y in planning stage y, respectively, as provided in [24,28,60]; CO&M is the NPV of opery y ation and maintenance (O&M) cost; and Csalvage is the salvage value of CNS;var based on the straight line calculation [61]. The O&M cost is modeled as: y y y y y y 5 CO&M;NS 1 ELoss 3 CELoss 1 PLoss 3 CPLoss CO&M

ð5:33Þ

Impact of neighborhood energy trading and renewable energy communities y

where CO&M;NS is the fixed O&M cost of the NSs in planning stage y, which is calcuy lated as a fixed percentage of variable investment cost ðCNS;var Þ, that is 2%, in the same y y planning stage in this model. ELoss and PLoss are total expected energy loss and power y loss in planning stage y, respectively, which are calculated for each load level. CELoss y and CPLoss are the unit cost of energy loss and power loss in planning stage y, respectively. More details of the formulations are provided in [28] and [60]. In this model, the reliability cost is computed using the standard indices system average interruption duration index (SAIDI) and system average interruption frequency index (SAIFI), as follows: H X

y Crel 5

y51

H X

y y SAIDI y 3 CSAIDI 1 SAIFI y 3 CSAIFI

ð5:34Þ

y51

where SAIDI y and SAIFI y denote the expected SAIDI and SAIFI in planning stage y, y y respectively; CSAIDI and CSAIFI are the value of customer reliability (VCR) [62] for SAIDI per customer-minute and SAIFI per failure-customer in planning stage y, respectively. For evaluation of network reliability using these indices, network voltage and thermal limits are also included when the network relies on the cross-connects [63]. In this chapter, only existing cross-connects are considered. Consequently, the reliability cost includes the cost of failing to supply the demand due to technical constraints. The technical constraints in (5.30) set the upper and lower bounds on the voltage and the current magnitudes and are expressed in a probabilistic manner. For example, the node voltages must be within the acceptable range with a probability of 95%. The statistical measures for the node voltages and branch currents are obtained using the efficient probabilistic distribution state estimator proposed in [46]. The constraints (5.30) are relaxed and added to the objective function as a penalty term ofPwhich the penaltyP factors are set a big number. n21 In general, the penalty term is equal to ni51 CΔvi ðΔvi Þ 1 i51 CΔIi ðΔIi Þ, where Δvi and ΔIi are the voltage and current magnitude violations at i-th bus and i-th branch, respectively; CΔvi and CΔIi are the cost functions associated with Δvi and ΔIi , respectively.

5.4.8 Solution strategy The proposed model is a nonconvex mixed-integer nonlinear programming problem. The available mathematical solvers face difficulties in solving these problems, especially in real size networks [26]. Also, solving these problems with mathematical solvers entails an enormous computational cost, which increases exponentially with the size of the problem (the curse of dimensionality) [64]. Moreover, the time dynamic nature of the problem worsens the computational burden of the numerical solution. As a result, in most of the cases, it is not feasible to solve a full dynamic programming problem [64]. Given that, a forwardbackward pseudo-dynamic algorithm is utilized to break the multistage problem up into a

159

160

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

sequence of single-stage problems and each single-stage problem is solved using a heuristic optimization approach. The adopted solution strategy has three steps: 1. the original multistage MSDEP problem is decomposed into single-stage problems, 2. the single-stage problems are solved by using a heuristic approach, and 3. the single-stage solutions are coordinated using the forwardbackward strategy to find the optimal plan. We use the forwardbackward approach developed in [24]. Briefly, the multistage planning problem is decomposed into single-stage planning problems. Considering a planning year as the reference year, the optimal plan to meet the forecast demand for this reference year is obtained, as shown in Fig. 5.22. The planning exercise proceeds for the previous planning years to achieve the optimal investments that should be commissioned in the years before the reference year. Then, the forward-filling approach is carried out from the reference year to the horizon year H. This procedure is repeated for all possible selections for the reference year and results are compared. A modified PSO (MPSO) algorithm is used in this study due to its promising capability in handling the large scale nonlinear programming problems [6468] and the stability and convergence of PSO in multidimensional complex spaces [67]. Also, PSO is more efficient than classical linear programming, as investigated in [69]. The idea of mutation in GA is added into standard PSO particle update rules [66,70]. Fig. 5.23 shows the flowchart of the proposed MPSO. The set of decision variables in each particle of MPSO for single-stage planning is shown in Fig. 5.24. The DSNS index is a number between 10 and 90 (in the step of 10) represents the level of demand for POE10 and POE90, respectively. The total y probabilistic cost of each particle, CProb , is calculated as shown in Fig. 5.25.

5.5 Application of the planning model to case studies and analysis of the results The proposed risked-managed approach is tested on three networks as below: • Situation A: where the consumers are only trading energy with the utility, for which the risk-managed method is employed to provide an optimum plan for the following cases: 1) IEEE 13-bus radial feeder, and 2) realistic 747-bus distribution network. The latter is chosen to verify the effectiveness of the proposed planning approach in real size distribution networks. • Situation B: where the consumers are able to trade energy amongst themselves in a NET platform. In this situation, the simulation results are discussed on the IEEE 33-bus radial distribution networks. The characteristics of the candidate network solutions are the same as in [24]. The parameters of the simulation are listed in Table 5.9, as provided by the local utility Ergon Energy Co. Ltd. The uncertainty level is 3% and increases by 3% each year.

Impact of neighborhood energy trading and renewable energy communities

Figure 5.22 The flowchart of the proposed approach for MSDEP.

161

162

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 5.23 The flowchart of the MPSO.

The parameters of the MPSO are particle population 5 50, maximum iterations 5 100, ψmax 5 4.05, K 5 0.99, and the mutation probability 5 80% and the mutation operator is applied to 10% of particle population [70]. These parameters are tuned for this particular MSDEP problem to guarantee the convergence and stability of the MPSO algorithm [67].

…

…

…

DSNS index

Conductor

Fixed ESS/PV

Reactive power comp.

VRs

Impact of neighborhood energy trading and renewable energy communities

…

Figure 5.24 The structure of a particle in MPSO.

y Figure 5.25 The approach to calculate CProb .

5.5.1 Situation A, Case 1: IEEE 13-bus radial feeder The proposed planning approach is tested on the IEEE 13-bus network, as shown in Fig. 5.26, are presented. Other network parameters are provided in [24] and [71]. The effective time series load profiles are constructed for each node using the historical data for demand and renewable generation [72]. Two capacitor banks are already installed at buses 6 and 10 with a capacity of 100 and 600 kvar in this network, respectively. A unit cost of $0.382/kWh [57] is considered for the DR program. The DR uncertainty is modeled using the truncated

163

164

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 5.9 The parameters for the MSDEP. Parameter

Value

Interest rate (%) SAIDI cost ($/min-customer) SAIFI cost ($/failure-customer) Cost of power loss ($/kW-year) Cost of energy loss ($/kWh) Failure rate of OH/UG line. (f/km-year) Failure rate of OH/PM transformer (f/year) Repair time OH/UG line. (min) Repair time of OH/PM transformer (min) Switching time (min)

5 1.14 88 235 0.04 0.14/0.05 0.02/0.005 180/300 900 60

OH, Overhead; UG, underground; OH/PM, overhead pole mounted; f, failures.

Figure 5.26 IEEE 13 bus test system. Table 5.10 DR statistical parameters for uncertainty modeling. DR parameters

Mean

SD

Min.

Max.

Power (kVA) Duration (hours)

17.5 5

8.3 2

5 2

30 8

Gaussian PDF with the parameters listed in Table 5.10 for each bus. The constants for temporary ESS/DG costing A1, A2 and A3 are considered as $100, 0.4/kWh, and 100/kW, respectively [73]. A load growth of 4% in average over planning years is added to the model, which is applied to both the load level of each end-user and the number of end-users. The total probabilistic cost, the total RMC, and other costs (the sum of reliability cost and the total NS probabilistic cost) are reported in Table 5.11. The least-cost belongs to the plan

Impact of neighborhood energy trading and renewable energy communities

Table 5.11 Comparison of results in different selections for Ref. year. Ref. year

Total probabilistic cost (k$)

Total RMC cost (k$)

Other costs (k$)

1 2 3 4 5

2973 2858 3041 3063 3397

369 315 270 265 268

2604 2543 2771 2798 3129

Figure 5.27 Comparison of total costs versus “Ref. year”.

with Ref. Year 2 with a cost of $2858k. Although the selection of 2 for Ref. Year increases the RMC cost by $45k, 50k, and 47k, it results in a saving of $228k, 255k, and 586k in “other costs” when compared to plans with Ref. Year 3, 4, and 5. Also, the overinvestment in NSs as in the plan with Ref. Year 4 does not provide the most cost-effective plan. The total RMC decreases when Ref. Year of planning increases from year 1 to year 5 (Fig. 5.27). The opposite holds for the total NS cost when Ref. Year of planning is changed from year 1 to year 5. The planning with the Ref. Year 2 ha the lowest total cost when compared to other selections for Ref. Year, as seen in Fig. 5.27. The location and size of deployed NSs and NNSs for the optimal plan with Ref. Year 2 are provided in Table 5.12. For example, a new fixed ESS of 70 kVA is installed in year 2 at bus 10. As seen, the NSs including fixed ESSs and capacitors are installed at a few buses, which are bus 4, bus 6, and bus 10 to bus 13. However, the NNSs are implemented across the network, including all buses except buses 1, 2, 5, and 8. As expected, the proposed risk-managed approach provides the optimal combination of NSs and NNSs. For example, at bus 4 with the highest loading in the network [24], fixed ESSs of 640 and 600 kVA ratings are installed in years 3 and 5, respectively as well as a 325 kvar capacitor bank in year 2. Regarding NNSs, a total

165

Table 5.12 The location and size of NSs and NNSs at the optimal plan: Ref. year 5 2. Upgrades

Transformer Fix ESS Capacitor DR

Planning years

kVA kVA location kvar location kWh

location Temporary NNS

kW location

1

2

3

4

5

0 0  325 13 8.4, 0.4, 0.3, 2.3, 0.3, 0.3 4, 6, 9, 10, 12, 13 0 

0 70 10 325, 150 4, 12 28, 255, 50, 26, 65, 241, 77, 96, 161 3, 4, 6, 7, 9, 10, 11, 12, 13 28, 10, 10 4, 10, 13

0 640, 13, 30 4, 6, 12 0  15, 153, 20, 20, 43, 125, 48, 61, 85 3, 4, 6, 7, 9, 10, 11, 12, 13 13, 5, 6 4, 10, 13

0 15, 580, 75 6, 10, 12 25 13 16, 99, 12, 12, 32, 77, 26, 27, 50 3, 4, 6, 7, 9, 10, 11, 12, 13 11, 3, 3 4, 10, 13

0 600, 25, 60 4, 6, 13 0  7, 64, 11, 13, 16, 16, 18, 35, 50 3, 4, 6, 7, 9, 10, 11, 12, 13 5, 4 4, 12

Impact of neighborhood energy trading and renewable energy communities

load of 579.4 kWh is shifted at bus 4 during contingencies in the network. Moreover, temporary ESS/DG are operated at bus 4 from years 2 to 5 with a total power of 57 kW. Furthermore, it is seen that the implementation of NSs and NNSs at buses that are located far from the main transformer is higher, that is buses 10, 12, and 13. The voltage magnitudes at all buses are within 65% of the nominal voltage for a 95% confidence interval for Ref. Year 2, as shown in Fig. 5.28. Interestingly, the voltage drops to a level just above the lower bound of 0.95 per unit at bus 2 in year 1, but the voltage regulator (VR) installed in this network improves the voltage at bus 3. The least-cost expansion plans for uncertainty levels of 1.5%, 3% (base case), and 6% are presented in Fig. 5.29. The total probabilistic cost increases by 18% and 34% when the uncertainty level is increased from 1.5% to 3%, and 6%, respectively. However, the RMC cost increases by 45% and 217%, respectively, when the uncertainty level increased from 1.5% to 3% and 6%, respectively. This advocates that the RMC cost is remarkably more sensitive to the uncertainty level than the other costing terms. The RMC contributes

Figure 5.28 Voltage magnitude and 95% confidence interval for “Ref. year” 5 2.

Figure 5.29 The total RMC and NS cost for different levels of uncertainty.

167

168

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

12%, 13%, and 17%, in the total probabilistic cost for the uncertainty levels of 1.5%, 3%, and 6%, respectively. Also, the deployment of NSs increases to cope with the uncertainties in the future years, as well as the NNSs. This means that for a high level of uncertainty, the investment in expensive NSs is a reasonable choice. The planning with only NSs is carried out with the same conditions as the base case, and it shows an increase of $153k in the total probabilistic cost. This is the value that the NNSs can provide for utilities.

5.5.2 Situation A, Case 2: A realistic 747-bus radial feeder To investigate the performance of the proposed MSDEP in real size distribution networks, a realistic 747-bus distribution network is studied. The details of this network are available in [46]. An average load growth of 1.5% is considered for this 5-year planning study. The upper limit for DR is 2, 10, and 200 kVA for residential, commercial, and industrial customers, respectively. It was observed that the optimal solution is for the plan with Ref. Year 1. The planning result with Ref. Year 1 is presented in Table 5.13. As seen, different values for POEx are found in different years as the best compromise between the total cost of NSs and NNSs in the optimal plan. It takes about 33 hours for the proposed planning approach to run the 5-year MSDEP in a realistic 747-bus distribution network in MATLAB on Intel CORE i74770 PC with clock speed 3.4 GHz and 16 GB RAM.

5.5.3 Situation B: IEEE 33-bus radial feeder The proposed planning approach considering NET is tested on the IEEE 33-bus network, as shown in Fig. 5.30. Three cases are studied, as follows: Case 1) no NET is considered Case 2) NET is enabled with fixed network charge throughout the different load levels Case 3) NET is enabled with a network charge that can vary in different load levels The investment, operating, and the total costs incurred by both the utility and end-users are presented in Table 5.14. The aim of planning in all three cases is to Table 5.13 MSDEP Results for 5-year planning with Ref. year 1. Upgrades

Transformer (kVA) Fix ESS (kVA) Capacitor (kvar) DSNSy y CRMC (k$) y CProb (k$)

Planning years

Total

1

2

3

4

5

25 5 490 POE90 183 32,323

0 340 210 POE80 166 30,614

0 890 500 POE70 369 29,355

63 2645 1480 POE50 609 28,284

0 1345 975 POE40 940 27,184

88 5225 3655  2268 147,760

Impact of neighborhood energy trading and renewable energy communities 25

26

27

28

29

30

32

31

33

7

8

9

10

11

12

13

14

24 23

L1 1

L2 2

L3 3

L4 4

L5 5

L6 6

15

16

17

18

Figure 5.30 Schematic diagram of IEEE 33-bus test system.

Table 5.14 Comparison of MSDEP results for one single-stage planning. Case

Total cost (M$)

Utility’s revenue (M$)

Total end-users’ cost (M$)

Number of upgraded (lines, transformers)

1 2 3

9.2 8.9 8.7

216.5 216.5 216.5

25.7 25.4 25.2

(2,1) (1,1) (0,0)

minimize the total cost incurred by the utility and end-users, here called the cost of electrification. While the utility’s revenue remains unchanged, the total cost incurred by the end-users and the cost of electrification are reduced by $300k and 200k when comparing case 1 with case 2 and case 2 with case 3. Utilities are not generally interested in NET programs because it reduced the total energy purchases from the utilities, which means a loss of revenue for utilities. However, that does not always hold. According to Table 5.14, the utility does not see any loss of revenue in case 2 and case 3 (cases with NET) when compared to case 1 (without NET). That is due to avoiding line and transformer upgrade which covers the loss of revenue due to selling less energy to end-users. That also holds true when comparing case 2 with case 3. The energy flow of the trades of the end-users is shown in Fig. 5.31, where the width of arrows indicates the value of the energy traded between a seller and a buyer. As seen, only a small amount of power is imported by the substation at this load level, which is only to cover the ohmic loss in the network. In this load level, the network is almost working autonomously, and 100% of the demand is supplied locally.

169

170

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 5.31 Flow of energy trades in the NET scheme.

5.6 Conclusions The first part of this chapter has illustrated an ADMM-based distributed optimization procedure for the day-ahead scheduling of a local energy community with generation, loads and battery storage systems incorporating the calculation of the losses and their allocation to each transaction. A centralized approach implies that the prosumers communicate all the details of the equipment features as well as the load and production forecasts. The distributed procedure reduces the amount of shared information: indeed, the only information that every prosumer must communicate is the profile of the exchanges with the external grid and with the other prosumers for updating the multipliers at each iteration and for the evaluation of the transaction efficiencies. The structure of the day-ahead scheduling procedures is consistent with the billing scheme and the meters of the LEC. The results confirm that, in the considered LEC framework, each prosumer achieves a reduction in costs or increased revenues by participating in the LEC, compared to the case in which it can only transact with an external energy provider. The second part of the chapter focuses on an investment plan model for the utility and end-users, in which nonnetwork solutions such as DR and temporary ESS/DG have been considered. The produced energy by these solutions along with the energy of PVs and ESSs, invested by customers, are traded amongst end-users within a platform of the neighborhood energy trades (NET). The model proposes how to treat uncertainties in future on PV generation and DR availability to increase the profitability of investments by utilities and customers. Also, a multiyear forwardbackward planning strategy of network provides a better planning decision compared to other methods. Other than the investment of the utility in grid reinforcement, the model also incorporates the optimal investment of end-users in PV and battery units considering NET. The results obtained for test cases show that the planning model prevents the utility from over/underinvestment taking into account the beneficial effects of the

Impact of neighborhood energy trading and renewable energy communities

LEC and NNSs through NET that offers a platform that provides energy at more convenient prices for both the buyers and the sellers.

Acknowledgment This chapter summarizes the results obtained in research activities supported in part by ECSEL JU under grant agreement No 737434 (CONNECT), by the EU Horizon 2020 program—MSC grant agreement No 675318 (INCITE), and GECO - Green Energy Community co-funded by EIT Climate-KIC. These activities have been carried out in collaboration with Stefano Lilla, Fabio Napolitano, Fabio Tossani, Gerard Ledwich, Anula Abeygunawardana (coauthors of the papers that present some of the obtained results).

References [1] C. Inês, P.L. Guilherme, M.G. Esther, G. Swantje, H. Stephen, H. Lars, Regulatory challenges and opportunities for collective renewable energy prosumers in the EU, Energy Policy 138 (2020). Available from: https://doi.org/10.1016/j.enpol.2019.111212. [2] EU, Directive 2018/2001 of the European parliament and of the council on the promotion of the use of energy from renewable sources, (2018). [3] EU, Directive 2019/944 on common rules for the internal market for electricity and amending directive 2012/27/EU, (2019). [4] E. Mengelkamp, J. Gärttner, K. Rock, S. Kessler, L. Orsini, C. Weinhardt, Designing microgrid energy markets: a case study: the brooklyn microgrid, Appl. Energy 210 (2018) 870880. Available from: https://doi.org/10.1016/j.apenergy.2017.06.054. [5] G. Belli, G. Brusco, A. Burgio, M. Motta, D. Menniti, A. Pinnarelli, et al., An energy management model for energetic communities of smart homes: the power cloud, in: Proc. 2017 IEEE 14th Int. Conf. Networking, Sens. Control. ICNSC 2017, IEEE, (2017) 158162. Available from: https:// doi.org/10.1109/ICNSC.2017.8000084. [6] J. Lee, J. Guo, J.K. Choi, M. Zukerman, Distributed energy trading in microgrids: a game-theoretic model and its equilibrium analysis, IEEE Trans. Ind. Electron. 62 (2015) 35243533. Available from: https://doi.org/10.1109/TIE.2014.2387340. [7] Y. Zheng, Y. Song, D.J. Hill, Y. Zhang, Multiagent system based microgrid energy management via asynchronous consensus ADMM, IEEE Trans. Energy Convers. 33 (2018) 886888. Available from: https://doi.org/10.1109/TEC.2018.2799482. [8] B. Yan, M. Di Somma, P.B. Luh, G. Graditi, Operation optimization of multiple distributed energy systems in an energy community, in: 2018 IEEE Int. Conf. Environ. Electr. Eng. 2018 IEEE Ind. Commer. Power Syst. Eur. EEEIC/I CPS Eur (2018). Available from: https://doi.org/10.1109/ EEEIC.2018.8494476. [9] Y. Liu, Y. Li, H.B. Gooi, J. Ye, H. Xin, X. Jiang, et al., Distributed robust energy management of a multimicrogrid system in the real-time energy market, IEEE Trans. Sustain. Energy 10 (2019) 396406. Available from: https://doi.org/10.1109/TSTE.2017.2779827. [10] C. Orozco, S. Lilla, A. Borghetti, F. Napolitano, F. Tossani, An ADMM approach for day-ahead scheduling of a local energy community, in: 2019 IEEE Milan PowerTech, Milan. Italy (2019) 16. Available from: https://doi.org/10.1109/PTC.2019.8810578. [11] S. Lilla, C. Orozco, A. Borghetti, F. Napolitano, F. Tossani, Day-ahead scheduling of a local energy community: an alternating direction method of multipliers approach, IEEE Trans. Power Syst. 35 (2020) 11321142. Available from: https://doi.org/10.1109/TPWRS.2019.2944541. [12] H. Le Cadre, P. Jacquot, C. Wan, C. Alasseur, Peer-to-peer electricity market analysis: from variational to generalized Nash equilibrium, Eur. J. Oper. Res. 282 (2020) 753771. Available from: https://doi.org/10.1016/j.ejor.2019.09.035.

171

172

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[13] E. Munsing, J. Mather, S. Moura, Blockchains for decentralized optimization of energy resources in microgrid networks, in: 2017 IEEE Conf. Control Technol. Appl., (2017) 21642171. Available from: https://doi.org/10.1109/CCTA.2017.8062773. [14] M.L. Di Silvestre, P. Gallo, M.G. Ippolito, E. Riva, Sanseverino, A technical approach to the energy blockchain in microgrids, IEEE Trans. Ind. Inf. 14 (2018) 47924803. Available from: https://doi.org/10.1109/TII.2018.2806357. [15] S. Boyd, N. Parikh, E. Chu, B. Peleato, J. Eckstein, Distributed optimization and statistical learning via the alternating direction method of multipliers, Mach. Learn. (2011). Available from: https://doi.org/10.1561/2200000016. [16] M. Roelofs, J. Bisschop, AIMMS language reference, (2013). [17] IBM/ILOG, Introducing IBM ILOG CPLEX optimization studio V12.8, (2018). [18] C. Orozco, A. Borghetti, F. Napolitano, F. Tossani, Multistage day-ahead scheduling of the distributed energy sources in a local energy community, 2020 IEEE International Conference on Environment and Electrical Engineering and 2020 IEEE Industrial and Commercial Power Systems Europe (EEEIC / I&CPS Europe), Madrid, Spain, 2020, pp. 17. Available from: https://doi.org/ 10.1109/EEEIC/ICPSEurope49358.2020.9160579. [19] C. Orozco, A. Borghetti, S. Lilla, G. Pulazza, F. Tossani, Comparison between multistage stochastic optimization programming and Monte Carlo simulations for the operation of local energy systems, in: 2018 IEEE Int. Conf. Environ. Electr. Eng., Palermo, (2018). Available from: https://doi.org/ 10.1109/EEEIC.2018.8494563. [20] G.C. Lazaroiu, V. Dumbrava, G. Balaban, M. Longo, D. Zaninelli, Stochastic optimization of microgrids with renewable and storage energy systems, in: EEEIC 2016-Int. Conf. Environ. Electr. Eng., (2016). Available from: https://doi.org/10.1109/EEEIC.2016.7555486. [21] S.S. Reddy, V. Sandeep, C.-M. Jung, Review of stochastic optimization methods for smart grid, Front. Energy 11 (2017) 197209. Available from: https://doi.org/10.1007/s11708-017-0457-7. [22] Y. Yuan, Q. Li, W. Wang, Optimal operation strategy of energy storage unit in wind power integration based on stochastic programming, IET Renew. Power Gener. 5 (2011) 194. Available from: https://doi.org/10.1049/iet-rpg.2009.0107. [23] B.R. Pereira Jr., A.M. Cossi, J. Contreras, J.R.S. Mantovani, Multiobjective multistage distribution system planning using tabu search, IET Gener. Transm. Distrib. 8 (2014) 3545. Available from: https://doi.org/10.1049/iet-gtd.2013.0115. [24] A. Abeygunawardana, A. Arefi, G. Ledwich, An efficient forward-backward algorithm to MSDEPP including batteries and voltage control devices, in: PES Gen. Meet. Conf. Expo. 2014 IEEE, (2014) 15. Available from: https://doi.org/10.1109/PESGM.2014.6939812. [25] A. Arefi, G. Ledwich, F. Boulaire, A. Abeygunawardana, R. Drogemuller, A flexible tool for integrated planning of active distribution networks, in: 23rd Int. Conf. Exhib. Electr. Distrib., Lyon, France, (2015). [26] M.E. Samper, A. Vargas, Investment decisions in distribution networks under uncertainty with distributed generation - Part II: implementation and results, IEEE Trans. Power Syst. 28 (2013) 23412351. Available from: https://doi.org/10.1109/TPWRS.2013.2239667. [27] K. Zou, A.P. Agalgaonkar, K.M. Muttaqi, S. Perera, Distribution system planning with incorporating DG reactive capability and system uncertainties, IEEE Trans. Sustain. Energy 3 (2012) 112123. Available from: https://doi.org/10.1109/TSTE.2011.2166281. [28] I. Ziari, G. Ledwich, A. Ghosh, G. Platt, Optimal distribution network reinforcement considering load growth, line loss, and reliability, IEEE Trans. Power Syst. 28 (2013) 587597. Available from: https://doi.org/10.1109/TPWRS.2012.2211626. [29] B. Zeng, J. Zhang, X. Yang, J. Wang, J. Dong, Y. Zhang, Integrated planning for transition to lowcarbon distribution system with renewable energy generation and demand response, IEEE Trans. Power Syst. 29 (2014) 11531165. Available from: https://doi.org/10.1109/TPWRS.2013.2291553. [30] M.F. Shaaban, E.F. El-Saadany, Accommodating high penetrations of pevs and renewable dg considering uncertainties in distribution systems, IEEE Trans. Power Syst. 29 (2014) 259270. Available from: https://doi.org/10.1109/TPWRS.2013.2278847. [31] C.L.T. Borges, V.F. Martins, Multistage expansion planning for active distribution networks under demand and distributed generation uncertainties, Int. J. Electr. Power Energy Syst. 36 (2012) 107116. Available from: https://doi.org/10.1016/j.ijepes.2011.10.031.

Impact of neighborhood energy trading and renewable energy communities

[32] S. Haffner, L.F.A. Pereira, L.A. Pereira, L.S. Barreto, Multistage model for distribution expansion planning with distributed generation-part II: numerical results, IEEE Trans. Power Deliv. 23 (2008) 924929. Available from: https://doi.org/10.1109/TPWRD.2008.917911. [33] A.S. Bin Humayd, K. Bhattacharya, Comprehensive multi-year distribution system planning using backpropagation approach, IET Gener. Transm. Distrib. 7 (2013) 14151425. Available from: https://doi.org/ 10.1049/iet-gtd.2012.0706. [34] K. Nara, T. Satoh, K. Aoki, M. Kitagawa, Multi-year expansion planning for distribution systems, IEEE Trans. Power Syst. 6 (1991) 952958. Available from: https://doi.org/10.1109/59.119234. [35] H. Lee Willis, R.W. Powell, H.N. Tram, Long-range distribution planning with load forecast uncertainty, IEEE Trans. Power Syst. 2 (1987) 684691. Available from: https://doi.org/10.1109/ TPWRS.1987.4335195. [36] P.M.S. Carvalho, L.A.F.M. Ferreira, F.G. Lobo, L.M.F. Barruncho, Distribution network expansion planning under uncertainty: a hedging algorithm in an evolutionary approach, IEEE Trans. Power Deliv. 15 (2000) 412416. Available from: https://doi.org/10.1109/61.847282. [37] D.T.C. Wang, L.F. Ochoa, G.P. Harrison, Modified GA and data envelopment analysis for multistage distribution network expansion planning under uncertainty, IEEE Trans. Power Syst. 26 (2011) 897904. Available from: https://doi.org/10.1109/TPWRS.2010.2057457. [38] S. Ganguly, D. Samajpati, Distributed generation allocation on radial distribution networks under uncertainties of load and generation using genetic algorithm, IEEE Trans. Sustain. Energy 6 (2015) 688697. Available from: https://doi.org/10.1109/TSTE.2015.2406915. [39] Y. Zhou, Y.P. Li, G.H. Huang, Y. Zhou, A robust approach for planning electric power systems associated with environmental policy analysis, Electr. Power Syst. Res. 95 (2013) 99111. Available from: https://doi.org/10.1016/j.epsr.2012.08.015. [40] S. Montoya-Bueno, J.I. Munoz, J. Contreras, A. Stochastic, Investment model for renewable generation in distribution systems, IEEE Trans. Sustain. Energy 6 (2015) 14661474. Available from: https://doi.org/10.1109/TSTE.2015.2444438. [41] V.F. Martins, C.L.T. Borges, Active distribution network integrated planning incorporating distributed generation and load response uncertainties, IEEE Trans. Power Syst. 26 (2011) 21642172. Available from: https://doi.org/10.1109/TPWRS.2011.2122347. [42] A. Soroudi, Possibilistic-scenario model for DG impact assessment on distribution networks in an uncertain environment, IEEE Trans. Power Syst. 27 (2012) 12831293. Available from: https://doi.org/10.1109/ TPWRS.2011.2180933. [43] M.R. Alam, M. St-Hilaire, T. Kunz, An optimal P2P energy trading model for smart homes in the smart grid, Energy Effic. 10 (2017) 14751493. Available from: https://doi.org/10.1007/s12053-0179532-5. [44] M. Khorasany, Y. Mishra, G. Ledwich, Market framework for local energy trading: a review of potential designs and market clearing approaches, IET Gener. Transm. Distrib. 12 (2018) 58995908. Available from: https://doi.org/10.1049/iet-gtd.2018.5309. [45] AS/NZS ISO 31000:2009, Risk management—principles and guidelines, (2009). [46] A. Arefi, G. Ledwich, B. Behi, An efficient DSE using conditional multivariate complex gaussian distribution, IEEE Trans. Smart Grid 6 (2015) 21472156. Available from: https://doi.org/ 10.1109/TSG.2014.2385871. [47] M.T. Au, J.V. Milanovi´c, Development of stochastic aggregate harmonic load model based on field measurements, IEEE Trans. Power Deliv. 22 (2007) 323330. Available from: https://doi.org/ 10.1109/TPWRD.2006.881455. [48] C. Muscas, F. Pilo, G. Pisano, S. Sulis, Optimal allocation of multichannel measurement devices for distribution state estimation, IEEE Trans. Instrum. Meas. 58 (2009) 19291937. Available from: https://doi.org/10.1109/TIM.2008.2005856. [49] S.N. Luko, Risk assessment techniques, Qual. Eng. 26 (2014) 379382. Available from: https://doi.org/ 10.1080/08982112.2014.875769. [50] D. Ionita, Current established risk assessment methodologies and tools, (2013). [51] C. Wilhelmsen, L.T. Ostrom, Risk assessment: tools, techniques, and their applications, John Wiley & Sons, Somerset, NJ, 2012. [52] I. Forster, Risk management, World Technologies, New Delhi, 2012.

173

174

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[53] A.M.A.K. Abeygunawardana, A. Arefi, G. Ledwich, Estimating and modeling of distribution network costs for designing cost-reflective network pricing schemes, in: Power Energy Soc. Gen. Meet. 2015 IEEE, (2015) 15. Available from: https://doi.org/10.1109/PESGM.2015.7286418. [54] F. Javed, N. Arshad, F. Wallin, I. Vassileva, E. Dahlquist, Forecasting for demand response in smart grids: an analysis on use of anthropologic and structural data and short term multiple loads forecasting, Appl. Energy 96 (2012) 150160. Available from: https://doi.org/10.1016/j.apenergy.2012.02.027. [55] A. Arefi, A. Abeygunawardana, G. Ledwich, A new risk-managed planning of electric distribution network incorporating customer engagement and temporary solutions, IEEE Trans. Sustain. Energy (2016). Available from: https://doi.org/10.1109/TSTE.2016.2573290. [56] Energex, Peak smart air-conditioning, 2014 (2014). Available from: https://www.energex.com.au/ residential-and-business/positive-payback/positive-payback-for-households/households. [57] M. Fahrioglu, F.L. Alvarado, Using utility information to calibrate customer demand management behavior models, IEEE Trans. Power Syst. 16 (2001) 317322. Available from: https://doi.org/ 10.1109/59.918305. [58] Ergon Energy Co. Ltd, Final recommendation south mackay area, (2014). [59] AEMO—Australian energy market operator, regional demand definition, 10 February, (2012). [60] I. Ziari, G. Ledwich, A. Ghosh, G. Platt, Integrated distribution systems planning to improve reliability under load growth, IEEE Trans. Power Deliv. 27 (2012) 757765. Available from: https://doi.org/10.1109/TPWRD.2011.2176964. [61] P.D.K. Jerry, J. Weygandt Donald, E. Kieso, Accounting principles, John Wiley & Sons, 2007. [62] AEMO—Australian energy market operator, value of customer reliability final report, (2014). [63] A. Coelho, C.L.C. de Castro, M.G. da Silva, A.B. Rodrigues, Inclusion of voltage drop and feeder loading constraints in the evaluation of reliability indices for radial distribution networks, IEE Proc. Gener. Transm. Distrib. 153 (2006) 661669. Available from: https://doi.org/10.1049/ip-gtd:20050419. [64] Y. del Valle, G.K. Venayagamoorthy, S. Mohagheghi, J.C. Hernandez, R.G. Harley, Particle swarm optimization: basic concepts, variants and applications in power systems, IEEE Trans. Evol. Comput. 12 (2008) 171195. Available from: https://doi.org/10.1109/TEVC.2007.896686. [65] K.E. Parsopoulos, M.N. Vrahatis, On the Computation of all global minimizers through particle swarm optimization, IEEE Trans. Evol. Comput. 8 (2004) 211224. Available from: https://doi.org/ 10.1109/TEVC.2004.826076. [66] P. Faria, J. Soares, Z. Vale, H. Morais, T. Sousa, Modified particle swarm optimization applied to integrated demand response and DG resources scheduling, IEEE Trans. Smart Grid 4 (2013) 606616. Available from: https://doi.org/10.1109/TSG.2012.2235866. [67] M. Clerc, J. Kennedy, The particle swarm-explosion, stability, and convergence in a multidimensional complex space, IEEE Trans. Evol. Comput. 6 (2002) 5873. Available from: https://doi.org/ 10.1109/4235.985692. [68] Y. del Valle, R.G. Harley, G.K. Venayagamoorthy, Comparison of enhanced-PSO and classical optimization methods: a case study for STATCOM placement, in: Intell. Syst. Appl. Power Syst. 2009. ISAP’09. 15th Int. Conf., (2009) 17. Available from: https://doi.org/10.1109/ISAP.2009.5352884. [69] M.I. Doma, Particle swarm optimization in comparison with classical optimization for GPS network design, J. Geod. Sci. 3 (2014) 8. Available from: https://doi.org/10.2478/jogs-2013-0030. [70] A. Arefi, M.R. Haghifam, S.H. Fathi, Distribution harmonic state estimation based on a modified PSO considering parameters uncertainty, in: PowerTech, 2011 IEEE Trondheim, (2011) 17. Available from: https://doi.org/10.1109/PTC.2011.6019326. [71] IEEE PES Test Feeder Working Group, IEEE PES distribution test feeders, 13-Bus test feed. (1991). Available from: https://site.ieee.org/pes-testfeeders/resources/. [72] Perth solar city annual report, (n.d.). Available from: http://perthsolarcity.com.au/annual_report/ index.html. [73] A.H. Mohsenian-Rad, A. Leon-Garcia, Optimal residential load control with price prediction in real-time electricity pricing environments, IEEE Trans. Smart Grid 1 (2010) 120133. Available from: https://doi.org/10.1109/TSG.2010.2055903.

CHAPTER 6

Fostering DER integration in the electricity markets Julia Merino1,2, Inés Gómez1, Jesús Fraile-Ardanuy3, Maider Santos1, Andrés Cortés1,2, Joseba Jimeno1 and Carlos Madina1 1

TECNALIA, Basque Research and Technology Alliance (BRTA), Derio, Spain Department of Electrical Engineering, University of the Basque Country, Bilbao, Spain 3 Information Processing and Telecommunication Center (IPTC-SISDAC), Universidad Politécnica de Madrid, Madrid, Spain 2

Abbreviations ACER aFRR BSP CBA CE CHP CS DER DG DR DSO EB GL EC EU EV FCR ICT IEM IGCC LEM LOLP mFRR MP NE P2P PABP PDD PV RES RI RR

agency for the cooperation of energy regulators automatic frequency restoration reserve balancing service provider cost-benefit analysis continental Europe combined heat and power coordination scheme distributed energy resource distributed generation demand response distribution system operator electricity balancing guideline European commission European Union electric vehicle frequency contrainment reserves information and communication technologies internal electricity market international grid control cooperation local energy market loss of load probability manual frequency restoration reserve marginal pricing northern Europe peer-to-peer pay-as-bid pricing probability density distribution photovoltaic renewable energy sources reference incident replacement reserve

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00008-X

r 2021 Elsevier Inc. All rights reserved.

175

176

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

SG SO SOC T&C TCL TSO VRES

synchronous generator system operator state of charge terms and conditions thermostatically controlled loads transmission system operator variable renewable energy sources

This chapter provides insights into the participation of distributed energy resources (DERs) in the electricity markets. The massive integration of distributed energy generation coming from renewable energy sources (RES) with intermittent character, together with the empowering of the customers and the emergence of new agents, such as electric vehicles, are forcing the power systems to evolve. In this context, the provision of flexibility services that has been traditionally done by a reduced number of big generation plants will be also done by the new DERs. Throughout the chapter, a review of the possibilities of the DERs to provide flexibility services, the needs of those services in the future power systems and the current market framework will be reviewed. Eventually, the need to develop market-based solutions (local energy markets) to overcome the limitations of DERs participation in the current electricity markets will be analyzed.

6.1 Distributed energy resources as providers of flexibility services According to EDSO for smart grids [1], “flexibility used by network operators will be referred to as ‘system flexibility services.’ System flexibility services are here defined as any service delivered by a market party and procured by the distribution system operators (DSOs) to maximize the security of supply and the quality of service in the most efficient way.” In this sense, this section shows the state of the art of the products and services that DERs can provide for voltage and frequency control.

6.1.1 Products and services for voltage and frequency control The power system needs to keep both the voltage and the frequency within preassigned limits to guarantee the security and quality of the supply and to avoid possible damage to the connected devices. In this sense, depending on the purpose of the delivered flexibility, flexibility services can be classified as [2,3]: (A) frequency services (mainly for balancing); (B) nonfrequency services; (C) congestion management (Fig. 6.1). Apart from these flexibility services, the NREL Report [4] also includes the primary frequency response and the inertial response in the “Frequency-Responsive Reserve Requirements.” Additionally, this report accounts for voltage control and black start as “Other Essential Reliability Services.”

Fostering DER integration in the electricity markets

Figure 6.1 Flexibility services (based on [2]).

To enable a market-based allocation of these grid services and thus enable market parties to effectively bid into these markets, products for grid services need to be defined. Next, a high-level description of the main flexibility services is made. 6.1.1.1 Balancing or frequency control According to the Energy Balancing Guideline (EBGL) [5], balancing means “all actions and processes, on all timelines, through which TSOs ensure, in a continuous way, the maintenance of system frequency within a predefined stability range and compliance with the amount of reserves needed with respect to the required quality.” Different reserve products are used to restore the frequency whenever it is necessary: • Frequency containment reserves (FCR): is the first control action to be activated, usually within 30 s, in a decentralized fashion over the synchronous area. • Frequency Restoration Reserve (FRR): in the centralized automated control, activated by the TSO in the time interval between 30 s and 15 min. It can be with automatic activation (aFRR) or with manual activation (mFRR). • Replacement reserves (RR): is a manually activated reserve (in Continental Europe) in the time from 15 min up to hours. 6.1.1.2 Congestion management According to Regulation (EC), No 714/2009 [6] a congestion is “a situation in which an interconnection linking national transmission networks cannot accommodate all physical flows resulting from international trade requested by market participants, because of a lack of capacity of the interconnectors and/or the national transmission systems concerned.” Until recently, congestions at transmission mainly occurred on the border or between TSO control areas. However, the increase of renewable energy sources

177

178

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

(RES) in the generation mix has resulted in structural congestion issues within certain regions countries. For the provision of congestion management services, two products are considered [7]. • Reserved, capacity-based product: a product to cope with structural constraints procured at a certain availability price. It is automatically activated when the service is needed and called upon by the relevant system operator. • Nonreserved, energy-based product: a product to cope with sporadic constraints procured at an energy price. It is manually activated. 6.1.1.3 Voltage control As stated in The EU Electricity Network Codes technical report [8], to achieve the desire voltage profile the voltage is controlled by controlling the generation, consumption, and flow of reactive power at all levels in the system. It is important to note that voltage is a local issue what means that all the reactive power exchanges need to take place nearly close to the point at which occurs a voltage deviation. Depending on the activation time voltage actions can be divided into a three-level hierarchy (from a few seconds to a couple of hours), from fast automatic primary (containment) and secondary (restoration) control to slower manual and economically optimized tertiary control [3,9]. Two reactive products are defined to cover the requirements for reactive power [7]. • Steady-state reactive power: a product to keep the voltage profile within the safe range under normal operation. The reactive power provision will be in function of a voltage setpoint at the injection point set by the system operator (SO). This product is limited to units capable to provide reactive power as a function of grid voltage. • Dynamic reactive power: a product to control the voltage under perturbances. The SOs request for a punctual regulation of reactive power exchange. This product is open to all technologies capable of following the request within specified time scales. 6.1.1.4 Inertial response The inertia of a power system is defined by ENTSO-E [10] as “the ability of a system to oppose changes in frequency due to resistance provided by the kinetic energy of rotating masses in individual turbine-generators.” Therefore, the inertia is an inherent characteristic of conventional synchronous generators (SGs) that prevents fast frequency variations in the first few cycles after a power imbalance. The replacement of conventional SGs by means of DER, which are mostly converter-interfaced with low or no inherent inertia1, could lead to frequency stability issues. 1

DERs connected to the power system through power electronic converters are electrically decoupled from the grid; therefore, they do not naturally contribute to system inertia.

Fostering DER integration in the electricity markets

To reduce the negative impact that the lack of inertia due to the massive presence of DERs can derive, it is of high importance to make the DERs behave in a similar way to synchronous generators regarding inertial response. For this purpose, a particular feature, known as virtual or synthetic inertia, in contrast to the (physical) inertia, has been proposed as a service [3]. 6.1.1.5 Black start Black start is the procedure to recover the power system from the total or partial shutdown. Black start providers are designated generators (known as Black Start units) that can restore electricity to the grid without using an outside electrical supply to ensure that the power system can always be effectively and economically restored. There is no commonly agreed approach among European TSOs for black start service and thus, black start is seldom clearly defined, provided, and remunerated [3,11].

6.1.2 Characterization of distributed energy resources as flexibility providers One of the major global trends that affect the evolution of the power system is decentralization (see Chapter 2). In this sense, the provision of flexibility services by DERs is crucial and brings challenges in planning and operating the power system in terms of changes in dynamic response. This is due to the presence of power electronic devices, new utilization patterns due to more dispersed generation units and new types of demand among others [12]. The provision of flexibility services by DERs is possible, on the one hand, thanks to technological advances in RES and storage combined with the deployment of automation and monitoring technologies and, on the other, to regulatory changes [13]. To assess the real provision capability of DERs to the flexibility services is of high importance to review the technical capabilities of different DERs to provide those flexibility services. Since the contribution to the flexibility services depends on the characteristics of the system and the DER, and on the requirements to be fulfilled for the provision of each specific service, all these aspects need to be considered when quantifying such flexibility service provision. The technical capability of DERs for providing flexibility services will depend on the grid coupling technologies and the associated control capabilities [14]. Depending on their physical behavior, DERs can be classified into eight different groups [13]: • Variable Renewable Energy Sources (VRES): The flexibility that they can provide in terms of active power comes from curtailing their power output (downwards flexibility). Normally these units are connected through power electronic devices whose reactive power can be potentially controlled to provide voltage regulation services.

179

180

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

Electrical storage (stationary): These DERs provide a high level of flexibility since they can provide both upwards and downwards flexibility when required. This flexibility is limited by their energy capacity, maximum, and minimum State of Charge (SOC) and the degradation suffered by the depth of discharge and number of charge/discharge cycles. • Electric Vehicles (EVs): EVs can be controlled by setting their charging start times, charging power and in the most advanced cases even injecting power to the grid (V2G). In the case of EVs, the flexibility depends not only on the technology itself but also on the users’ behavior including EV parking periods, SOC at arrival time and desired SOC at departure time. • Conventional generators: Typical conventional generators connected at distribution level are the ones used to provide back-up energy in buildings or industrial sites. These generators run on fossil fuels (gas, oil) and are fully controllable within their operating constraints (maximum and minimum power, ramping times, maximum on and off periods). • Combined Heat and Power (CHP): They can produce heat to be used in industrial processes and thermal purposes in buildings. As a by-product, they also produce electricity that increases the overall energy production efficiency and profitability. Being in most cases heat demand-driven resources, the flexibility highly depends on the capability of the heat demand process to support changes in the heat input. • Thermostatically Controlled Loads (TCL): TCLs are loads, for which the main purpose is to convert electricity into heating or cooling energy. Examples of these loads are: air conditioning systems, space heating devices, and water heaters. They are controlled by a thermostat that sets the desired temperature and the electricity consumption depends on the internal thermodynamic processes (thermal losses and gains, the thermal capacity of the medium being heated/cooled, etc.). Typical control techniques include the modification of temperature set points or the duty cycles of the devices. The flexibility they can provide is limited. • Load shifting devices: These are loads that once started they consume power following a fixed power profile. Residential loads such as washing machines, dryers and dishwashers as well as certain industrial processes form part of this category of DERs. The flexibility is provided by selecting the start time of the process within a certain time window set by the end-user. • Load curtailment devices: There are certain loads whose power consumption can be curtailed to some extent without a significant impact on the end-user. Loads such as dimmable lighting are considered within this category. Some studies in the literature analyze the provision of services from the perspective of the DER’s availability [13,15,16]. Depending on the required flexibility service, some DERs can be more suitable than others.

Fostering DER integration in the electricity markets

Ancillary services

Wind

PV

Stationary storage: batteries

Mobile storage: Evs

CHP

TCL

Shiftable loads: WET appliances

Shiftable loads: industrial processes

Curtailable loads

FFR FCR FRR RR

FFR FCR FRR RR

RM

RM

FRTC CMVC PVC SVC

FRTC CMVC PVC SVC

TVC

TVC

Fast frequency reserve Frequency containment reserve Frequency restoraon reserve Restoraon reserve

Ramp margin (ramp control) Fault ride-through capability Congeson management voltage control Primary voltage control Secondary voltage control Terary voltage control

Very good capabilies Good capabilies Lile capabilies Very lile capabilies No capabilies

Figure 6.2 Capabilities of DERs to provide future system services (based on [13]).

• Frequency control services: In general, this control (based on active power) can be provided by almost all DER. Conventional generators as well as storage show good capabilities because of their high availability and flexibility. By contrast intermittent energy systems (wind, PV, and hydropower plants), as well as flexible loads, even if they can control their active power has limited availability. • Voltage control, optimization of grid losses and congestion management are mainly dependent on the reactive power control capability of the grid-coupling technology and also on the active power control capability of the DER unit. Inverter-coupled storage units show a very good reactive power control capability. • Black Start needs the same control capabilities as a unit to work in islanded mode. Besides, it demands a grid-independent system start. This is possible for the same DER units if the necessary storage support is assumed to be implemented. A qualitative analysis regarding the capability to provide system services by different DERs was performed in [13]. The analysis was carried out for both the current and the expected situation for the year 2030, considering the evolution of technologies and requirements for the system services provision. Fig. 6.2 shows the qualitative mapping at the 2030 horizon; the green color indicates good technical capabilities, while red indicates no capabilities to provide the indicated services.

6.2 The regulatory framework for the participation of distributed energy resources in different electricity markets This section analyses the current regulation existing in Europe for the provision of flexibility services by different providers. According to regulatory requirements, barriers for the participation of the DERs still exist, because of the type of the accepted

181

182

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

bids or the technical specifications required for the participation in the electricity markets. However, in several countries it is already possible for the DERs, by participation in the markets or because the SOs can directly contract that flexibility to the providers.

6.2.1 European regulatory context The widespread deployment of the DERs, the integration of the RES and the development of new technological solutions have led to different operations of the current electrical systems. Moreover, the roles to be developed by the participants in the electricity markets are also changing drastically. For example, consumers are empowered to actively participate in the electricity market and generate their electricity. Therefore, the regulatory framework must be adapted and consistent with the real needs as the electricity markets evolve. Some of the main regulatory milestones in the last five years are summarized below, mainly, focusing on the issues regarding the integration of DERs in the future electricity market, and the updated roles to be performed by the participants. 6.2.1.1 Clean energy package for all Europeans In November 2016, the European Commission (EC) proposed a legislative package, the so-called Clean Energy for all Europeans Package, which aimed at improving the electricity flow across borders, updating supply/demand mechanisms and establishing new market designs to accommodate the new RES-based energy system. In 2019, the European Union (EU) completed a comprehensive update of its energy policy framework to facilitate the transition away from fossil fuels towards cleaner energy and launched a reviewed version of this legislative package. The Clean Energy for all Europeans Package consists of eight legislative acts. According to the agreement between the Council and the European Parliament (May 2018
May 2019) and the entry into force of the different EU rules, the Member countries have now to transpose the new directives into national laws [17]. Some of the most relevant documents published within this Package include; the Electricity Market Regulation [18], Directive on the Internal Market for Electricity [19] and the Renewable Energy Directive [20]. Based on these documents, Member States shall ensure that their national legislation does not unduly hamper cross-border flows of electricity, consumer participation and the investments into flexible energy generation or energy storage. Next, some of the most relevant implications regarding the participation in the electricity markets and the role to be performed by Transmission System Operators (TSOs) and Distribution System Operators (DSOs) are summarized.

Fostering DER integration in the electricity markets

6.2.1.1.1 DSOs, TSOs, and cooperation between DSOs and TSOs The DSOs have to cost-efficiently integrate new electricity generation and new loads. Article 32 in [19] establishes that the Member States should allow and provide incentives to DSOs to procure flexibility services, including congestion management in their areas. In particular, the DSOs should be able to procure such services from providers of distributed generation (DG), demand response (DR) or energy storage. DSOs shall procure such services under transparent, nondiscriminatory and market-based procedures unless the regulatory authorities have established that the procurement of such services is not economically efficient or that such procurement would lead to severe market distortions or higher congestion. The DSOs shall be adequately remunerated for the procurement of such services. According to article 57 of [18], the DSOs and TSOs shall cooperate in planning and operating their networks. Moreover, they will cooperate to achieve coordinated access to the DERs that may support the particular needs of the DSOs and TSOs. 6.2.1.1.2 RES integration The Article 12 in [18] establishes that SOs shall give priority to generating installations using RES to the extent permitted by the secure operation of the national electricity system, based on transparent and nondiscriminatory criteria and where such powergenerating facilities have an installed electricity capacity of less than 400 kW. 6.2.1.1.3 Active consumers Article 15 in [19] establishes that end-users are entitled to act as active customers, to sell self-generated electricity, to participate in flexibility provision and efficiency schemes, directly or through aggregation. Simultaneously, they are financially responsible for their imbalances. All types of customers (industrial, commercial, and households) should have access to the electricity markets to trade their flexibility and self-generated electricity and to participate in all forms of DR. Additionally, the active consumers who own an energy storage facility shall not be subject to any double fees, including network fees for remaining energy stored within their premises or when providing flexibility services to TSOs/DSOs and they shall be allowed to provide stack flexibility services, if technically feasible. 6.2.1.1.4 Aggregation Article 17 in [19] determines that the Member States shall allow all end customers to participate in all electricity markets. TSOs and DSOs shall treat market participants engaged in the aggregation of DR in a nondiscriminatory manner alongside producers. Customers should be allowed to make full use of the advantages of aggregation over larger regions and benefit from cross-border competition. The aggregation model to be implemented could include choosing market-based or regulatory principles that

183

184

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

provide solutions to comply with the Directive. Flexibility products should be defined on all electricity markets, including the markets for procuring flexibility services and capacity markets, to encourage the participation of DR. 6.2.1.2 European green deal Just three years after the publication of the Clean Energy Package, the EU published its roadmap for a complete energy transition in Europe by 2050, the European Green Deal [21]. The European Green Deal is a new growth strategy that aims to transform the EU into a society with a modern, resource-efficient, and competitive economy where there are no net emissions of greenhouse gases in 2050. Fig. 6.3 illustrates the elements that are under consideration within the Green Deal, under the principle of mobilizing research and fostering innovation leaving no one behind. The Commission expects to adopt a new and more ambitious EU strategy on adaptation to climate change. One of the main critical issues to reach climate objectives in 2030 and 2050 is to accomplish a further decarbonizing of the energy system. Member States presented their revised energy and climate plans by the end of 2019. In line with the Regulation on the Governance of the Energy Union and Climate Action, these plans should set out their ambitious national contributions to EU-wide targets. An increased cross-border and regional cooperation will help in achieving the benefits of the clean energy transition at an affordable price. Furthermore, the clean energy

Figure 6.3 Overview of the European green deal (based on [21]).

Fostering DER integration in the electricity markets

transition should involve and benefit consumers and, for that, the RES will have an essential role. The European Green Deal also outlines investments in smart infrastructure needed and financing tools on how to ensure a fair and inclusive transition [21]. 6.2.1.3 Electricity network codes and guidelines Europe’s cross-border electricity networks are operated according to rules that help in governing the work of TSOs and determine how the access to electricity is given to users across the EU. The EU-wide rules effectively manage the electricity flows between areas in the internal energy market (IEM). This way, ENTSO-e with the support of the Agency for the Cooperation of Energy Regulators (ACER) developed a set of network codes to facilitate the harmonization, integration, and efficiency of the European electricity market. Network codes are divided into Connection, Operation, and Market clusters [22]. Regarding the market family, three main guidelines have been published to set down these rules: • Capacity allocation and congestion management (Commission Regulation (EU) 2015/1222 of 24 July 2015). • Forward capacity allocation (Commission Regulation (EU) 2016/1719 of 26 September 2016). • Electricity Balancing (Commission Regulation (EU) 2017/2195 of 23 November 2017). The EBGL, which entered into force in December 2017, defines the framework for common technical, operational, and market rules for a cross border balancing market across Europe. To guarantee a well-functioning balancing market, it is essential the harmonization of the balancing products and close cooperation of the TSOs on regional and European level. For this purpose, five different initiatives (described in [23] and summarized below) have been developed to define the methodologies for the implementation of common European platforms for the different balancing services: Trans European Reserves Exchange (TERRE)

Manually Activated Reserves Initiative (MARI)

Platform description RR Platform to set up the European RR balancing energy market. The RR Platform will enable the exchange and optimized activation of a standard product for balancing. European manual Frequency Restoration Reserve (mFRR) Platform, where the standard product of the mFRR Platform is defined by the “standard bid characteristics,” the “variable bid characteristics” and additional “bid characteristics” defined in the Terms & Conditions (T&C) for the balancing service providers (BSPs).

185

186

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Platform for the International Coordination European Automatic Frequency Restoration of Automated Frequency Restoration and Reserve Platform (aFRR). PICASSO tries to Stable System Operation (PICASSO) propose a suitable market model for the common aFRR market. International Grid Control Cooperation European Platform for the Imbalance Netting (IGCC) process (IN Platform), as the process agreed between TSOs that allows avoiding the simultaneous activation of FRR in opposite directions. Platform for Frequency Containment Platform for procurement and exchange of FCR Reserves (FCR) to integrate balancing markets.

6.2.2 Current status of DERs as flexibility providers in several European countries The possibilities for DER to become flexibility providers vary among the European countries due to regulations, so, many projects are currently analyzing these differences and the possible ways the situation can evolve. For example, the CoordiNet project [24] carried out an exhaustive survey to assess the current status of the provision of flexibility by the DERs. Table 6.1 shows whether the countries allow the participation of the DERs in the provision of flexibility services, and if so, which is the service they can provide. As can be seen, DERs are already providing flexibility in many countries but there are others, like Greece, in which no service can be provided by DERs yet. However, this is changing as the Network Codes are being implemented. Although in countries like Spain, Sweden, Germany, Italy, Netherlands, and Belgium, DERs are already providing system services, not all types of DER can participate. Table 6.2 shows the DERs that are allowed per country. While countries like Sweden, Germany, The Netherlands and Belgium allow the participation of any type of DER in the provision of system services, other ones, like Italy and Spain define several restrictions to the participation of some of them. Nevertheless, countries like The Netherlands note that the possibility for different types of DER may vary depending on the service to be provided. Table 6.1 Allowed participation of DERs in the provision of system services. Service/country

Spain

Greece

Frequency control Voltage control Congestion management

Yes Yes Yes

No No No

a

Voltage control is not a remunerated service in Sweden.

Sweden

Germany

Italy

Netherlands

Belgium

Yes

Yes Yes Yes

Yes No No

Yes Yes Yes

Yes No No

a

Yes

Fostering DER integration in the electricity markets

Table 6.2 Types of DER allowed in the provision of system services (information from [25]). DER/country

Spain

Greece

Sweden

Germany

Italy

Netherlands

Belgium

Demand response Prosumers Distributed generation only Storage

No




Yes

Yes

No

Yes

Yes

No Yes





Yes Yes

Yes Yes

Yes No

Yes Yes

Yes Yes

No




Yes

Yes

Yes

Yes

Yes

6.2.3 Barriers to market access of DERs As shown in Section 2.2, many countries allow the provision of flexibility by DERs, while others do not. Although the Clean Energy Package states that the regulatory framework should incentivize and compensate expenses with the procurement of flexibility, the national regulations are still being adapted. The Clean Energy Package recommends that, to the extent possible, procurement of services by TSOs and DSOs should be market-based. This is still a barrier for many products and services, especially at the DSO side. Additionally, it is important to note that the role of the aggregator and its responsibilities are still under development. Therefore, this lack of specification becomes a barrier to the deployment of aggregators. Most common barriers that the DERs have to cope with when accessing to the current structure of electricity markets have been identified and categorized according to the three objectives established within the Clean Energy Package in [25]: • Optimal utilization of resources: In most countries, DSOs do not use flexibility from DER yet, while the provision to the TSOs still is limited by the size and type of DER. Besides, DSOs are not usually incentivized to procure services for the management of the system, which is a fundamental condition for them to procure flexibility services. • Secure and Efficient Operation: Coordination between TSOs and DSOs is essential as DSOs start to use DER flexibility. Additionally, the activation of DERs by the TSOs might create constraints, so much data must be exchanged and coordinated among both system operators. • Facilitation of market development: Aggregation is not deployed yet and the rules are unclear. Product definitions and market mechanisms still need to be developed. Other relevant barriers are also identified for the nondiscriminatory participation of DER in the markets where system services are traded, such as technological limitations, minimum size requirements that allow only the access to big power plants, the exclusion of several types of DERs when the access should be allowed for all the resources that could potentially provide those services, etc.

187

188

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

6.3 Flexibility needs in power systems The expected massive deployment of DERs in the future power systems will force them to partially supply the reserves that are needed in the grid. This section revises the current methods used by the responsible parties to establish the frequency and voltage control reserves nowadays (the TSOs) as well as a quantitative assessment on reserves’ sizing.

6.3.1 Current practices in the estimation of flexibility requirements 6.3.1.1 Frequency control (balancing) reserves The ultimate cause of frequency changes in electrical power systems is the imbalance between generation and load that can be produced by several reasons: small but fast fluctuations in generation or consumption, stochastic errors coming from the forecasts of RES or the demand, imbalances between the actual production and consumption curves and the stepwise results of the market-based balancing schedule or unexpected disturbances created by a trip of a generation unit or a major component of the system. Different approaches can be followed to estimate the frequency reserves needed in a system to maintain the frequency within acceptable limits. These three approaches are deterministic, the probabilistic and the dynamic. The definition of these approaches as well as its uses to size the frequency reserves can be found in [26] and are summarized in Table 6.3 below. Based on these general procedures, the estimation methods for the different frequency control reserves: FCR, FRR, and RR as done nowadays are summarized below. 6.3.1.1.1 Frequency containment reserves ENTSO-E determines the required FCR needs for every synchronous area using the RI (deterministic approach). However, the responsibility of the provision of those Table 6.3 Summary of methodologies for reserves’ estimation. Deterministic

Probabilistic

Dynamic

Conservative calculation is done from a known condition, usually, the worst for the system named the Reference Incident (RI)

It uses various possible incidents as well as the stochastic and deterministic factors together with their potential impact and likeliness to occur

Used for recalculation of the deterministic and probabilistic in shorter intervals to consider the effect of fast changes due to RES generation and load

Fostering DER integration in the electricity markets

FCR reserves is shared among the different TSOs. In the Northern Europe Area (NE) the RI is the N-1 criterion, the worst incident that is likely to happen, that can be the trip of a big unit or AC line or HVDC link. This RI is volatile and is calculated regularly [27]. In NE, the RI has been selected using a deterministic method, considering the maximum expected instantaneous power deviation between generation and demand in the synchronous area. The particularization of the contribution of every country to the FCR reserves is done by introducing a coefficient affecting the total FCR. This coefficient is the ratio between the energy produced in the affected country the year before (considering importing/exporting power flows) over the energy produced (also the year before) in the whole synchronous area. Therefore, this coefficient is dynamic and has to be updated annually. In the Continental Europe area (CE), with a bigger number of generating units, lines, and loads, it is more likely that a second disturbance would occur before the first one has been completely recovered. In that case, the N-1 criterion has a higher probability of not being enough and the N-2 criterion has shown to be the best practice by applying probabilistic techniques. In the CE area, the N-1 incident is currently considered as 3000 MW. 6.3.1.1.2 Frequency restoration reserves The FRR reserves are calculated for every control area belonging to a synchronous region. The FRR rules, for example, the aFRR/mFRR ratio, have to be agreed between the TSOs in the same control block. The ENTSO-E recommendations to settle the minimum values of reserves in the CE and NE given by ENTSO-E are based on the combination of both deterministic and probabilistic methods. The deterministic method uses the RI, which has to be combined with the historical records of the TSO (for at least one year). From this approach, the minimum requirement of FRR reserves established for the UCTE area [28] is shown in Eq. (6.1): R5

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi aULmax 1 b2 2 b

ð6:1Þ

where a and b are known coefficients (10 MW and 150 MW respectively) and Lmax is the anticipated consumer load in the control block. However, by common agreements between neighboring blocks and considering probabilistic indexes, the minimum requirements for reserves can be reduced. 6.3.1.1.3 Replacement reserves The RR can be also agreed between the different TSOs belonging to the same control block but they have to be, at least, enough to replace the operation of the FCR and FRR, both upwards and downwards.

189

190

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

6.3.1.2 Voltage control reserves The objective of the voltage control is to keep the values of the voltages in the system within the limits established by the regulations. The TSOs must ensure the voltage levels during normal operation as well as an adequate response in case of an event [28]. These reserves are sized by the combination of deterministic plus dynamic approaches. For the dynamic alignment of the voltage reserve needs, it is periodically simulated the worst event that the system can suffer in real-time thus correcting the initial guess made by the deterministic procedure. According to ENTSO-E Operational Handbook [28], the recommended practice to dimension the voltage control reserves is to apply the N-1 criterion to several critical facilities. Each generator is obliged, by the TSO in its control area, to fulfill with the needed voltage control requirements, and to provide automatic response to voltage deviations. The voltage control reserves can be split into two different categories: static and dynamic (See section 1.1.3).

6.3.2 Estimation of future needs of reserves in power systems with high shares of DERs In this subsection, a review of how can be done the estimation of flexibility services (frequency/voltage control) in Europe by 2030 is evaluated. Due to the massive connection of renewable energy sources in distribution grids, these reserves’ needs will be higher as also system imbalances are expected to grow. 6.3.2.1 Frequency control reserves The FCR are calculated for each country based on the RI and the corresponding share of net generation and consumption per country with regards to the net generation and consumption of the full synchronous area. The methodology presented was developed in the SmartNet project to size the frequency reserves’ needs (FRR) and was based on the procedure followed by the Belgian TSO, ELIA, according to ENTSO-E guidelines [29] which combines the probabilistic and deterministic approaches. The calculation of the total frequency control reserves considers the main sources of uncertainty according to Eq. (6.2): TR 5 GLmax 1 Derr 1 Werr 1 PVerr 1 AR

ð6:2Þ

where GLmax is the maximum generation loss according to the RI, Derr , Werr , and PVerr are the demand deviation error, the wind forecasting error and the PV forecasting error respectively and AR is the additional reserve to cope with further deviations of forecasts and market programs as well as to consider the interconnections with neighboring areas. The demand deviation error increases the reserves’ needs by a

Fostering DER integration in the electricity markets

percentage of the forecasted demand (e.g., 2% in Spain) and the wind and PV deviation errors consider an increase in the reserves to cover a percentage of wind/solar forecasting error (e.g., 99%). The sum of the maximum generation loss plus the load forecasting error plus the additional reserve AR represents the system imbalance (SI). The percentile used is 99% as the reserves should cover the imbalances 99% of the time. After this total frequency reserves estimation, they need to be split into different products (FRR/RR). The amount of aFRR is based on the variability of the residual quarter-hourly imbalances in the system in the following quarter hours. Instead of considering the total deviation errors, as it is done to calculate the total reserves, the differences of sequential deviation errors in each quarter of an hour are used, as can be seen in Fig. 6.4. The mFRR is thus the difference between the FRR and the aFRR. The proposed methodology was tested by estimating the currently required reserves and considering the consistency with the data published by the corresponding TSOs. To extrapolate the results to 2030 they were considered as assumptions a constant ratio or reserves over the total capacity of the system as well as constant system imbalances. Historical forecast data is needed for the calculation of the generation and demand forecasting errors, the imbalance of the system and the 2030 forecast, the installed capacity and forecasted capacity as well as wind and PV capacities. As some data was only available with hourly granularity, the impact on aFRR has to be adjusted. For this purpose, data from Germany was analyzed for 2015 and 2016. The average deviation for aFRR when using hourly data instead of quarter-hourly data was around 40%, meaning that hourly data gives 40% fewer reserves than quarter-hourly data. However, results are still limited as these differences will not be the same for each country. The application of this methodology to reference countries (Denmark, Italy, and Spain) can be seen in Table 6.4. In the

Figure 6.4 aFRR assumption (based on [29]).

191

192

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 6.4 Estimation of FRR reserves for 2030 for Spain, Denmark, and Italy. Spain

Upwards (MW) Downwards (MW)

Denmark (DK1)

Italy

aFRR

mFRR

aFRR

mFRR

aFRR

mFRR

783 669

1523 1028

262 257

426 334

1470 1414

3191 5473

case of Denmark, it has been only calculated for West Denmark (DK1), that belongs to the CE synchronous area.

6.3.2.2 Voltage control reserves The fundamental shift from bulk power generation and transmission system into a system with high levels of embedded RES may increase the range of reactive power control that is required to maintain system voltages within limits. The voltage control reserves are largely dependent on the location and the grid topology and thus, the estimation of these reserves is complicated usually because it exists an important lack of data. In future scenarios, we can have an increased need for localized reserves. In the same way, as FRR reserves are currently separated between control areas to better control the power flows, some new reserves could be beneficial to help to alleviate congestion and to provide voltage control at the distribution level. As an example, the reactive power control reserves an estimative estimation of reactive power needed in the German grid by 2030 can be seen in [15]. In this study, the need for reactive power was calculated by comparing the result of hourly optimal power flow for a year without controllable devices (only demand) and with controllable devices and an extra demand to be covered by them, identifying the need to increase the reactive power control reserves. In the same study, the short circuit power development trend in the German electricity grid was calculated based on an aggregated European transmission grid model and a showcase 110 kV distribution grid. Based on them, the short circuit power in Germany will increase by 20% on average by 2033. Minimum and maximum short circuit capacity levels will not be exceeded. The main cause for the reserve needs in distribution compared to transmission would be the congestion created by high or low levels of production by DER units, combined with variable levels of local consumption. Depending on the local regulations and the market prices, there could be situations where contracting local reserves could be used instead of the curtailment of DER for solving the congestion issues. These reserves could be voltage-controlled or activated by tripping signals in case the power flows in the transformers reach specific limits.

Fostering DER integration in the electricity markets

6.4 The market value of flexibility in the distribution system This section will analyze, for the flexibility products traded in the market (Section 1), subject to market regulations (Section 2) and according to the needs of procurement calculated by the DSOs (Section 3), the benefits obtained and recovered depending on the market prices and, from a higher-level perspective, the benefits for the different stakeholders along the full value chain.

6.4.1 Flexibility market beneficiaries Flexibility can add value throughout the entire electricity supply value chain. Generators can sell their electricity when the system needs it most, consumers can get better prices for the electricity they consume, and storage systems can take advantage of their flexibility to inject and take energy from the grid to pay off their investments as soon as possible. In the case of small resources, aggregators will also put their intelligence at the service of these units to exploit to the maximum the opportunities that appear in the different markets (markets for capacity, energy or system services). Some of the potential sources of incomes for flexibility providers are load shifting to avoid high electricity prices, temporary power limitation to facilitate grid congestion management, peak shaving to reduce the power term in the electricity bill, active or reactive power control to provide system services, creation, and management of local energy communities, CO2 trading, arbitrage between day-ahead/intraday and balancing markets, etc. However, to be able to make use of this flexibility potential, TSOs and DSOs must have an appropriate regulatory framework, which, on the one hand, provides them the right incentives to fully exploit the available flexibility and, on the other, defines the best way for them to coordinate with each other. TSO-DSO coordination is needed to ensure optimal use of the inherent flexibility of DER and, thus, the market design must allow close collaboration between both system operators. A coordination scheme (CS) is defined as “the relation between TSO and DSO, defining the roles and responsibilities of each system operator when procuring and using system services provided by the distribution grid [30].” According to this definition, a CS highlights two important ingredients for increased coordination: (1) the assignment of responsibilities to and the interaction between system operators, (2) the focus on specific market phases (for instance procurement, prequalification) and how this market phase should be organized through a proper market design [7]. Since there is no single solution that fulfils the needs of all power systems (due to local circumstances, market maturity or regulatory conditions) there are many possible TSO-DSO CSs. Besides, certain features of the CS may have to mature, so it is convenient to present several alternatives, with a flexible categorization structure.

193

194

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

6.4.2 Cost-benefit analysis of market participation of DERs When selecting the most suitable CS to be implemented, the analysis of them from an economic perspective can be of utmost importance. A cost-benefit analysis (CBA) is an analytical method for evaluating the costs and the benefits of a decision to determine the economic advantages and disadvantages of implementing that decision and to assess whether its benefits outweigh the costs or not. The best way to identify such advantages/disadvantages is to use a set of metrics and to assign a weight to each of them, or converting all metrics into a monetary unit and, then, there is no need to assign weights as such, but just converting all the metrics into money. The selection and definition of those metrics are critical because they must fulfill several characteristics to ensure that they represent the complexity of the analysis. An example of a detailed CBA performed in the SmartNet project can be found in [31], where four different TSO-DSO CSs and different real-time market architectures were compared, to find out which one could deliver the best compromise between costs and benefits for the system. These CSs to perform the CBA analysis were: • Centralized ancillary services market model (CS A): Single market operated by the TSO with resources connected at transmission and distribution levels, extending its balancing service market with locational information. • Local ancillary services market model (CS B): The DSO operates a local market and the TSO has access to the DER through this local market but the DSO has priority. • Shared balancing responsibility model (CS C): The TSO is only responsible for the balancing resources at transmission, transferring the balancing responsibility for the distribution grid to the DSO. • Common TSO-DSO ancillary services market model (CS D). TSO and DSO are jointly responsible for the operation of the common market with both resources connected at the transmission and the distribution levels. A comparison between the benefits drawn by the system with the costs needed to implement each TSO-DSO CS was performed based on the following metrics: • Total mFRR cost: This metric included the total balancing cost of the market. The energy activated was remunerated at the nodal price resulting from the clearing process. The mFRR activations in the balancing market were aimed to solve the network imbalance and to avoid congestions predicted in advance for the next time step. • Total aFRR cost: This was the cost of re-balancing the system after the mFRR market. In this case, the bids submitted to the market were ordered according to a system-wide merit order and the resulting price was applied as the marginal price.

Fostering DER integration in the electricity markets

• Cost of unwanted measures or cost of re-dispatching (UM): This was the cost of emergency actions taken by network operators caused by unpredicted network congestions. These measures were activated on available flexibility and they were valued at the correspondent bid price. • Information and Communication Technologies (ICT) costs: The term ICT cost included the software for the aggregation and market clearing process. Only those ICT costs that were directly related to the implementation of each CS were considered. Communication costs were assumed to be very similar in all Css and, therefore, differences came from the pieces of software needed for aggregation and market clearing. To represent the conditions envisaged in different ENTSO-e visions [13], three European countries were selected (Denmark, Italy, and Spain) and a plausible scenario for 2030 was created for each of them. Then, the costs and benefits of the different CSs were assessed for each of them and compared against a baseline, which was assumed to be the centralized ancillary services market model. Metrics were elaborated and applied for comparing the TSO-DSO interaction schemes for each national case and, as result, the different schemes were independently scored for each country, so that the most convenient architecture, which is different for each national case, could be identified. In addition to the metrics monetized, the CO2 emissions were also investigated as a complementary indicator, because they can provide extra information for the CBA. However, CO2 emissions were not monetized because the bids sent to the mFRR market already included an estimation of the cost of the CO2 ton. Results for the Spanish CBA analysis and the cost breakout is shown in Fig. 6.5.

Figure 6.5 CBA breakdown costs CS comparison for Spain.

195

196

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Several significant general conclusions were drawn from the analysis and its application to the reference countries: 1. The effectiveness of the TSO-DSO CS depends on the level of services requested by the DSO: a. In the case of few congestions at the distribution level, the centralized market model has higher economic performance concerning the local and common market models. This issue was observed in Denmark, where the uncertain contribution of wind power (mostly located at transmission-level) is higher than the contribution from PV (at distribution level). Therefore, forecasting errors are expected to be higher, while the congestions at the distribution level are likely to be less common. b. When distribution congestions were significant, the adoption of the local and common market models resulted to be beneficial. This happened in the case of Italy, where generation located in distribution grids, such as PV, contributes more to the electricity supply and, hence, congestions are expected to happen more frequently. c. In any case, the most relevant cost component was the mFRR cost in all the cases, while the unwanted measures and the ICT costs accounted for a small share. Moreover, results discouraged the investigation of new methods for dispatching the aFRR because higher complexity would result in higher IT costs. 2. The implementation of two-step markets (the local and shared balancing responsibility market models) is generally less efficient than optimizing in a single step (common market model): a. Regarding the local market model, the results are pretty similar to the ones returned by the common TSO-DSO market model, although slightly more expensive in the scenarios simulated. This is something to be expected because the common market obtains an overall optimum, while the local market model solves distribution and transmission services separately. b. On the contrary, the shared balancing responsibility model is the least efficient CS in all the countries. c. Both the local ancillary services market and shared balancing responsibility mode may suffer from scarcity and/or illiquidity of resources, which would further decrease their efficiency. d. However, in rare circumstances (i.e., severe congestions at transmission-level) the selection of two-step market architectures can be more beneficial than other schemes, as market separation potentially prevents the spreading of high nodal prices among distribution and transmission systems. 3. The main finding regarding the ICT cost was that the technological costs in upgrading market architectures from the centralized market model to the local, shared balancing responsibility and common market models are almost the same (subject to uncertainties) and much lower than operating costs. The variation between countries is minor.

Fostering DER integration in the electricity markets

4. In addition to these general conclusions, some country-specific deductions could be extracted: a. In Italy, where there are big congestions at the distribution level, the upgrade from the centralized to the local or common market models is convenient and not jeopardized by ICT costs. b. In Spain, with average congestions at the distribution level, the ICT costs are comparable to the benefits brought by adopting the local or common market models instead of maintaining the centralized one. c. In Denmark, with low congestions at the distribution level and high forecasting errors, the implementation of DSO-inclusive CSs failed. 5. The aggregators will bear a large portion of ICT costs (communications with DERs, aggregation software, etc.). Therefore, the related cost analysis indicates that the aggregator’s IT systems will be the most expensive ones. It may be possible that DER communication/activation costs turn out to be too large for a profitable aggregation business, but this issue applies to all CSs. 6. Regarding the CO2 emissions, the main conclusions achieved by the analysis were that the differences among countries depend on their energy mixes. CO2 emissions in Italy were the highest ones since Denmark has small demand and mostly renewable energy mix and the energy mix in Spain is also predominantly carbon-free (nuclear and large portions of renewables). However, the differences among CSs were less than 7%. Because of the obtained outcomes during the project, and taking into account the scenarios analyzed within it, the adoption of the shared balancing responsibility market model seemed to be the least efficient one and, therefore, technical reasons could advise continuing centralizing balancing responsibility to TSOs. However, depending on the impact of the congestions at the distribution level, the congestion management responsibility could be shared between TSO and DSOs, as it is already addressed by the Clean Energy Package. More advanced CSs incorporating distribution constraints showed higher economic performances, but their performance could be undermined by big forecasting errors. Hence, it is of paramount importance to improve the forecasting techniques, to increase the marketclearing frequency and to shift the gate closure as much as possible towards real-time. The local congestion markets should have a “reasonable” size and guarantee a free competition by enough actors, to prevent scarcity of liquidity and the power exercised by the local markets. This may create the need for small DSOs to pool-up to ensure the required market size. In a more than likely scenario in which the fit-and-forget reinforcement remuneration approach is abandoned and the forecasting errors are more accurately calculated the common TSO-DSO market model could be the most efficient of the CSs analyzed. The appropriate allocation of costs and benefits among the actors is of the utmost importance when selecting the CS, since this issue may strengthen or threaten its deployment. Therefore, an appropriate business model analysis must be performed.

197

198

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

6.5 Local energy markets This subsection analyses the need for developing local energy markets aligned with the European Directive for the engagement of customers. Linked to those markets, new roles will lead the energy trading mechanisms, such as the aggregators. Different approaches to flexibility management and relationships between different stakeholders involved will be reviewed.

6.5.1 Local energy markets To fulfill the targets set out by the Paris Agreement, renewable energy would need to make up at least 86% of the new electricity supply by 2050 [32]. To reach this large share of renewable energy and improve global energy efficiency, the EU is promoting local energy markets (LEMs) developing end-customer focused policies. An LEM can be defined as a series of trading actions where participants can be aggregated to supply flexibility services in the short-or long-term for a particular geographical location, DSO, TSO, and electrical network [33]. Hence, by introducing a large quantity of DERs within the power system, flexibility services like system balancing, congestions management, and portfolio optimization can be envisaged [34]. LEMs must integrate with existing electricity markets, avoiding overlapping, and defining local balance responsibility. These markets should send price signals to all participants, according to the status of the grid, increasing the price if there is a shortage of energy or reducing it, if there is an imbalance between generation and consumption, producing a surplus. However, DSOs need to define a clear flexibility portfolio with adequate procurement, measurement, and settlement procedures in which all participants are treated equally [33]. Based on the Universal Smart Energy Framework (USEF) report [35], there are four potential flexibility customers: prosumers, Balance Responsible Parties (BRP), DSOs, and TSOs. DSOs and TSOs are interested in acquiring the flexibility services to maintain in a safe condition network capacity and stability. BRP look for reducing its operational cost through its flexibility portfolio and financially regulate any imbalance that arises. Lastly, prosumers would employ their flexibility capabilities to reduce the electricity bill by combining their local generation and storage capacity, which indirectly reduces network stress. For example, in [36], the role of batteries and how it is affected by market design rules (e.g., prices and peer-to-peer trade) is investigated employing two market approaches called Flexi User and Pool Hub. The first analyses the ability of local storage capacity to sell surplus energy from renewable sources to peers, whereas the last, does the analysis at the community level. These peer-topeer mechanisms have the advantage of not requiring a central entity that manages the DERs, but this reduces the possibility to sell flexibility services on a large scale to TSOs or DSOs. Moreover, smaller market players would not have access to wholesale markets. To overcome this barrier, the peer-to-platform approach [34] can be used to

Fostering DER integration in the electricity markets

manage local problems through the aggregator in a centralized fashion, who would supervise the end-users in terms of energy production and consumption, clearing, and contract fulfilment. In [37], a decision-making problem is proposed to schedule flexible energy resources based on the requirements of the DSO using a new aggregator type called Smart Energy Service Provider (SESP). On the other side, for large scale emerging projects, in [38], the authors analyzed four recent private projects between several European utilities (Piclo Flex, Enera, GOPACS, and NODES) based on flexibility markets to obtain an overview of their operational framework, which is summarized in six yes or no questions. These six controversies look for answering if: (1) a particular energy flexible market follows the current order of EU electricity markets, (2) there is a third party operating such market, (3) exist a reservation payment, (4) there is a common flexible portfolio, there is a cooperation framework between (5) TSO
DSO and (6) distribution system operators. Moreover, since a large number of stakeholders in LEMs use the power network to supply energy and flexibility services, and some existing approaches for LEMs design ignore its physical characteristics, power flow, and operational constraints (e.g., voltage and thermal limits), it is necessary to consider these conditions for avoiding unreasonable market-clearing results.

6.5.2 Roles in a local energy market Rivero et al. [39] define a role as an intended behavior of a precise market party with certain responsibilities and unique properties that cannot be shared. In this sense, as the LEMs are envisaged to provide new services supported at a given time for a given span from a specific location within the network, it is necessary to define new roles in the market framework that can operate alongside existing ones. However, this does not mean that the participants’ positions in the LEM could be overlapped due to the nature of their portfolios [40]. This means, for instance, that the LEM operator’s role can be taken over by the DSO or the Aggregator, and an Aggregator can also act as its own BRP. The role and responsibilities of the involved agents in a LEM are described below. The scheme in Fig. 6.6 exemplifies how the agents in a LEM can interact. 6.5.2.1 Prosumers A prosumer can be considered as an end-user that utilizes all types of devices that either demand or produce energy and can be actively managed. Hence, every device can respond to price and other signals from a third-party entity (e.g., the Aggregator) and provide flexibility to the LEMs via such agent. However, the control space of this agent is limited to the prosumer’s preferences over its assets.

199

200

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 6.6 LEM scheme and interaction.

6.5.2.2 Aggregator An aggregator is a grouping of agents in a power system, which acts as a single entity when engaging in the electricity market. The Aggregator’s role is to gather flexibility from the Prosumer’s devices to sell it to the DSO, the BRP, directly or through the BRP to the TSO. Thus, the Aggregator seeks to maximize its profit by supplying that flexibility to reduce grid congestions, deferring the need for network reinforcements, limit any penalties due to balance supply, and buy energy when prices are low. Moreover, it is in charge of supervising its customers in terms of power production and consumption, settlements, and contract fulfillment. Through this entity, the endusers are less exposed to the risks involved in participating in the LEMs [33]. 6.5.2.3 Supplier Retailers are existing commercial entities purchasing electrical energy from their associated BRP or directly from the market for its clients [41]. Both the retailer and its customers arrange the commercial terms for the supply and procurement of energy. Furthermore, this entity can reduce its balancing costs by optimizing its portfolio. 6.5.2.4 Balance responsible parties (BRP) BRPs are defined as traders in the energy exchange market for representing the members of its portfolio [33]. This entity is responsible for maintaining the energy supplyand-demand balance of its customers over a given time, and it could be subject to be penalized for imbalance cost if it does not accomplish this target. Therefore, the BRP could contract flexibility to optimize its portfolio to match its energy commitments.

Fostering DER integration in the electricity markets

However, it is particularly interested in the amount of flexibility, no matter the location of the given flexibility in a specific distribution network [40]. Given its nature, the BRP can be represented by retailers, generators, or Aggregators. 6.5.2.5 DSO The DSO is responsible for maintaining the secure operation and energy quality of the distribution network cost-effectively. The DSO can obtain flexibility for different operational purposes (e.g., congestion management, voltage control, loss minimization, and so on) and planning purposes (e.g., the network investment deferral). For example, a DSO would prefer to observe a flat load profile on its networks, to employ the available network capacity maximally, thus minimizing the investments in its assets. 6.5.2.6 TSO The role of the TSO is to transport large energy blocks at a high voltage level from a centralized generator to distributed Prosumers and DSOs in a region. The TSO is responsible for maintaining power balance by using regulating capacity, reserve capacity, and incidental emergency capacity. Due to the increase in renewable sources in the energy matrix, there is also a growth in the demand for flexibility and capacity. Therefore, flexibility services provided by the LEC could help to obtain a more economical operation of the TSO’s tasks.

6.5.3 Components of functional local energy markets Flexibility can be enabled by a local market platform where the stakeholders can trade flexibility capacity and energy in the distribution grid level. LEMs mechanism can be implemented in different ways, allowing suppliers and consumers to trade at energy at variable prices, taking into account the particular constraints produced by this local context. To develop a functional market, it is important to define the following components: • Bids formats: There are two different options, open or sealed bids that is whether the bids are visible by one or more sides of auction during its process. Open bids allow adjusting offers for future bids, facilitating the flexibility procurers to bid if they first know the number of flexible resources available for different time slots. Additionally, all transactions are visible to DSOs therefore; they can carry out technical constraints’ validation. The main drawbacks are the uncertainty of the bids and some privacy challenges [42]. In sealed bids, suppliers offer supply schedules and the market supply curve is generated, with a lack of other bids knowledge. The market-clearing price is determined at the price where supply is equal to demand. Sealed bids provide the same price/benefit for every agent when the price is formed at the end of the clearing mechanism.

201

202

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

Bidding type: Again, there are two types available, one or two-sided bidding. English dynamic increasing outcry of Dutch dynamic decreasing one, are examples of one-sided bidding. In the first outcry, flexibility is sold to the highest bidder with bids taking place in ascending order (with an initial reserve price to be met). In the second one, bidding starts at a high price and it is progressively lowered until one buyer accepts it. One of the primary advantages of this bidding type is speed because the bidding process takes less time. On the contrary, in a continuous double auction, all agents bid and the matches are performed continuously, based on the current distance between asks and bids. This second option is more complicated to be developed, but it is more efficient in the price formation and it is more participative, promoting the empowering of end-users [42]. • Clearing rule: It determines the final price, using a single iteration, which is simpler, or involving different iterations, allowing achieving better commitments among all involved agents in the market. • Matching can be done in fixed time slots, allowing to connect this LEM to the wholesale market or continuously, allowing to obtain more accurate forecasting of demand and supply, enabling more efficient use of the distributed variable renewable energies. • Price formation can be determined by marginal pricing (MP) or pay-as-bid pricing (PABP). Once the market-clearing price is determined (when supply is equal to demand), all the submitted bids at a price lower than (or equal to) the marketclearing price are accepted and paid. Under the MP scheme, accepted suppliers are paid by the market marginal price (which is the last accepted price-offer) and all accepted suppliers receive the same price. In PABP, the suppliers are paid their bids they have offered. • Market design can be centralized, where buying and selling agents present their offer to a central market, determining the market-clearing price for all agents, or peerto-peer (P2P) market, where buying and selling agents trade between each other with orders made on a PABP basis individually and anonymous, using blockchain smart contracts. Finally, sharing information required to perform LEM is also a critical issue, affecting the level of communication required, from no communication among market agents to one-way or two-way communications.

6.6 Conclusions The expected increase in the installation of DERs at the distribution levels together with the replacement of larger conventional units at transmission are forcing the electricity markets to evolve. The DERs development has as main advantages the reductions in CO2 emissions (as many of them rely on RES) and an increase of the

Fostering DER integration in the electricity markets

self-sufficiency of end-users. However, the DERs increase the uncertainties in the grid and can create congestions and voltage issues in the distribution grid. The DERs will, on the one hand, force the operators to increase their reserve needs in the coming years but, on the other hand, they will also become providers of flexibility services, expanding the portfolio of candidate resources to participate in the electricity markets. There are still limitations for the participation of DERs in the electricity markets due to the lack of specific regulations but also linked to the technical characteristics of each type of DERs itself, that pose restrictions to the flexibility services they can provide. Some countries have already started to allow the participation of DERs in electricity markets, alone or through aggregation. The current status of European regulations together with a country-specific analysis to know what type of DER can provide a certain flexibility service in which country has been provided in this chapter. The massive deployment of DERs is also going to change the needs of reserves shortly (2030 horizon) and a methodology to estimate these reserves has been drafted. Additionally, different coordination schemes for the provision of flexibility services have been analyzed from an economic perspective (cost-benefit analysis), highlighting the better approach (more centralized, more decentralized, different responsibility allocations) depending on the costs for the provision of the FRR reserves, the possible congestions and the deployment of the ICT infrastructure. Recently, local energy markets (LEMs) have arisen as an opportunity to trade flexibility among the different participants in an economically efficient way, giving a key role to the consumers, as promoted by last European regulations. Insights about the different roles involved and the main characteristics to design functional LEMs have been included.

References [1] EDSO, The role of DSOs in tomorrow’s electricity market, (n.d.). ,https://www.edsoforsmartgrids. eu/wp-content/uploads/public/EDSO-views-on-Flexibility-FINAL-May-5th-2014.pdf.. [2] EDSO, TSO
DSO report: an integrated approach to active system managementwith the focus on TSO
DSO coordination in congestion management and balancing (2019). ,https://www.edsoforsmartgrids.eu/wp-content/uploads/2019/04/TSO-DSO_ASM_2019_190304.pdf.. [3] K. Oureilidis, K. Malamaki, K. Gallos, et al., Ancillary services market design in distribution networks: review and identification of barriers, Energies 13 (2020) 917. [4] NREL, An introduction to grid services: concepts, technical requirements, and provision from wind, (n.d.). ,https://www.nrel.gov/docs/fy19osti/72578.pdf.. [5] European Commision, Commission regulation (EU) 2017/2195 of 23 November 2017 establishing a guideline on electricity balancing, (n.d.). [6] European Commission, Regulation (EC) No 714/2009 of the European Parliament and of the council of 13 July 2009 on conditions for access to the network for cross-border exchanges in electricity and repealing regulation (EC) No 1228/2003, (n.d.). [7] A. Delnooz, J. Vanschoenwinkel, E. Rivero, C. Madina, CoordiNet deliverable 1.3—definition of scenarios and products for the demonstration campaigns., (n.d.).

203

204

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[8] Florence School of Regulation, The EU electricity network codes (ed.), (2019). ,http://fsr.eui.eu/ publications/the-eu-electricity-network-codes-2019-ed/.. [9] R. D’hulst, J. Merino, E. Rikos, D. Kolodziej, K. Heussen, D. Geibelk, et al., Voltage and frequency control for future power systems: the ELECTRA IRP proposal, in: 2015 International Symposium on Smart Electric Distribution Systems and Technologies (EDST), IEEE, Vienna, Austria, (2015) 245
250. ,http://ieeexplore.ieee.org/document/7315215/. (accessed 02.07.2020). [10] ENTSO-E website, (n.d.). ,https://www.entsoe.eu/.. [11] T. Leiskamo, Definition of flexibility products for multilateral electricity markets (MSc. Thesis), (2019). [12] E. Hillberg, A. Zegers, B. Herndler, et al., Flexibility needs in the future power system, (2019). ,http://rgdoi.net/10.13140/RG.2.2.22580.71047. (accessed 02.07.2020). [13] J. Le Baut, G. Leclerq, G. Vigano, et al., SmartNet Deliverable 1.2—characterization of flexibility resources and distribution networks, (2017). [14] M. Marroquin, P. Jiménez, E. Rivero, et al., CoordiNet deliverable 1.4—DER modeling, (2019). [15] Dena—German Energy Agency, System services study 2030, (2014). ,https://www.dena.de/en/ topics-projects/projects/energy-systems/dena-ancillary-services-study-2030/.. [16] M. Braun, Technological control capabilities of DER to provide future ancillary services, (n.d.). [17] European Commision, Clean energy for all Europeans, Energy—European Commission. (2017). ,https://ec.europa.eu/energy/en/topics/energy-strategy-and-energy-union/clean-energy-all-europeans. (accessed 23.02.2019). [18] Regulation (EU) 2019/943 of the European parliament and of the council—of 5 June 2019—on the internal market for electricity, (n.d.) 71. [19] Directive (EU) 2019/944 of the European parliament and of the council—of 5 June 2019—on common rules for the internal market for electricity and amending Directive 2012/ 27/ EU, (n.d.) 75. [20] Directive (EU) 2018/2001 of the European parliament and of the council—of 11 December 2018—on the promotion of the use of energy from renewable sources, (n.d.) 128. [21] European commission, Communication from the commission to the European parliament, the European council, the council, the european economic and social committee and the committee of the regions. The European Green Deal, (n.d.). ,https://ec.europa.eu/info/sites/info/files/european-green-deal-communication_en.pdf.. [22] Network Codes Home, (n.d.). ,https://www.entsoe.eu/network_codes/. (accessed 3.06.2020). [23] Electricity Balancing, (n.d.). ,https://www.entsoe.eu/network_codes/eb/. (accessed 3.06.2020). [24] CoordiNet Porject, (n.d.). ,https://coordinet-project.eu/.. [25] L. Lind, J.P. Chaves Ávila, CoordiNet deliverable 1.1—market and regulatory analysis: analysis of current market and regulatory framework in the involved areas, (n.d.). ,https://coordinet-project. eu/publications/deliverables.. [26] J. Merino, I. Gómez, E. Turienzo, C. Madina, I. Cobelo, A. Morch, et al., Ancillary service provision by RES and DSM connected at distribution level in the future power system. SmartNet Deliverable D1.1, (2016). ,http://smartnet-project.eu/wp-content/uploads/2016/12/D11_20161220_V1.0.pdf.. [27] European Commission, Commission regulation. Establishing a guideline on electricity transmission system operation. Draft version, (2015). ,https://ec.europa.eu/energy/sites/ener/files/documents/ Draft%20GLSysOP-2611_clean.pdf. (accessed 3.03.2016). [28] ENTSO-E, Continental Europe operation handbook. P1
Policy 1: load frequency control and performance., (2009). ,https://www.entsoe.eu/fileadmin/user_upload/_library/publications/entsoe/Operation_Handbook/Policy_1_final.pdf. (accessed 3.03.2016). [29] ELIA, Evolution of ancillary services needs to balance the Belgian control area towards 2018, (n.d.). ,http://www.elia.be/B/media/files/Elia/Grid-data/Balancing/Reserves-Study-2018.pdf.. [30] H. Gerard, E.I. Rivero Puente, D. Six, Coordination between transmission and distribution system operators in the electricity sector: a conceptual framework, Uti. Policy 50 (2018) 40
48. Available from: https://doi.org/10.1016/j.jup.2017.09.011. [31] I. Gómez, S. Riaño, C. Madina, M. Rossi, et al., SmartNet deliverable 4.3—Cost-benefit analysis of the selected national cases, (n.d.).

Fostering DER integration in the electricity markets

[32] International Renewable Energy Agency (IRENA), Global energy transformation: a roadmap to 2050 (2019 edition), (2019). ,https://www.irena.org/publications/2019/Apr/Global-energy-transformation-A-roadmap-to-2050-2019Edition.. [33] A. Ramos, C. De Jonghe, V. Gómez, R. Belmans, Realizing the smart grid’s potential: defining local markets for flexibility, Uti. Policy 40 (2016) 26
35. [34] P. Olivella-Rosell, P. Lloret-Gallego, Í. Munné-Collado, R. Villafafila-Robles, A. Sumper, S. Ottessen, et al., Local flexibility market design for aggregators providing multiple flexibility services at distribution network level, Energies. 11 (2018) 822. Available from: https://doi.org/10.3390/ en11040822. [35] USEF: the framework explained, (2015). ,https://www.usef.energy/download-the-framework/#.. [36] A. Lüth, J.M. Zepter, P. Crespo del Granado, R. Egging, Local electricity market designs for peerto-peer trading: the role of battery flexibility, Appl. Energy 229 (2018) 1233
1243. [37] P. Olivella-Rosell, E. Bullich-Massagué, M. Aragüés-Peñalba, A. Sumper, S.Ø. Ottesen, J.-A. Vidal-Clos, et al., Optimization problem for meeting distribution system operator requests in local flexibility markets with distributed energy resources, Appl. Energy 210 (2018) 881
895. [38] T. Schittekatte, L. Meeus, Flexibility markets: Q&A with project pioneers, Uti. Policy 63 (2020) 101017. [39] E. Rivero, D. Six, P. Mallet, M. Sebastian-Vian, J. Stromsather, M. Baron, Evolvdso: assessment of the future roles of the dsos, future market architectures and regulatory frameworks for network integration of dres, (2015). ,https://doi.org/10.5281/ZENODO.28763.. [40] X. Jin, Q. Wu, H. Jia, Local flexibility markets: literature review on concepts, models and clearing methods, Appl. Energy 261 (2020) 114387. [41] J. Villar, R. Bessa, M. Matos, Flexibility products and markets: literature review, Electr. Power Syst. Res. 154 (2018) 329
340. [42] O. Kilkki, S. Repo, G. Mendes, et al., DOMINOES—deliverable D2.3 scalable local energy market architecture (second release), (2019). ,http://dominoesproject.eu/wp-content/uploads/2020/ 01/D2.3_DOMINOES_Scalable-Local-Energy-Market-Architecture_v1.1_PU.pdf..

205

CHAPTER 7

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios M. Caruso1, P. Gallo2, M.G. Ippolito2, S. Nassuato3, N. Tomasone3, E.R. Sanseverino2, G. Sciumè2 and G. Zizzo2 1 Exalto Energy & Innovation Srl, Palermo (PA), Italy Department of Engineering, University of Palermo, Palermo (PA), Italy 3 Regalgrid Europe S.r.l., Treviso (TV), Italy 2

Abbreviations REC EMD CEC V2G V2V G2V DR P2P DSO UVAM MSD ICT DLT BLORIN BTF pBTF PoW PoS PoA SC EVM RES ESS VN PPA MAN

Renewable Energy Communities Internal Market Directive Citizens’ Energy Communities Vehicle-to-Grid vehicle-to-vehicle Grid-to-Vehicle Demand Response Peer-to-Peer Distribution System Operator Unità Virtuali Abilitate Miste Dispatching Services Market Information and Communication Technology Distributed Ledger technologies Blockchain for Renewable Energies Bizantine Fault Tolerance practical Bizantine Fault Tolerance Proof of Work Proof of Stake Proof of Authority smart contract Ethereum Virtual Machine Renewable Energy Sources Energy Storage System Virtual Node Power Purchase Agreements Metropolitan Area Network

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00009-1

r 2021 Elsevier Inc. All rights reserved.

207

208

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

7.1 Introduction The European legislator has recently supported the emerging roles of “prosumers” and “energy communities”, indicating the desire to give citizens a leading role in the ambitious process of the energy transition towards the goal of climate neutrality by 2050. The EU Directive 2018/2001 (RED-II) [1] on Renewable Energy Sources (RES) represents an important tool aimed at enhancing the decentralization of electricity production in the context of the liberalization of the energy market. In addition to raising the share of renewable energy to be fed into the network by 2030 to 32%, the Directive establishes the role of “prosumers” and “REC Renewable Energy Communities” in articles 21 and 22. This Directive is associated with that on the electricity market, the Internal Market Directive (EMD) 2019/944, which regulates the “Citizens’ Energy Communities” (CEC). Both have in common the purpose of offering environmental, economic and social benefits. In order to avoid confusion and overlaps, it has been suggested that in the transposition of the two Directives in the Italian law, it would be better to define a single type of Community that best exploits the potential of each formulation described in the Directives. As far as the energy sources are concerned, the REC solution should be preferred, because it excludes fossil sources, while CEC is a technology-neutral concept. Also, in relation to Governance, the setting of the REC opens to individuals, local authorities, small and medium-sized enterprises, and this appears preferable to that of the CEC approach in which also large companies have a role. Finally, while the REC indicates the need for a local dimension of the interventions, in the CEC there are no limits and they could virtually extend to the whole national territory. The Blockchain, born in 2008 with Bitcoin, is an Information and Communication Technology (ICT) belonging to the Distributed Ledger Technology (DLT) family and that, today, is arousing the interest of all stakeholders in the power and energy sectors. The chapter is structured as it follows: • In Section 2, a brief overview of the blockchain technology and of its main characteristics is given. • In Section 3, current trends and possible evolutions of the energy blockchain, especially in the DR and V2G fields, are presented and discussed and the most significant energy blockchain projects are briefly described. • In Section 4, a laboratory setup for testing the energy blockchain in the framework of the BLORIN (Blockchain for Renewable Energies) research project is presented. • Finally, Section 5 contains the conclusion of the chapter.

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

7.2 The blockchain technology 7.2.1 What is the blockchain The Blockchain is a technology based on DLT, hence characterized by a distributed register; in these systems, all the nodes of a network have the same copy of a dataset that can be read and modified independently from the individual nodes. While in distributed databases all nodes that have a consistent copy of the data can consult and modify it through the authorization of a central authority, in Distributed Ledger systems, changes to the register are regulated by consensus algorithms. These algorithms aim to reach a consensus between the various versions of the register. In addition to consensus algorithms, blockchains also make extensive use of cryptography to maintain the security and immutability of the registry, or the use of smart contracts to execute or validate transactions. Details on smart contracts, the different consensus algorithms and the process for achieving consensus over a block of transactions are reported, with specific reference, in the following sections. As a result, the blockchain is a DLT in which the register is structured in blocks, each block is linked to the previous one thanks to particular cryptographic functions called “Hash Functions” and a new block can be added only after the last one and after the consensus process. Hash functions convert a message of any length, in this case the data related to a block, into a message of a fixed length that uniquely identifies it. They create a string associated with the data of the reference block so that, once the function is applied, it is no longer possible to return to the source data. The peculiarity of these functions is that they are easy to calculate and always give the same result using the same input data. Blockchain technology uses these functions to create the chain of blocks: on each block, there is a hash calculated using the transaction data contained in the block itself plus the hash of the previous block; in this way, each block stores a chained reference to previous transactions on all nodes of the network. The direct consequence is that a single block becomes impossible to modify without modifying the whole chain: hence the immutability of the blockchain. Bitcoin was the first application of the blockchain technology. The latter was founded with the aim of creating an electronic payment system that allows two counterparties to negotiate directly with each other without the need for a third party of trust. The traditional electronic payments systems suffer from the inherent weaknesses of a model based on the trust of unknown users in a third party that acts as a guarantor for both. The result is an intermediation cost that increases the cost of the transaction, limiting the minimum size of transactions that can be carried out and reducing the possibility to execute small transactions. Bitcoin was created to address these problems, proposing a new technology based on cryptographic evidence instead of trust, which allows the exchange of currency between two unknown users without the need for a third party to act as a guarantor for both. Thanks to the use of the blockchain, in

209

210

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

bitcoin, transactions are recorded and ordered in time, and there is a guarantee that recorded events are difficult to delete or modify even if some malicious users try to do so since each change would require a lot of resources and/or the approval of most of the network nodes and not a single authority as in the case of classic centralized systems. For this reason, users should not rely on third parties, who often make decisions for their own interests. So this new technology is particularly useful in cases where two people want to make a deal, but do not trust each other. The main novelty introduced by the blockchain is the distribution of data and trust in the network while overcoming the problems afflicting decentralized systems. In a blockchain network, each node can see all transactions, and transactions cannot be deleted or changed once added to the ledger, thus creating a network that allows transparency between users and traceability of all transactions. The use of blockchain technology thus makes it possible to achieve: 1. Transparency, because each block added to the blockchain is accessible to all participants and is in the archive of all participants; 2. Immutability, since a block, thanks to the use of the hash functions, once added, can be modified only by modifying the whole chain or after the approval of most of the participants in the network; 3. Confidence, since there is a shared reading among all the participants; 4. Efficiency, since it does not require intermediaries compared to the classic transaction management system, thus simplifying processes, infrastructures and increasing operational efficiency; 5. Security and Control, since it allows the use of encryption allowing greater data protection and a lower risk of fraud. Thanks to the features and advantages it offers compared to classic database systems, today many companies are experimenting with the use of this new technology by developing and testing many applications in different sectors.

7.2.2 Consensus algorithms As the name “consensus” suggests, they consist of finding an agreement on the validity of something, in this case, transactions and, then, the blocks to be inserted in the blockchain. This mechanism makes it possible to eliminate the trusted third party of traditional centralized systems, so it can be said that it is the most important part of the blockchain execution. The consensus mechanism is executed through consensus algorithms, which are more or less complicated and can be based on different principles such as solving a puzzle or a simple voting mechanism. Consensus algorithms must ensure the so-called Byzantine Fault Tolerance (BTF), so they must allow the distributed system to work even if some nodes fail or act dishonestly. The complexity of the algorithm depends on whether the blockchain is permissioned or permissionless.

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

Usually, these algorithms are more computationally-intensive in the permissionless blockchain, as users are unknown and therefore the consensus cannot even take into account neither the user’s identity nor the validity of transactions sent by them. While, they are simpler in the permissioned blockchain, as users are known and there is usually a need to verify that a user is malicious. The most famous consensus mechanism that ensures the BTF is the Proof of Work (PoW), used in Bitcoin [2], where the user who adds a new block on the chain is the one who solves a computationally-intensive problem before the others. In Bitcoin, users competing for the insertion of new blocks are called “miners”, because the system, as a result of the effort made to solve the algorithm, pays the user a fixed number of new bitcoins. Being a permissionless blockchain, in Bitcoin anyone can be a network node and can theoretically run the consensus algorithm and compete for new blocks insertion, earning both the bitcoins issued by the system and the fee that users pay to miners to have their transaction inserted into a new block. The higher the computational power, the higher the probability of validating a block, but once the algorithm is resolved and a new valid block is found, it must be approved by 51% of the nodes of the entire network, otherwise the block will be rejected and the miner will not be remunerated. In addition, as the network grows in size, to maintain high security, the power consumption needed to perform the consensus operations also increases. Theoretically, a malicious user could insert a false transaction within a block and validate it, but the block would not be approved by the other nodes and therefore discarded. Due to the high energy consumption for the execution of the algorithm, to date about 570 kWh per transaction [3], and the impossibility to own 51% of the network nodes, a malicious user can’t insert a transaction that is not true. Therefore, thanks to this mechanism, not only are malicious users eliminated from the system but also miners are forced to validate the blocks correctly, otherwise they would not be remunerated and the mechanism would become economically unsustainable due to the high costs in terms of energy consumption. An alternative way to PoW to reach consensus, requiring much less computational effort and this energy consumption, is the Proof of Stake (PoS) consensus. Unlike PoW, where the algorithm rewards miners who solve the algorithm with the aim of validating transactions and creating new blocks, with PoS, the creator of a new block is chosen depending on its wealth, also called “stake”. The larger the stake, that is, the amount of digital currency owned by a user, the greater the probability that the system is not being hacked. Furthermore, the more an individual is exposed to a cryptocurrency, the more likely it is that he or she is behaving optimally. In the blockchains that use PoS there is no reward for generating the new block, all digital currencies were created at the beginning and their number never changes. In fact, in these systems, the miners are called “forgers”. They do not earn on the reward for generating a new block, but on transaction fees. Finally, in blockchains that use PoS, participants

211

212

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

with significant cryptocurrency are selected on a pseudo-random basis not to extract but to coin the blocks and add them to the blockchain. The pseudo-random selection process starts after the system has analyzed several factors to ensure that only individuals with a higher stake are selected, but also others with a lower stake. To date, Ethereum’s blockchain uses PoW to validate blocks, but for several years its creators have been planning to switch to PoS with a protocol called “Casper” because of the high power consumption needed to feed the PoW process. Today, a Bitcoin transaction requires the same amount of electricity to power an American family for 19 days [3]. And these energy costs are paid in fiat coins, leading to constant downward pressure on the value of the digital currency. For this reason, Ethereum developers would like to leverage PoS for a “greener” and therefore less expensive distributed consensus. One of the weaknesses of PoS is security. In a PoW system, due to the high cost of executing an attack, you would risk spending more than you would earn, so malicious users are excluded. A system that uses PoS without particular penalties might be cheaper to attack. For this reason, the Casper protocol provides in some circumstances for the loss of the deposit that the (malicious) user has deposited to participate in the creation of new blocks and of his privilege to be part of the consensus mechanism. Another solution could be to increase the price of the cryptocurrency, thereby increasing the cost of attacks. The kind, and related complexity, of a consensus algorithm strongly characterizes each blockchain technology since it is on the nature and strength of the distributed consensus that the security of the blockchain is based. It is the consensus management model that determines the difference between permissionless and permissioned blockchain. In a permissionless blockchain, trust between users, which are anonymous, is ensured by the consensus algorithm. While, in a permissioned blockchain, where the users are known and allowed by an administrator of the network, it is possible to use lighter consensus algorithms. Usually, in a permissioned Blockchain, a central authority determines who can access it, defines who is authorized to be part of the network and the roles a user can have. Within these networks, the validators are known, so there is no risk of a 51% attack resulting from miners collusion. As a result, transactions are cheaper, as they are verified only by a few nodes, called validators, which are considered trusted. When the consensus is performed by a group of known and reputable nodes the mechanism is called Proof of Authority (PoA). Validators are member nodes authorized to participate in validating transactions and adding blocks to the blockchain. Knowing the identity of the miners means that their work does not have to be controlled and verified by the other nodes, further reducing the time and cost of execution. Since trust is not derived from the consensus algorithm used, simpler and faster algorithms can be used in permissioned networks. The most used are the practical-Bizantine-Fault-Tolerance (pBFT) algorithms, characterized by low complexity and high practicality in distributed systems. For the pBFT system to work, the number of malicious nodes must be less than or equal to 1/3 of all system nodes in a

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

given vulnerability window. The more nodes in a pBFT network, the more secure it becomes. When a user sends a transaction request, it is received by a validator, then one of the validators is randomly elected “leader”. When a new block is generated, the leader orders the transactions that should be included in the block and sends this list of ordered transactions to all other validators on the network. When each validator receives the list of the ordered transactions, it starts executing the transactions one by one. As soon as all transactions are executed, it calculates the hash code for the block. Then it transmits its response (the resulting hash) to the other validators on the network and starts counting their responses. If he sees that 2/3 of all validators have the same hash code, he will commit the new block on the blockchain. The consensus mechanisms used for permissioned blockchain as well as PoS-based mechanisms offer the advantage of being more environment-friendly than PoW-based algorithms, as validators are not competing to find the solution to the mathematical quiz and therefore do not need to use the computational power typical of other types of consensus. In general, each consensus algorithm has disadvantages and advantages, each one is valid for some applications, others are not but, certainly, for the execution of an energy blockchain, algorithms very expensive in terms of computational resources are to be avoided.

7.2.3 Smart contracts Contracts are useful tools that have always been used when two or more parties want to establish, regulate or terminate a relationship of any kind. In the digital era, the blockchain technology offers many advantages in terms of transparency, information symmetry and disintermediation, but it also offers the possibility to implement the so-called “smart contracts” (SC). A SC is defined as the “translation” into a code of a contract between the parties that can verify automatically specified and predefined conditions and execute actions when conditions between the parties are fulfilled and verified. So, it is a tool that guarantees the same trust as a traditional contract, but with the advantage of automatically verifying the terms of the contract, performing actions such as money transfer and assessing the situation, if one of the parties does not comply with the terms. SCs deployed by a blockchain are shared on network peers and are usually executed to implement transaction logic and verify their validity, or to execute transactions automatically if certain conditions are met. SCs are stored on the blockchain, thus distributed, and no one has complete power over the contract. Moreover, being implemented on the blockchain, they inherit some interesting properties: • immutability: after the creation, a SC can never be modified. No one can change the contract code, not even the creator of the contract itself; • the contract is distributed on the network and validated by the blockchain. A single node cannot force the contract against the specified features because the other nodes in the blockchain would not approve it.

213

214

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

In the blockchain world, the SCs were introduced with Ethereum [4], a global and open-source permissionless blockchain for decentralized applications that implements SCs for cryptocurrency transaction execution. The programming language used for Ethereum SCs is “Solidity”, a Turing complete language, but not suitable for the implementation of complex codings such as the evaluation of optimal power flows or other optimization codes. Ethereum SCs are compiled in “bytecode” and are read and executed by a virtual machine called “Ethereum Virtual Machine” (EVM). Each node of the Ethereum network has its EVM, which represents the environment where SCs are executed. To write a transaction within the Ethereum blockchain, a SC is executed, and a fee is paid for this execution, whose currency is expressed in “gas”. The amount of gas to be paid is directly proportional to the length and computational complexity of the SC to be performed, it is also possible to establish the value “gas price”, that is, the amount of Ether (the Ethereum cryptocurrency) per unit of gas you are willing to pay; The modification of this variable, affects the priority of recording of the transactions on the blockchain, in fact, with a higher gas price, the transaction will be recorded in a lower amount of time than a transaction made with gas price of lower value. There is therefore an increase in the speed of registration of the transaction, but with a higher price. In the energy sector, the timing and the cost of transactions is a significant challenge. As already said, Ethereum is a permissionless blockchain. An important difference between permissionless and permissioned blockchain concerns the complexity of the consensus algorithms necessary to ensure the consistency of the data stored in the blockchain. The permissioned blockchains use lighter consensus algorithms, whereas the permissionless blockchains use algorithms with a higher computational complexity, but necessary to ensure the integrity of the exchanged data. The use of permissionless blockchains provides greater security in case data can be released publicly but, due to privacy and the computational time of the heavier consensus algorithms they are not adequate for handling energy transactions. For this reason, energy-related applications are more efficiently handled using permissioned blockchains, despite in the literature there are many solutions that use permissionless ones. A permissioned blockchain that is well suited for application in the energy sector is Hyperledger Fabric [5], a platform designed for the use in business contexts, with a highly modular and configurable architecture that enables innovation and versatility for a wide range of use cases in the industry. Fabric, unlike Ethereum, supports SCs written in generic programming languages like Java, Go and Node.js. The SCs are called “chaincodes” and they work as a reliable distributed application that acquires its security/trust from the blockchain and the underlying consensus among the peers. In an Hyperledger Fabric network, the SC represents the main element of the network, because it is the component that

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

implements the logic of each kind of transaction and verifies the integrity of each transaction sent by network users. The chaincodes are tools that allow the users to interact with the blockchain ledger, in fact any operation, even a simple query, is performed in Fabric through the SC. The SCs are installed on each peer, and every time a user sends a transaction proposal on the network, this transaction proposal is processed by the SCs of all peers. Then, if the transaction sent by the user is consistent with the logic implemented by the SC and verified by all peers, the new transaction after the consensus process is recorded in a new blockchain block. In conclusion, we can say that the SCs are software programs that allow the transfer of ownership of any digitally identifiable good from one person to another or the implementation of the transaction logic for a given application. In peer-to-peer (P2P) electricity transactions, as an example, the SC can be used to execute energy transactions directly when the trading conditions set by the prosumers are met, or to assign to the prosumers digital tokens corresponding to the energy injected into the grid and destroy them as the user consumes electricity [6]. Tokens are issued by the blockchain platform. They can either represent, as in this case, the representation of a digital asset, or they can be used as a payment (payment token), or they can prove the ownership of a right (service or utility token). Such right can be related to the property of a financial asset (service token) or to the possibility to obtain a service/benefit (utility token) through the platform. In the DR applications, tokens can be used to implement the logic of the DR programs, to certify the energy service provided by users through remuneration [7]. For V2G applications the SCs can be used to certify the car charging or to implement the logic for the network support services provided by the electric vehicles [5]. In both cases, the car charging can be paid by e-tokens that can be attained by the user as he offers the grid an energy service through the battery of the vehicle.

7.2.4 State of art of blockchain applications for P2P, DR and V2G Thanks to its characteristics of decentralization, immutability, trust, transparency, traceability and total network distribution, the blockchain could be the solution to many problems if applied to the energy field. This distributed technology is well suited to the energy applications because with the advent of RES, the grid is also becoming a distributed system, and the blockchain could, therefore, help to support the grid change and enable new business models for the energy market. For this reason, today many projects in the energy field are using blockchain technology. The possible applications are many, but the most popular are: Electrical Energy Trading, DR tracing and V2G.

215

216

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

The Electrical Energy Trading comprises the applications in which two parties exchange electricity including the P2P energy exchanges. Some papers focus on this application proposing a blockchain-based framework to enhance the P2P electricity exchanges between the prosumers of a microgrid [8 10]. While the work in [11] proposes an authorized blockchain that uses Hyperledger Fabric to provide a network of P2P energy transactions in order to cope with the growing volume of renewable energy, using Energy Storage System (ESS), the work in [12] proposes an energy trading platform using also ESS and energy token exchange solving the security problem through the blockchain private network and implement the P2P system that enhances the transparency and immutability by executing automatic transactions without involving third parties using smart contracts. The proposed architecture consists of a private blockchain network in which energy is exchanged through ESS and energy tokens through the blockchain. Users register sales and purchase offers on the blockchain, after which the Distribution System Operator (DSO) compares the total amount of the purchase and sale. Trade is realized if the amount of energy to buy and sell is equal. So the smart contract moves the token and the energy according to the accomplished trade, and then the result is recorded on the blockchain. The blockchain technology also offers the possibility to enable the exchange of energy services rather than energy exchange. In order to respond to power fluctuation due to the penetration of the RES in the electrical system, DR management represents an efficient solution. In general, DR indicates a structured program intending to modify the electric load diagram in response to problems in the network, such as network congestion or temporary unavailability of power caused by failures or intermittent production from nonprogrammable RES. By using the blockchain, the DR programs can be performed directly between the system operator and the users, unlike what happens today where the aggregator mediates between user and system operator. In this case, we are talking about the exchange of an energy service: the system operator requests a load reduction/increase and the users response with a load reduction/ increase, the transaction is not a direct energy exchange between two parties, but a request to provide a service, where users will be remunerated if they comply with the request. One of the main issues in the provision of the traditional DR service is the lack of transparency between the different stakeholders, but through the blockchain it is possible to handle this problem. Thanks to the blockchain, it is possible to directly involve the small prosumers in the capacity and balancing markets by aggregating them in virtual load units. The issue of the decentralized management of DR is discussed in many works. As an example, the work in [13] proposes a blockchain platform for storing the energy usage and generation information of the prosumers of a microgrid, and SCs running on Ethereum blockchain evaluate the expected flexibility of each prosumer, the associated remuneration or penalty and the rules for the whole energy balancing in the considered microgrid. Another interesting work is the recent

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

H2020 project DELTA [14], which proposes a Virtual Node (VN) platform to manage problems on the network due to peak loads and imbalance through end-user cooperation and interactivity. In this case, no SC manages the flexibility of the prosumers, but a multiagent system running on the VNs which updates energy profiles for each prosumer in the aggregator’s portfolio, with load forecasting and dispatch optimization tools that provide the necessary information required for self-balancing and preventing the internal loss of energy or stability. Instead, the work in [7] proposes a blockchain-based framework in which an Hyperledger Fabric network for DR programs execution has been developed. In this case, the SC is used to implement the execution logic of the DR programs and to evaluate the customers’ baseline in a trusted, shared and transparent way. Alongside DR, applications involving electric vehicles and blockchain technology are currently of huge interest [15]. Electric vehicles can be considered as mobile accumulators that can exchange energy with the electricity grid or other electric vehicles. This leads to three scenarios: Vehicle-to-Vehicle (V2V), Vehicleto-Grid (V2G) and Grid-to-Vehicle (G2V). In this way, electric vehicles can contribute to DR, improve grid resilience and reduce peak loads through charging and discharging. On the other hand, however, the penetration of an uncoordinated charge can lead to grid overload. In the literature, there are many works with the aim of integrating electric vehicles into the grid and optimally coordinate charging and discharging using blockchain technology. In [16], the authors propose the use of a private blockchain for the implementation of a protocol that allows owners of electric vehicles to find the cheapest charging station within a defined area and preserve the privacy of the electric vehicle. The protocol is based on a blockchain where electric vehicles send their request and the charging stations send bids similar to those of an auction. The owner of the electric vehicle then decides on a particular charging station based on the bids it receives. In this case, the blockchain is used as a decentralized and immutable warehouse for transparency and verifiability of bids. Also in [17], the authors propose a private blockchain to implement a secure pricing scheme for charging and discharging electric vehicles, but in this case, all is executed through optimal SCs to meet the needs of electric vehicles, individual energy consumption preferences and maximize the utility of the network operator. Through the execution of dedicated SCs between an aggregator and owners of electric vehicles, the energy exchange and cryptocurrency exchange process are performed automatically and securely. While, one of the most recent works in this field [18], shows a model of electricity exchange based on a blockchain with the aim of performing P2P transactions between electric vehicles in V2G networks. Moreover, the authors, in order to verify the feasibility of the proposed work, designed and released the SC implementing the mechanism developed on Ethereum, finally verifying the effectiveness of the proposed scheme.

217

218

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

7.3 The energy blockchain: current trends and possible evolutions The blockchain technology offers the possibility to change the approach of the “market integration” into “market substitution” allowing peer-to-peer transactions. Out of the many energy blockchain applications, such as: 1. Peer-to-Peer and decentralized energy trading 2. Wholesale energy trading 3. Energy trading support for small generators and end-consumers 4. Metering and billing 5. Vehicle to Grid 6. Demand Response (DR) for primary/secondary/tertiary regulation 7. Green certificates and carbon trading 8. Grid management 9. Power Purchase Agreements (PPA) Some will be considered in greater detail, considering those that are more relevant to the aim of this chapter.

7.3.1 Peer-to-peer energy exchanges among prosumers The future distribution system will be organized into small clusters of energy resources. Such clusters can be technologically identified as Microgrids; these grids host controllable loads, storage units, conversion systems (also to integrate electrical as well as other energy resources) and through an ICT infrastructure they can control them. The members of a Microgrid will, in the future, decide from/to whom they would buy/sell energy, based on different criteria (not only economic) also considering social and environmental issues. As an example, a consumer can specify the maximum price he would pay for renewable energy and the users can give higher priority to the sale/purchase of energy from relatives or someone living in the neighborhood. The platform will register the interest of buyers and sellers in an order portfolio. The users will be able to modify their real-time price preferences. In the future, end-users will collect historical information on prices and thus learn and adapt their offer strategies. The prices from the buyer side are determined based on the demand/offer law, with consequent price volatility. As a consequence, it is necessary to protect the elder and marginalized citizens. Moreover, the equilibrium prices will not only be determined by simple cost functions, but also by social values. Individual preferences and social behavior of market participants require further analysis to develop efficient market models and price mechanisms. The energy retailers or grid operators can see jeopardized their interests. The increase of energy self-sufficiency may imply a reduction of earnings, while at the same time, the costs related to the functioning and maintenance of the electrical network may increase. As an example of the great interest in this topic, the most significant projects on the application of the blockchain to P2P energy exchanges are briefly described below.

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

7.3.1.1 The Brooklyn Microgrid The first energy blockchain application is the highly cited Brooklyn Microgrid in New York supported by Siemens Digital Grid and LO3 Energy [19,20]. The first plants have been connected among them in 2016 and the activity then expanded getting to the limits imposed by regulations that were restricting the right of end-users to sell the excess electricity to neighbors. A recent petition against governor Cuomo opened up this possibility at the end of 2019. Prosumers can now sell the excess electricity directly using SCs based on Ethereum and pBFT consensus, implemented on Tendermint. The first process included five prosumers and five closeby consumers and brought to the first energy transaction ever registered on a blockchain worldwide. The energy surplus is registered by smart meters suitably designed for managing energy data and measurements. These are then translated into energy token that can be exchanged on the local market. The token indicates that a given amount of energy has been produced by the photovoltaic panels and can be transferred from the portfolio of a consumer to the end-users by the blockchain technology. The microgrid users interact with the platform specifying their individual price preferences under the form of availability to pay or sell electricity. The platform can visualize energy prices in real-time. In the initial phase of the project, the users manually acted on the platform, they could for example make an offer at a given price and a given hour. The public register keeps the terms of the contract, the parties that carry out the contracts, the volumes of energy injected and the consumption measured by the measuring devices as well as the chronological order of the transactions. Any member of the community can have access to all historical transactions in the log and verify the connections. More than 300 homes and small businesses, including about 50 photovoltaic prosumers and a small wind generator, have signed up for the second phase of development, with fully automated devices. The Brooklyn MicroGrid project aims to serve as a testbed for the development of new business models that promote consumer involvement in energy community projects. Local energy exchanges open up the possibility of saving on energy costs. 7.3.1.2 Other energy trading projects Along time many interesting experiences took place. Below, some are cited. Enyway was founded in 2017 in Hamburg with the mission of facilitating a democratic use of energy through the connection between producers and consumers. It was defined as the AirBnb of energy, because it allows consumers to buy electricity from individual owners of a wind turbine or a solar roof. Users choose the projects from which to buy energy on the map of Germany, with the help of images and data, as in the case of apartments/rooms offered on AirBnb. They also collected resources through crowdfunding to build a 1.3 MW plant (built by BayWa r.e) without incentives, which allows to supply energy to a large number of small financiers. This

219

220

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

approach is interesting because it expands the production from renewables without the need for incentives. In France, the Sunchain experience is expanding. It allows energy exchanges between prosumers and consumers using the blockchain not associated with a cryptocurrency. Power Ledger is an Australian company that has developed a renewable energy trading platform enabled by blockchain technology. Since November 2018, the Power Ledger blockchain has been used to track solar energy transactions between 18 families in Fremantle, Western Australia allowing residents to set their own prices and exchange solar energy generated from their roofs with neighbors who do not have solar energy. In Japan, the Power Ledger blockchain platform will be used to create, track, trade and provide for the regulation of renewable energy credits generated by solar systems on the roof of buildings. The so-called “non-fossil value” credits allow electricity retailers to obtain certificates of renewable energy for electricity that has been fed back into the grid by photovoltaic systems. In the next phase of their collaboration, Power Ledger’s decentralized accounting technology will ensure that every transaction is immutably recorded in real-time, in order to prevent renewable energy certificats from being used more than once. As Japanese solar incentives have now come to an end, Power Ledger plans to explore new ways to benefit from rooftop photovoltaic installations. 7.3.1.3 Grid stabilization and Vehicle to Grid applications In Germany, an experiment took off between eServices Sonnen (Sonnen was acquired in 2019 by Shell), and the network operator TenneT with the aim of using a network of thousands of residential solar batteries to stabilize the network in the presence of an excess of electricity generated by the wind turbines of northern Germany. The experimentation that began with a pilot project in 2017 ended successfully in 2019. At the end of 2018, there were 120,000 plants equipped with photovoltaics and batteries in Germany and by 2030 the decentralized storage systems could reach a total capacity of 10,000 MW. Another trial was held in the Netherlands in collaboration between the Vandebron company and the operator of the TenneT network. In particular with this Proof of Concept, Vandebron works with electric vehicle owners to make car batteries available to contribute to the balancing of the network. Vandebron has developed a solution to provide this service to its customers without compromising battery availability. The blockchain solution is designed to allow each car to participate in the project by recording their availability and action in response to TenneT signals. The trial initially involved 150 Tesla, the making available of which allowed the owners to receive compensation. Energy flows could be followed via app.

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

7.3.1.4 PPA management Another application that may have important developments, at the moment tested by Power Ledger over a 250 kW plant in Australia, concerns the use of a blockchain system to guarantee transparency in the management of data concerning transactions inside PPA.

7.3.1.5 The BLORIN project The BLORIN project (financed through Action 1.1.5. within the P.O.FESR 2014 2020—European cohesion funds) [21] aims to identify: • the definition of scenarios for DR and Vehicle to Grid programs application in microgrids and physical islands; • the Smart Grid elements capable of managing local Energy Communities; • the theoretical objectives to be pursued by all actors and the methodologies to be used; • the definition and development of ICTs to support this system based on a blockchain platform. In particular, the project aims at implementing three testbeds for the proof of concept of blockchain applications to DR and V2G operation. The testbeds will constitute a virtual laboratory comprising the Sicilian islands of Lampedusa and Favignana as well as the power network of the University of Palermo, Italy. The two testbeds on Lampedusa and Favignana will allow to assess the impact of the blockchain for managing DR and V2G in small isolated networks. The considered models of Local Energy Communities will include the management of: (1) photovoltaic generators combined with storage systems, (2) Electric Vehicles recharging systems, (3) the modulation of demand through DR programs. The project will comprise nodes of the blockchain implemented on the physical nodes of the whole architecture with specified technical features. The latter being: low power consumption; delay compatible with regulation times (scalability), privacy-preserving capability. All features are attained by using a permissioned blockchain platform. Sensors and actuators, which are generally low-end devices, provide their contribution to the network running client applications that interact with a permissioned blockchain, like Hyperledger Fabric.

7.3.2 Challenges of using the blockchain technology for DR and V2G applications Finally, for all the considered applications, there are some specific and general challenges concerning the use of the blockchain technology [22,23] that are worth mentioning and that are listed in Table 7.1.

221

222

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 7.1 Specific and general challenges. Application

Specific challenges

General challenges

Peer-to-Peer and decentralized energy trading

Implementation of efficient market algorithms as smart contracts; balancing demand to supply and need for machine learning algorithms; Interoperability of current smart meters with blockchain platform; Involvement of all market actors

• Compatibility with low power nodes • Regulatory challenges • Costs of integration • Latency • Privacy and General Data Protection Regulation compliance • Scalability (management of large amount of data) • Bandwidth • Power consumption

Wholesale energy trading Energy trading support for small generators and end-consumers Metering and billing

Vehicle to Grid (V2G) Demand Response for primary/secondary/ tertiary regulation

Green certificates and carbon trading

Grid management

Power Purchase Agreements

Interoperability of current smart meters with blockchain platform; compatibility with existing market structure Interoperability of current smart meters with blockchain platform; metering data verification from multiple actors for consensus; Interoperability with current existing V2G control technology Communication of current smart meters with blockchain platform; metering data verification from multiple actors for consensus; latency; bandwidth compatibility Communication of current smart meters with blockchain platform; metering data verification from multiple actors for consensus Higher throughput and transaction speeds that would allow real-time verification; interoperability with current existing metering technology; generation of massive datasets Interoperability of current smart meters with blockchain platform; metering data verification from multiple actors for consensus

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

7.4 Laboratory setup for energy blockchain testing 7.4.1 Simulation and emulation of smart prosumers The simulation and testing of a new energy blockchain network are of considerable importance before they are applied to real scenarios. For this reason, beyond the network of nodes, there is the need to simulate the prosumers that will be part of the microgrid and then of the new blockchain network. In general, a prototype system for the laboratory emulation of a smart prosumer interacting with a blockchain node requires the following components: 1. Some switchable resistors: for the simulation of step-varying passive loads; 2. A controllable electric motor: for the simulation of active loads; 3. An energy production system: for the simulation of RES generators; 4. A storage system: for simulating the energy exchange with the grid; 5. An energy management system: for the coordinated control of all components; 6. Monitoring and smart measurement system: for monitoring, measuring load and generation profiles and communications with the blockchain; 7. Protections and isolation transformer: for protection against electric shock and overcurrents. Fig. 7.1 shows the prototype of a smart prosumer made to comply with the above requirements and installed at the Power Systems Laboratory of the University of Palermo.

Figure 7.1 Smart prosumer prototype at the Power Systems Laboratory at the University of Palermo.

223

224

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 7.2 Interaction between the different parts of the HW/SW prototype for testing the BLORIN blockchain platform.

Fig. 7.2 shows the different components for the experimental part of the BLORIN project. The BLORIN platform will aggregate different resources. The main component interfacing the measuring and control units to the blockchain is the SNOCU, a proprietary independent device produced by Regalgrids that allows to connect assets generation, storage or pure consumption, to the platform that enables different energy services (DR and V2G) or energy communities. In Fig. 7.2, the energy provider coincides with the grid operator, how it usually happens in most of the small islands in Mediterranean Sea. A dedicated smart contract will: • read data from the Blockchain for different purposes; • compute the baseline of each prosumer; • publish a DR event; • register and certify consumptions/generation during DR events; • calculate remunerations. The smart contract and the data collected from the smart meters or outputted by the SC are written on the BLORIN platform. The above actions can be invoked by any of the nodes enabled for the specific function (indicated in Table 7.2). The SNOCU, a device developed for monitoring and control of inverters and storage systems, is equipped with an ARM microprocessor and 1GB of memory (Raspberry PI3 like). ARM microprocessors are widely used in mobile applications due to their low power consumption characteristics, but on the other hand they are not so powerful. So due to limited computing capacity it will probably not be able to

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

Table 7.2 Roles and Smart contract execution. Smart contract functions

Who

Reads data from the Blockchain for different purposes; Writes data on the Blockchain for different purposes Compute the baseline of each prosumer; Publish a Demand Response (DR) event; Register and certify consumptions/ generation during DR events Calculate and give remuneration

Regulator, energy provider, end-users, grid operator, aggregator Energy provider, grid operator, aggregator Grid operator, aggregator Grid operator, aggregator Grid operator, aggregator Grid operator, aggregator

host an Hyperledger blockchain full node. As claimed in [23], it is required to recompile the Linux kernel to support Docker, since in Hyperledger Fabrik the chaincode are run inside Docker containers. Moreover, several tools must be precompiled, such as protoc and grpc, used in these containers for armv7l (32-bit ARM) and aarch64 (64-bit ARM) architectures. Finally, the size of some buffers needs to be reduced and timeouts need to be increased for the execution on RP3. For example, in [23] the authors decreased cNameArr buffer size from 256 MB (which is 1/4 of the available memory on RP3) to 1 MB, and increased request execution time- out from 30 seconds to 10 minutes. Nonetheless, once these issues are solved as suggested in [22], considering the DR bandwidth requirements [24], it seems that even a low power node would be enough for managing the features of DR programs. In [24], it is required 1 second roundtrip time for 1000 bytes messages and thus 16 kbps of minimum bandwidth will be needed for a single DR transaction. However, for the system to be scalable up to the numbers (hundreds) of end-users that could be involved downstream each MV/LV substation, and then HV/MV station, the bandwidth required must be scaled up to consider simultaneous transactions. Therefore, considering that downstream of a MV/LV substation with 500 endusers, whose 5% taking part to a single DR event, and the typical size of a metering message, the required bandwidth can sum up to 400 kbps, thus still leading to manageable numbers inside a standard Internet infrastructure. Even scaling up to HV/MV substations, considering the same percentage of users joining the program, the number would raise to 4000 users and thus to 64 Mbps. This computation considers bandwidth requirements as if the system was centralized; using the blockchain these requirements may increase requiring up to 10 times higher demand in bandwidth of the blockchain-based solution compared to real-time AMI. These requirements are still affordable on current MAN (Metropolitan Area Network) infrastructures.

225

226

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

7.4.2 The smart contracts in the BLORIN project for DR and V2G implementation In the BLORIN project, two main business models are taken into account, respectively for DR and V2G. Both business models will be established on the two physical islands. For this reason, in both cases, the technical objectives required by the grid operator are grid stabilization and balancing. The SC implementing the DR program will aim to modulate the loads to limit the power peak in certain hours of the day. Therefore, the grid operator, being available a load forecast, for each substation makes the desired planning of load/storage modulation and sends the request to the blockchain directed to the loads/storage downstream of the relevant substation. The nodes read the request and act accordingly. Similarly, for V2G, the grid operator sends the request on the blockchain and the V2G recharge stations in the area acquire the request and perform the required modulation to balance the energy flow upstream. In the following pseudo code, the entry point of the SC function for DR and V2G implementations are reported (Fig. 7.3): In the case of DR, the baseline of each customer must be assessed and, based on that, a remuneration can be calculated and acknowledged. Another application concerns the P2P energy exchange, which could take place in an energy community. In this case, energy resources like loads and vehicles’ batteries, as well as batteries installed close to photovoltaic plants, must be managed downstream of a given substation, that is, to minimize the hourly energy unbalance. The remuneration would be acknowledged to the whole community and shared based on the individual contribution to reaching the target.

func (t *SmartContract) Invoke(stub shim.ChaincodeStubInterface) pb.Response { function, args := stub.GetFunctionAndParameters() fmt.Println("invoke is running " + function) if function == "recordEnergy" { return t.recordEnergy ( stub , args ) } else if function == "evaluateBaseline" { return t.evaluateBaseline (stub ,args) //only for DR application } else if function == "notifyevent" { return t.notifyevent ( stub , args ) //either V2G or DR event } else if function == " queryEnergyModul" { return t.queryEnergyModul ( stub , args ) } else if function == "getToken" { return t.getToken (stub, args) } else if function == "queryEarnedToken" { return t.queryEarnedToken ( stub , args ) } fmt.Println ( "invoke did not find func:" + function ) return shim . Error ( " Received unknown function invocation") }

Figure 7.3 Smart contract pseudo code for DR and V2G management.

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

7.4.2.1 Future applications of the energy blockchain: the blockchain for energy communities RED-II [1] defines the role of prosumers and Renewable Energy Communities (REC) with the aim of increasing the decentralization of electricity generation in the context of the liberalization of the energy market. The RECs, which exist since several years in some northern European countries such as Denmark and Germany, are made up of aggregates of producers, consumers and possibly also prosumers, where renewable energy is produced by units owned by a community of citizens, with the aim of providing environmental, economic and social benefits to the community’s members rather than financial profits. Within a REC it is allowed to produce, store, consume and sell renewable electricity produced within the community and to access all electricity markets, either directly or through aggregation. The members of the REC are holders of connection points in low voltage electricity networks downstream the same MV/LV transformer substation, such that their participation in the REC does not involve any modification and/or upgrading of the existing electricity network. They can be final customers, producers and/or final customers and producers provided that they all belong to the same perimeter. In a REC there is usually an agent who applies to the authority to obtain the expected benefits from the selfconsumption of renewable energy produced within the community itself. The authority, on the basis of the measurement data collected at the collection and injection points of the consumption and production plants, assesses the collectively consumed electricity and grants incentives to the representative of the community, who is generally the owner of one or more production plants within the community. In Italy, for example, the decree-law 162/19, which transposes the European RED-II directive, states that the REC representative must submit a request to the authority (GSE) with all the information necessary to identify the energy community [25]. The GSE then recognizes an incentive equal, on an hourly basis, to the sum of two terms: 1. The product between the unit amount refunded (totaling 0.822 ch/kWh for the year 2020) and a quantity of energy equal to the minimum between the electrical energy supplied by the installations covered by decree-law 162/19 and the total electrical energy taken from the connection points that are part of the same REC; 2. The product between the coefficient of avoided losses (1.2% in the case of production plants connected to medium voltage networks or 2.6% in the case of production plants connected to low voltage networks), the hourly zonal price and a quantity of energy equal to the minimum between the electrical energy injected by the plants allowed by decree-law 162/19 and the total electrical energy taken from the connection points forming part of the same REC. This incentive is granted to the agent of the community, who then distributes it to users who consumed at the time in which the incentive was granted. The first term of the two mentioned above indeed is as larger as the minimum between injected and consumed

227

228

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

electricity is higher, namely as the effect of levelling (in terms of energy) is increased. Therefore, the more users consume at the time the renewable energy plant produces, the better it is. Besides, the self-consumption of energy allows to obtain several advantages both for the user of the energy community and for the network operator. The user will have an economic return due to the energy not purchased and the incentives, while the network operator will mainly have the benefit of the reduction of power flows on the power lines with the consequent reduction of losses. In addition, the electricity produced and consumed on site could, in perspective, reduce the need to upgrade existing networks or build new networks, as it would contribute to the reduction of the maximum power required at the connection points. A further advantage is the optimization of the use of delivery cabins and connection points, reducing connection costs. However, currently, in Italy the latter two benefits are not fully implemented due to the energy (and not power) related incentives. As a result these latter benefits can only be considered in a future perspective. The blockchain technology applied to the implementation of the REC concept would first of all allow full transparency of the incentive process for selfconsumption. Users would be identified on the blockchain as part of the REC and the hourly incentive process described above could be carried out through suitable smart contracts, in charge of distributing the benefits to all those that have taken part to the hourly energy balancing. In this way, not only the production/consumption network would be decentralized, but also the user management system. There is no need for a representative of the community, but all users would be in direct contact with the authority, which, hour by hour, through the smart contract could recognize the incentives directly to the users while attaining the required technical objectives.

7.5 Conclusions In this chapter, the role of the blockchain technology in the energy sector has been examined and discussed, with particular reference to DR and V2G applications. The future challenges of the energy blockchain comprise several aspects to be still discussed and problems to be solved, specifically with regards to the integration with the existing V2G technologies and DR programs. Many research projects and pilot installations are currently testing the potentiality of the blockchain as a disruptive technology for power systems in many countries. However, the blockchain’s potential of disintermediation would limit the role of some of the existing actors in the energy market. This puts a strong burden of regulators and member states that on one hand push for the opening of the energy market to endusers, on the other, are concerned about data privacy and the possible transformation of the energy market actors. The BLORIN project aims to overcome the privacy-preserving issue,by creating several ‘channels’ that are visible to only some of the users and not to all

Challenges and directions for Blockchain technology applied to Demand Response and Vehicle-to-Grid scenarios

of them. The use of Hyperledger Fabric also allows the use of different consensus algorithms and thus modulate the required computational requirements of peripheral units. The project aims at proving that it is possible to implement a blockchain DR and V2G platform to optimally manage energy resources in energy districts and attain the power balancing objective or any other objective that can be beneficial to the power grid operation. The prototype of smart prosumer designed and carried out within the BLORIN project will permit to explore, with more detail the applicability of the blockchain to the particular reality of energy communities in small islands where DR and V2G programs can be used for fostering the development of RES and the transition to cleaner and greener power systems.

Acknowledgment This work has been supported by the project BLORIN financed within the call POFESR Sicilia 2014 2020 Azione 1.1.5 “Sostegno all’avanzamento tecnologico delle imprese attraverso il finanziamento di linee pilota e azioni di validazione precoce dei prodotti e di dimostrazione su larga scala”.

References [1] Directive (EU) 2018/2001 of the European Parliament and of the Council of 11 December 2018 on the promotion of the use of energy from renewable sources. Available from: ,https://eur-lex.europa.eu/legalcontent/EN/TXT/?uri 5 uriserv:OJ.L_.2018.328.01.0082.01.ENG&toc 5 OJ:L:2018:328:TOC., 2018. [2] S. Nakamoto, Bitcoin: a peer-to-peer electronic cash system, (2008), accessed (31.05.20). Available from: ,https://bitcoin.org/bitcoin.pdf.. [3] Digicomist, Bitcoin Energy Consumption Index, digicomist. net. [Online]. Available from: ,https://digiconomist.net/bitcoin-energy-consumption., 2020 (accessed 31.05.20). [4] V. Buterin, A next generation smart contract & decentralized application platform, 2009, (accessed (29.05.20). Available from: ,https://blockchainlab.com/pdf/Ethereum_white_paper-a_next_generation_smart_contract_and_decentralized_application_platform-vitalik-buterin.pdf.. [5] C. Liu, K.K. Chai, X. Zhang, Y. Chen, Peer-to-peer electricity trading system: smart contracts based proof-of-benefit consensus protocol, Wireless Networks. (2019). Available from: https://doi. org/10.1007/s11276-019-01949-0. [6] W. Tushar, T. Saha, C. Yuen, D. Smith, V. Poor, Peer-to-peer trading in electricity networks: an overview, IEEE Trans. Smart Grid 1 (2020) 1 15. [7] M.L.D. Di Silvestre, P. Gallo, E. Riva Sanseverino, G. Sciumè, G. Zizzo, Aggregation and remuneration in demand-response with a blockchain-based framework, IEEE Transactions on Industry Applications 56 (2020) 4248 4257. Available from: https://doi.org/10.1109/TIA.2020.2992958. [8] M. Sabounchi, A decentralized P2P electricity market model for microgrids, IEEE Power Energy Soc. Gen. Meet. (2018) 1 5. Available from: https://doi.org/10.1109/PESGM.2018.8586257. [9] B.C. Neagu, G. Grigora¸s, O. Ivanov, An efficient peer-to-peer based blokchain approach for prosumers energy trading in microgrids, in: 2019 8th International Conference on Modern Power Systems (MPS), Cluj Napoca, Romania, 2019, pp. 1 4. Available from: https://doi.org/10.1109/MPS.2019.8759743. [10] A. Alam, M.T. Islam, A. Ferdous, Towards blockchain-based electricity trading system and cyber resilient microgrids, in: 2019 International Conference on Electrical, Computer and Communication Engineering (ECCE), Cox’s Bazar, Bangladesh, 2019, pp. 1 5. Available from: https://doi.org/10.1109/ ECACE.2019.8679442.

229

230

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[11] R.K. Kodali, S. Yerroju and and B. Y. K. Yogi, Blockchain based energy trading, TENCON 2018 2018 IEEE Reg. 10 Conference, Jeju, Korea (South), 2018, pp. 1778 1783. Available from: https://doi.org/10.1109/TENCON.2018.8650447. [12] S.J. Pee, E.S. Kang, J.G. Song, J.W. Jang, Blockchain based smart energy trading platform using smart contract, in: 2019 International Conference on Artificial Intelligence in Information and Communication (ICAIIC), Okinawa, Japan, 2019, pp. 322-325. Available from: https://doi.org/ 10.1109/ICAIIC.2019.8668978. [13] C. Pop, T. Cioara, M. Antal, I. Anghel, I. Salomie, M. Bertoncini, Blockchain based decentralized management of demand response programs in smart energy grids, Sensors 18 (162) (2018) 1 21. [14] CERTH HIT Hypertech Innovations, Archi Ilektrismou Kuprou, University of Cyprus, KIWI Power Ltd., Joint Research Centre (JRC) European Commission, E7 Energie Markt Analyse, Universidad Politecnoca de Madrid, Norges Teknisk-Naturvitenskapelige Universitet (NTNU), Carr Communication International Communications Consultants (CARR), “DELTA,” delta. eu. Available from: , https://www.delta-h2020.eu/., 2020 (accessed 15.06.20). [15] M. Baqer Mollah, J. Zhao, D. Niyato, K.-Y. Lam, X. Zhang, A.M.Y.M. Ghias, et al., Blockchain for future smart grid: a comprehensive survey, IEEE Internet Things J. (2020). [16] F. Knirsch, A. Unterweger, D. Engel, Privacy-preserving blockchain-based electric vehicle charging with dynamic tariff decisions, Computer Sci Res Dev. 33 (2018) 1 9. Available from: https://doi. org/10.1007/s00450-017-0348-5. [17] Z. Su, Y. Wang, Q. Xu, M. Fei, Y. Tian, N. Zhang, A secure charging scheme for electric vehicles with smart communities in energy blockchain, IEEE Internet Things J. 6 (3) (2019) 4601 4613. Available from: https://doi.org/10.1109/JIOT.2018.2869297. [18] H. Liu, Y. Zhang, S. Zheng, Y. Li, Electric vehicle power trading mechanism based on blockchain and smart contract in V2G network, IEEE Access. 7 (2019) 160546 160558. Available from: https://doi.org/10.1109/ACCESS.2019.2951057. [19] E. Mengelkamp, J. Gärttner, K. Rock, S. Kessler, L. Orsini, C. Weinhardt, Designing microgrid energy markets: a case study: the Brooklyn microgrid, Appl. Energy 210 (2018) 870 880. [20] L. Orsini, S. Kessler, J. Wei, H. Field, 10 - How the Brooklyn microgrid and transactive grid are paving the way to next-gen energy markets, The Energy Internet (2019) 223 239. Available from: https://doi.org/10.1016/B978-0-08-102207-8.00010-2. [21] BLORIN Project website, 2020. Available at: , https://www.blorin.energy/ . . [22] M. Andoni, V. Robu, D. Flynn, S. Abram, D. Geach, D. Jenkins, et al., Blockchain technology in the energy sector: a systematic review of challenges and opportunities, Renew. Sustain. Energy Rev. 100 (2019) 143 174. [23] D. Loghin, G. Chen, T.T.A. Dinh, B.C. Ooi, Y.M. Teo, Blockchain goes green? An analysis of blockchain on low-power nodes, Perform. Anal. Comput Syst. (2019). arXiv. org . cs . arXiv:1905.06520. [24] A. Losi, P. Mancarella, A. Vicino, Integration of Demand Response Into the Electricity Chain: Challenges, Opportunities, and Smart Grid Solutions, Wiley, 2015. ISBN: 9781848218543. [25] ARERA, Regulatory authority for energy networks and environment, Guideline for the regulation of economic items related to electricity subject to collective self-consumption or sharing within renewable energy communities, Consultation Document (2020). 112/2020/R/EEL, 1 April.

CHAPTER 8

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community Anna Pinnarelli1, Daniele Menniti1, Nicola Sorrentino1 and Angel A. Bayod-Rújula2 1

Department of Mechanical, Energy and Management Engineering, University of Calabria, Rende, Italy Department of Electrical Engineering, University of Zaragoza, Zaragoza, Spain

2

Abbreviations ACMG AC microgrids BMS battery management system CAES compressed air storage CEP community energy provider DBS DC bus signaling DCMG DC microgrid DSO distributed system operator DAB dual active bridge DHB dual half-bridge ESS energy storage system FC fuel cell IoT Internet of things ICT information and communication technology LA lead acid LV low voltage MS micro source MPPT maximum power point trekking nMS nanogrid management system Nonsum Comm non consumer community PEI power electronic interface PEM proton exchange membrane PCC point of common coupling P2G power to gas P-to-H power-to-hydrogen RES renewable energy source SC super capacitor SMES superconductive magnetic energy storage SOC state of charge SRC series resonant converter VSC voltage source converter Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00004-2

r 2021 Elsevier Inc. All rights reserved.

231

232

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Nomenclature Ci Imax_CS Imax_HS Imax_LS Imax_M_abs Imax_M_inj Imin_CS Imin_HS Imin_LS Imin_M_abs Imin_M_inj Iref_CS Iref_HS Iref_LS Li Lij Rij Si j SLi Smax j SOCmax_ HS SOCmax_CS SOCmax_LS SOCmax_M_abs SOCmax_M_inj SOCmin_ CS SOCmin_ HS SOCmin_LS SOCmin_M_abs SOCmin_M_inj Vdc_M_abs_H Vdc_M_abs_L Vdc_M_inj_H Vdc_M_inj_L VDC_nombus Vdc_SC_H Vdc_SC_L Vdc_SH_H Vdc_SH_L Vdc_SL_H Vdc_SL_L VDC_us Vref_M_abs Vref_M_inj Vref_max Vref_min Vref1 Xj

price of the wholesale market at time step i maximum current for central slave operating condition maximum current for high slave operating condition maximum current for low slave operating condition maximum current for master absorption operating condition maximum current for master injection operating condition minimum current for central slave operating condition minimum current for high slave operating condition minimum current for low l slave operating condition minimum current for master absorption operating condition minimum current for master injection operating condition reference current for central slave operating condition reference current for high slave operating condition reference current for low slave operating condition modified community power profile at time step i j-th nanogrid initial profile at a time step i profile of j-th nanogrid must track at time step i variation from the initial profile of j-th nanogrid at time step i modified community power profile at time step i j-th nanogrid maximum regulation value maximum SOC for high slave operating condition maximum SOC for central slave operating condition maximum SOC for low slave operating condition maximum SOC for master absorption operating condition minimum SOC for master injection operating condition minimum SOC for central slave operating condition minimum SOC for high slave operating condition minimum SOC for low slave operating condition minimum SOC for master absorption operating condition maximum SOC for master injection operating condition high level DC voltage for master absorption operating condition low level DC voltage for master absorption operating condition high level DC voltage for master injection operating condition low level DC voltage for master injection operating condition DC bus nominal voltage high level DC voltage for central slave operating condition low level DC voltage for central slave operating condition high level DC voltage for high slave operating condition low level DC voltage for high slave operating condition high level DC voltage for low slave operating condition low level DC voltage for low slave operating condition DC bus voltage DC bus reference voltage for master absorption operating condition DC bus reference voltage for master injection operating condition maximum DC bus voltage reference minimum DC bus voltage reference DC bus voltage reference j-th nanogrid technical parameters

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

8.1 Introduction The growing increase in the demand for electricity, combined with the need to reduce the environmental impact in the production of electricity in accordance with the world’s eco-sustainable policies, has promoted plans for expansion and enhancement of current electricity grids, with consequent developments in the ICT (Information and Communications Technology) sector, creating a convergence of scientific and industrial interests on the use of these technologies. Then, the development of a real structural transformation process of the energy cycle must start from the production of electricity, the storage, transport, distribution, sale, up to the intelligent consumption of energy, according to appropriate criteria. In practice, for years now, the electricity is produced not only by large production plants, but also locally by users who have installed production systems based on renewable sources (mainly photovoltaic). Many users in recent years have decided to equip themselves with storage systems integrated with its own production plants; all this is flanked by home automation systems for the remote management of electrical loads. In this context, the possibility of a smart management of the energy produced, stored and consumed, in order to guarantee continuity in the power supply of users, but on the other also to guarantee a support to the network operator in case of need, becomes increasingly important. Therefore, passive users become active and the power flows become bidirectional, which implies a smart management of the grid itself, to ensure optimal exploitation of renewable energy sources (RES) and continuity in the power supply of users [1,2]. In this context, smart microgrids are one of the solutions that have become more widespread among the scientific community to tackle the problem of integration and management of RES, especially when these are highly random, nonprogrammable, and difficult to predict. Of course, the original need to manage only energy sources by means of microgrids is now overcome by the opportunity to manage on-site also storage systems and critical and noncritical loads, as well as the possibility of managing in an aggregate way several microgrids connected together to form virtual energy systems. Therefore, the objectives to be pursued can be different, such as satisfy the demand for loads, store energy for future needs, provide network services, guarantee continuity for critical loads [3]. Local management of electricity production and demand allows to reduce losses along power lines; moreover, demand peaks can be reduced by scheduling the schedulable electrical loads appropriately so as to avoid their simultaneous switching on in the peak time bands, and this can be achieved through the use of monitoring and control systems for energy flows (smart-meter), as well as automated user-side electrical load management systems. This also leads to a reduction in costs for users who will activate the loads not in conjunction with peaks in demand (when energy has a higher cost), but in different time slots (when energy has a lower cost) [1
3].

233

234

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

8.2 Energy storage systems as distributed flexibility 8.2.1 The flexibility in a distribution grid According to the Council of European Energy Regulators (CEER—2018) flexibility is defined as “the capacity of the electricity system to respond to changes that may affect the balance of supply and demand at all times” and has both technical and commercial dimensions. From an overall perspective, flexibility is related to maintaining energy supply and frequency stability and from a local perspective to maintaining bus voltages and securing transfer capacities. Energy systems need flexibility to match with the energy demand which varies over time. This requirement is pronounced in electric energy systems in which demand and supply need to match at each time point. In a traditional power system, this requirement is handled through a portfolio of different kinds of power plants, which together can provide the necessary flexibility in an aggregated way. Once variable renewable electricity is introduced in large amounts to the power system, new kinds of flexibility measures are needed to balance the supply/ demand mismatches. Flexibility represents the ability to increase or decrease production in generation plants or consumption in loads. It is a valuable resource for the security of the electricity system and in the energy markets and will be even more so in the future. Our electrical system already has a certain degree of flexibility today [4]. It will potentially have even more in the future considering for example: • the flexibility of the loads that represents the possibility of controlling, that is modulate (including detachment or insertion) electrical loads. • the flexibility of generation that represents the possibility of modulating the power fed into the electrical grid. • the flexibility given by the storage tanks: some plants, such as hydroelectric plants with pumping basins, batteries, storage of thermal or mechanical power, allow to store energy and use it when it is necessary for the system and/or when it is more convenient economically. Flexibility has become such an essential resource that markets have been created in European countries for the supply of network ancillary services. Increased level of flexibility is essential in power systems with high penetration of RES to maintain the balance between the demand and generation. In addition, the flexibility provided by energy storage systems and flexible conventional resources (i.e., generating units) can play a vital role in the compensation of the RES variability. To decide which are the best technologies to adopt, it is first necessary to evaluate the technical characteristics of each technology to verify to what extent they meet short, medium, and long-term flexibility needs, but it is also necessary to consider, for each

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

solution, the ability to address system objectives, such as economy, sustainability, energy independence, and security. Since each technology has its own technical characteristics and responds differently to these objectives, it is necessary to adopt a mix of solutions modulated over time. It is necessary to deploy new technologies that increase the flexibility dimension in distribution systems. Batteries and Demand Response represent optimal medium/long-term solutions, for breadth of use and consistency with the system objectives, although in the short term they present criticality of use mainly due to the current investment cost for batteries and accessibility and organization of the system for Demand Response.

8.2.2 The main energy storage system technologies Electric storage technologies include a broad category of devices. A classification of storage systems frequently adopted in the literature refers to the specific form of energy and distinguishes the storage systems in: • electrochemical storage (lead acid (LA) batteries, lithium ion batteries, zebra, nickel-metal hydride etc.); • mechanical type storage (compressed air storage (CAES), high and low speed mechanical flywheels, pumping hydroelectric basins); • electrostatic storage (supercapacitors); • electromagnetic storage (superconductive magnetic energy storage (SMES)) • chemical storage (hydrogen). The storage systems, in fact, are generally chosen based on the function they are called to perform. The energy storage technology can serve at various locations where electricity is produced, transported, consumed, and held in reserve (back-up). Depending on the location, storage can be large-scale (GW), medium-sized (MW) or micro, local systems (kW). Research and technological development is needed to enable the wider application of many known technologies, and to develop new ones. Some of the key technologies, not all of which are at the stage of commercial application are: • Large bulk energy (GW):
Thermal storage, pumped hydro;
CAES;
Chemical storage (e.g., hydrogen—large scale .100 MW, up to weeks and months) • Grid storage systems (MW) able to provide:
Power: super-capacitors, SMES, flywheels,
Energy: batteries such as LA, Li-Ion, NaS, and Flow batteries
Energy and Power: LA and Li-Ion batteries

235

236

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning


Hydrogen Energy Storage/CAES/Pumped Hydro Energy Storage (PHES) (small scale, 10 MW , P , 100 MW, hours to days) • End-user storage systems (kW):
Power: super-capacitors, flywheels
Energy: batteries such as LA and Li-Ion
Energy and Power: Li-Ion batteries In this chapter, the attention is focused on Li-Ion battery, supercapacitor, and Power-to-Hydrogen. 8.2.2.1 Li-Ion battery [5] Lithium/ion batteries have numerous variations and are characterized by high specific power, which is why they are also widely used in electric traction. The main disadvantage of these solutions is the high cost determined by the need for security systems that must be adopted to guard against potential overload situations. Lithium batteries are high energy systems and therefore must be treated with the utmost care. Electrical, mechanical, and thermal abuse can cause processes, such as thermal leakage, which can damage the cell and, in the worst case, also determine the gasification and release of flammable vapors of the solvent present in the electrolyte. For these reasons, Li-Ion cells are often equipped with Battery Management System (BMS) for the management of cell sizes such as voltage, current, temperature, which regulates the charge and intervenes if the operating parameters exceed the set limits. Furthermore, again for safety reasons, the cells are often equipped with sturdy metal containers. Lithium-ion cells have a specific energy between 130
180 Wh/kg, corresponding to an energy density of 270
380 Wh/l (the highest of all electrochemical storage systems). The specific power can reach peak values of 1800 W/kg (with reduced specific energy), for cells specifically designed to work at high power. Lithium-ion-polymer cells have very similar specific energy and energy density values (140
150 Wh/kg), while the specific power can reach 2800 W/kg. The energy efficiency is very high for both technologies, with values up to 95% depending on the operating conditions. The lifetime in cycles of the cells is 500 cycles with a discharge depth of 100% and is linked with logarithmic law to the discharge depth. A negative aspect of lithium-ion cells is that related to the degradation generally suffered by these devices over time, which translates into a progressive reduction of its capacity starting from the moment of manufacture, regardless of the number of charge/discharge cycles. The working temperature range is very wide, and it can go from 230 C (for some commercial cells up to 260 C) up to 60 C, although the recommended temperature is 30 C.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

Having a very high-power density and a long-expected life, they can be used in Power Quality applications. 8.2.2.2 Supercapacitor [6,7] Supercapacitors, known as electrochemical capacitors, have been developed mainly in the last 15 years and belong to the category of electrostatic accumulators. Their physical structure is characterized by the presence of two electrodes immersed in an electrolytic solution and separated by a permeable membrane. This type of storage arouses considerable interest as it has intermediate functional characteristics with respect to those of electrochemical batteries and traditionally built capacitors. The batteries, in fact, are characterized by a high energy density and a low power density and are suitable for slow charging and discharging processes (duration of hours). Traditional capacitors, on the other hand, have a low energy density and a highpower density and can therefore be used in extremely rapid charging and discharging processes (duration of fractions of a second). Due to their intermediate characteristics of energy density and power density, supercapacitors are suitable for charge-discharge processes lasting around one minute. Therefore, they can be effectively adopted as support storage systems to assist electrochemical batteries during short-term load peaks both for electric vehicle applications and for stationary storage applications in distributed generation plants. Furthermore, the supercapacitors can be coupled to the production systems from renewable sources, with the aim of compensating the fluctuations in the power generated, due to the uncertainty of the primary source, thereby improving the quality of production. They are particularly suitable for Power Quality applications, where the storage systems are used to improve the quality of the supply and ensure the goodness of the waveform of the supply voltage and for the supply of loads particularly sensitive even to slight anomalies in the supply voltage. The strengths of supercapacitor storage are precisely those that in traditional batteries represent weak points, namely they can withstand very high charging and discharging powers, so that if the power is available, they can be charged even in a few seconds. The same can happen in the discharging process if the needs of the connected load require it, however they can be loaded or unloaded at very small powers. This feature makes them very useful where there are sudden peaks of power both in production and in absorption alternating with minimum powers, which are the conditions that often occur in networks and in the production of renewable energy. 8.2.2.3 PEM based power-to-hydrogen [8
11] A hydrogen-based energy storage system includes the following main components: • A conversion system designed to convert incoming electricity into hydrogen at the output based on Proton Exchange Membrane (PEM) technology.

237

238

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

A hydrogen storage system. Hydrogen is stored in gaseous form in pressure tanks up to 70 MPa. • A system to convert hydrogen into electricity, consisting of a fuel cell based on PEM technology. The Power-to-Power type is considered as the direct conversion into electricity, with fuel cells which, in addition to consuming nonprogrammable renewable energy when it is in excess, can return it when necessary. Although the overall efficiency is lower than that of a battery (depending on the types of electrolyzers and fuel cells, 40%
60% of the energy is lost in the process, compared to 20% dissipated by a battery system), the conversion in fuel can be convenient in terms of plant size and capacity to maintain stored energy for long periods. It can be used to make the prosumer completely self-sufficient (or almost) by the distribution network in terms of energy needs in normal conditions—keeping the connection in primis to ensure a possible backup supply, especially for relatively small power plants. It plays a role of primary importance to guarantee energy storage when production from renewable sources exceeds demand. The energy thus stored in the form of hydrogen can be used when the demand from the network cannot be satisfied by the energy produced from renewable sources.

8.2.3 The flexibility services provided by energy storage systems Energy storage can supply more flexibility and balancing to the grid, providing a back-up to intermittent renewable energy. Locally, it can improve the management of distribution networks, reducing costs, and improving efficiency. More in general, energy storage can provide many valuable services across the whole energy system. In fact, energy storage is essential to balance supply and demand. Peaks and troughs in demand can often be anticipated and satisfied by increasing or decreasing generation at short notice. In a low-carbon system, intermittent RES makes it more difficult to vary output, and rises in demand do not necessarily correspond to rises in RES generation. Higher levels of energy storage are thus required for grid flexibility and stability and to cope with the increasing use of intermittent wind and solar electricity [12
15]. Energy storage can be integrated at different levels of the electricity system and for different purposes: • Generation level: Arbitrage, balancing, and reserve power, etc. • Transmission level: frequency control, investment deferral, etc. • Distribution level: voltage control, capacity support, etc. • Customer level: peak shaving, time of use cost management, etc. In a future low-carbon energy system, storage will be needed at all points of the electricity system. In fact, storage serves several purposes in today’s power system, with the main following functionalities:

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

1. Transmission grid central storage (National and Europe Level): a. Balancing demand and supply: seasonal/weekly fluctuations; large geographical unbalances; strong variability of wind and solar; b. Grid management: voltage and frequency regulation; complement to classic power plants for peak generation; participate in balancing markets; cross-border trading; c. Energy Efficiency: better efficiency of the global mix, with time shift of offpeak to on-peak energy; 2. Level of distribution grid storage (City Level): a. Balancing demand and supply: daily/hourly variations; peak shaving; b. Grid management: voltage and frequency regulation; substitute existing ancillary services; participate in balancing markets; c. Energy Efficiency: demand side management; interactions grid-end-user; 3. End-user Storage (House-hold level): a. Balancing demand and supply: daily variations; b. Grid management: aggregation of small storage systems providing grid services; c. Energy Efficiency: local production and consumption; behavior change; increase usage of PV and local wind; efficient buildings; integration with district heating /cooling and CHP. In addition, the main energy storage functionalities such as energy time-shift, quick energy injection and quick energy extraction are expected to make a large contribution to security of power supplies, power quality and minimization of direct costs and environmental costs. Therefore, the dynamic behavior of storage is even more important than its long-term capacity. Energy storage can be used to mitigate these effects in terms of power quality and regulation. The storage systems are best suited for this service due to a rapid response time and high-power ramping rate, as the fluctuations require action within s to min, and a high cycling capability, because continuous operation is required.

8.3 The energy storage system in a nanogrid: the configuration The real-time management of different sources distributed throughout the territory and electrical loads requires a system of control of the power flows, therefore an exchange of information between the users and the controller, which would be difficult to achieve through a centralized system. In particular, in case this centralized system covers a very large area, the information from many users distributed throughout the territory should be collected, thereby implying a high load in terms of data exchanged and computational charges that would not guarantee security and reliability of the system, which should be ready to prevent and/or contain the dynamics of any problems in a very short time.

239

240

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

While a distributed system, unlike a centralized one, is much more efficient, also maintaining, in case of need, a coordination central system in the management of energy resources. Indeed, a distributed system is based on the use of automated control devices for the real-time management of local power flows, and the voltage control on the loads; high processing capacities that guarantee greater reliability; use of devices for adaptive control of the information transmitted, which allow easy diagnosis of transmission errors; and higher flexibility in the configuration of the communication network between the various control devices [3,16]. Smart microgrids are a solution to ensure the supply and management of electricity flows not only in areas already reached by the national distribution system, but can also be exploited in remote areas, rural areas and islands that are difficult to reach. In these areas, local solutions for the supply of electricity to groups of small users using microgrid systems represent the ideal solution. Therefore, one of the fundamental characteristics of intelligent microgrids is that of being able to operate both in grid-connected and isolated mode with respect to the distribution network, guaranteeing continuity to user loads in both the operating modes. Naturally, the control and management of the microgrid can be coordinated optimally if and only if, in addition to the generation systems and loads, also storage systems and intelligent measurement systems are integrated, which combined with appropriate strategies for control can provide solutions to different operating modes and operating conditions in which the microgrid can operate. In this context, reference is made to a configuration of a microgrid that is capable of integrating critical and noncritical electrical loads with renewable and eco-sustainable energy sources in addition to conventional ones and different storage system technologies, conventional such as Li-ION battery, flow battery and supercapacitor and/or unconventional such as Power to hydrogen, hydraulic system and biodiesel cogeneration system, by interfacing all to the distribution electricity network, if present. In order to guarantee such a solution and all the features described above, a DC distribution system is chosen which allows the various energy sources, storage systems, cogeneration systems and critical electrical loads to be interfaced by means of appropriate static power converters and, the microgrid to the distributor’s electrical grid, as illustrated in Fig. 8.1. It shows the general configuration of the microgrid, called the nanogrid [16]. The nanogrid is defined as: “an hybrid power supply system serving a single user/ real estate unit, of nominal power not exceeding 5 kW and able to feed loads in island mode, in areas without distribution network; or an hybrid power supply system serving a single user/real estate unit, of nominal power up to 100 kW, if usually is connected to distribution electric network and able to feed loads in island mode.” The ability to handle with a single system (nanogrid) different types of generation and storage systems, constitutes an important opportunity for the individual citizen to be

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

Figure 8.1 Nanogrid general configuration.

involved both Consumer (storage without generation) and Prosumer (generation and storage). In fact, the nanogrids, in grid connected mode, are the ideal instrument to satisfy the requirements of DSO therefore, can contribute to the delivery of different types of services (energy, power, and voltage regulation) and to meet needs with different time horizons ranging from the few milliseconds (ultracapacitors) days and/or months in case of Power to Gas technologies (P2G) or hydraulic storage systems of adequate size, by the time of the hour with regard to lithium technologies. The proposed architectural nanogrid solution guarantees the possibility of supply the critical electrical loads with maximum continuity and quality both in gridconnected and in stand-alone mode, mainly using RES, while conventional energy sources may be present for eventual emergency operation; the storage systems allow to store surplus energy deriving from nonprogrammable renewable sources, which cannot be immediately used by electric loads. In this way, it will be possible to produce most of the energy at local level, and, having different storage systems available, it is possible to choose the most suitable one to use according to the needs of users and the nanogrid itself. Of course, the use of electricity to power critical electrical loads must be a priority, while the power supply of noncritical electrical loads may be dependent on an analysis that can take into account the cost of energy or the storage capacity and production levels of the nanogrid [17].

241

242

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Furthermore, within the nanogrid, not only the management of electrical energy flows can be controlled, but the control systems can also include the management of thermal energy flows; this management can be implemented in an interdependent manner. Indeed, among the various technologies present in the nanogrid, there can be cogeneration systems that produce electricity and heat at the same time; there may be storage technologies in the form of thermal energy, or fuel-cell or hydrogen, which allow electrical and thermal storage; similarly in the case of energy production from biodiesel systems; of course, conventional storage systems such as lithium-ion batteries, supercapacitors and flow batteries, are also present together with the new Vehicle-togrid technologies, which support eco-sustainable energy sources (PV, wind, hydroelectric, biodiesel, cogeneration systems, etc.) [18]. Finally, the set of generation systems, storage systems, and electrical loads within the nanogrid can also be managed appropriately for the purpose of providing support services to the electricity network or to an aggregation of users, in order to achieve certain objectives, which can be for instance energy saving within the community, economic convenience, reduction of energy imbalances between the various nanogrids. The Power Electronic Interface (PEI) operates as rectifier or inverter, depending on the DC bus voltage and the power required by the aggregator. The aggregator is the intermediary between the nanogrids, that is the virtual nonsumers community—in the sense that the entire community can become essentially self-sufficient in terms of net kWhs purchased from the grid, or in some cases, become a net generator of energy- and the operators of the utility grids that is the DSO and the TSO. This intermediation is intended to provide ancillary services. In this sense, the aggregator coordinates the nanogrids so that power flows measured at each meter or POD are equal to a desired value valuated solving the Optimal energy management for the nonsumers community. The goal of the PEI control is the regulation of the power flow between the nanogrid and the utility grid. When none request from the aggregator is running, the PEI operates based on the DC bus signaling (DBS) control strategy, tracking the DC bus voltage reference. On the contrary, when the aggregator asks for a desired power flow, the PEI control switches from the voltage control to a current one to follow a current reference value.

8.3.1 The nanogrid as enabling technology Fig. 8.1 shows the more general configuration of the nanogrid: a DC network that integrates one or more sources from RES, storage systems, and electrical loads and interfaces them with the electrical network. Among the loads of the NG we can distinguish between AC electrical loads and critical loads connected to the DC bus by means of a special converter. Each unit is interfaced to the DC bus by means of a suitable power converter, DC/DC or AC/DC. In detail, taking as reference the

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

definition reported in [19], the term nanogrid will mean: “a hybrid power supply system serving a single user/real estate unit, with a nominal power of not more than 5 kW and capable of powering its loads on the island, in areas without a distribution network; or a hybrid power supply system serving a single user/real estate unit, with a nominal power of up to 100k W, if it is normally connected to the electricity distribution network and capable of powering its loads even in an intentional island.” The possibility of being able to manage different types of generation and storage systems with a single system (nanogrid), constitutes an important opportunity for the involvement of the individual citizen user, who is both Consumer and Prosumer or new end-user as consumagers (consumer with installed storage system) and Prosumagers (prosumer with integrated storage system). In fact, the nanogrids, in grid connected mode, are the ideal tool to meet the needs and requests of the DSO and, therefore, can contribute to the provision of different types of services (energy, power, and voltage regulation) and to the satisfaction of needs with different time horizons, ranging from a few milliseconds (ultracapacitors) to days and/or months in the case of P2G and biodiesel technologies or water/thermal storage systems of adequate size, passing through the hours for regarding lithium technologies. The nanogrid is interfaced to the electrical network by means of a single bidirectional DC/AC converter, previously indicated as PEI. The PEI provides the necessary flexibility to guarantee the functioning of the nanogrid as a single aggregate system, to maintain a specific quality of power and energy production. In addition, the PEI connects the nanogrid to the LV electricity grid with the aim of regulating the bidirectional power flow between the NG and the electricity grid. The PEI also plays an active role in providing ancillary services to the network for maintaining the quality of the power supply, during conditions of stress of the electrical system [20]. Another important peculiarity of the nanogrid is its ability to integrate different storage system technologies, which can: • facilitate the interconnection between the different microsources from nonprogrammable RES (solar, wind, fuel-cell, etc.) and can make them more reliable and efficient; • operate the microsources correctly, efficiently, and permanently at full power, with the surplus of energy produced available to recharge the storage systems; • compensate for sudden changes in load (supercapacitors); • facilitate the control of the flow of active and reactive power for the regulation of the voltage to the Point of Common Coupling (PCC) with the electrical network (ancillary services); • provide the “black-start” capability (provide power for plants that need electricity to start, in the absence of the main network) for power plants; • have a particular use in applications such as temporary and short interruptions of high-value industrial processes and/or plants, to support voltage, power factor correction and other aspects for the quality of the power supply.

243

244

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

A DC nanogrid is considered, taking into account that the DC microgrids (DCMGs) are characterized by a simple electrical structure with a lower number of power converters, lower system costs, and a higher efficiency respect to AC microgrids (ACMGs) [21
23]. The NG is configured as a unipolar DC distribution system with two conductors; therefore, the DC bus consists of a positive and a negative pole (the reference, 0 V). However, it can also be configured as a bipolar, positive pole and negative pole system, with reference to ground to the central of the DC bus. Furthermore, it can be classified as a remote microgrid operating only on the standalone mode but also as a utility microgrid operating both in stand-alone mode and in grid-connected mode, by providing ancillary services and a high level of integration to the electrical network itself, high quality of power, and continuity to critical loads. In case of an energy community, so as the nonsumers community, each nanogrid receives the power profile requests from an aggregator; in this case, the only electronic power converter that interfaces the nanogrid with the utility grid (PEI) regulates the power flow to the reference value communicated by the aggregator, so as to meet the needs of the utility grid or the results of the local market of the energy community. At the same time, the other nanogrid converters operate according to the DBS control strategy, responding only to the voltage variations of the DC bus. It is worth to underline that the control strategy implemented is adaptive that is the distributed DBS control is used either if the nanogrid provides a service to the aggregator—for example a balancing service—or not. The strength of the DBS control strategy is that this strategy does not change when the transition from “no service is provided” to “a service is provided” occurs.

8.3.2 Nanogrid configuration schemes with integrated energy storage systems From the study of the various most promising storage technologies to create a distributed storage system, the technologies, Li-Ion, supercapacitor, and Power-toHydrogen, have been identified that are worthy of being analyzed when integrated into a hybrid system such as the nanogrid, thus identifying the following tree hybrid system configurations, starting from the baseline configuration as illustrated in Fig. 8.2. In the baseline module, the AC/DC converter performs the functions of the PEI and the DC/DC converter is used for direct management of the MicroSource (MS). The two converters are connected via the common DC bus and are connected to the nanogrid management system (nMS) control and supervision system which manages their correct operation. In Grid-Connected configuration, the PEI has the task of managing the exchange of energy flows, in a bidirectional way, between the nanogrid and the network and ensuring compliance with regulatory requirements.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

Figure 8.2 Nanogrid configuration: (A) NG0-baseline, (B) NG1—Li-Ion, (C) NG2—Li-Ion and supercapacitor, (D) NG3—supercapacitor and power-to-hydrogen.

Furthermore, using the IEC 61850 communication standard as required by current legislation, it is possible to receive and therefore comply with a programmed energy exchange profile requested by the distributor and offer ancillary services. Furthermore, in general, an internet connection is also provided to receive other requests for exchanging profiles from a so-called “Energy Community”. In particular, the MS can be managed in Maximum Power Point Trekking (MPPT) mode, where the control logic allows the extraction of maximum power or in reduced power mode in which the extraction of power is limited for the purpose of BUS stability DC. In Fig. 8.3 the block scheme representation of NG0, NG1, NG2, and NG3 nanogrid configuration is reported.

8.3.3 Modeling and control 8.3.3.1 Modeling To design and test the type of control developed, it is necessary to model the entire system starting from the individual converters. For this reason, the different subsystems that can make up the nanogrid have been implemented, both to study the behavior of the individual converters with the relative controls and to study their interaction with

245

246

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 8.3 Nanogrid configuration block—schemes.

the entire system and between the different converters. The ANSYS Simplorer simulation software was used to implement these models. The structure of the nanogrid starts from a common DC-bus on which all the different sources are interfaced through the relative converters. Each converter is controlled with its own logic according to the voltage on the DC bus. Each storage system requires an appropriate interface device with the nanogrid, which is compliant with the dual functionality of absorption from the DC bus and electricity delivery to the DC bus. The bidirectionality of energy flows typical of storage systems must be appropriately managed through proper interfaces and controllers. From the analysis of the various configurations of DC/DC converters used to interface storage systems in the nanogrid, it was identified the solution to use that galvanically insulates the set of various storage systems, in particular for supercapacitor and Power-to-Hydrogen technologies, from the DC bus of the nanogrid. Such an architecture is particularly suitable when the individual storage systems have operating voltages between 30 and 60 V approximately, as in the cases analyzed. Following a summary of each equivalent model is provided. 8.3.3.2 PEI DC/AC converter model As regards the converter of the network interface (PEI), a four-phase bidirectional single-phase DC/AC with a Full-Bridge IGBT structure with LCL type output filter was chosen (see Fig. 8.4). 8.3.3.3 MS DC/DC converter model As for the MS converter, a two-way bidirectional DC/DC with an IGBT Half-Bridge structure with an LC-type output filter was chosen (see Fig. 8.5). 8.3.3.4 Li-Ion battery model The equivalent model of the Li-Ion Battery (see Fig. 8.6) can be found in [24]. For the simplicity of the implementation, SOC Tracking and Runtime Prediction block are not considered in the model because the SOC is calculated during the dynamic

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

Figure 8.4 PEI converter model.

Figure 8.5 MS converter model.

Figure 8.6 Li-Ion battery equivalent model.

247

248

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

simulation, and based on its value, all the parameters of the model are determined. Typical electrical parameters adopted in the model are: Voc, Rseries, Rtransient_S, Rtransient_L, Ctransient_S, Ctransient_L (see Fig. 8.7). 8.3.3.5 Li-Ion DC/DC converter In the context of DC microgrid, the connection of Li-ion batteries to the nanogrid bus can take place via a bidirectional DC-DC interface [25,26]. The most used topologies are Full-Bridge, Dual Half-Bridge (DHB), Synchronous Buck-Boost, and Dual active bridge (DAB) types. From the literature and from a technical-economic evaluation, the most topology solution used is DHB. So, the DC/DC converter that interfaces the Li-Ion Storage to the DC Bus is a two-way bidirectional buck-boost type with an IGBT Half-Bridge structure with an LC-type output filter (see Fig. 8.8).

Figure 8.7 Li-Ion battery simplorer equivalent model.

Figure 8.8 Li-Ion DC/DC converter model.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

8.3.3.6 Supercapacitor model The supercapacitor modeling is categorized in base of principal functioning as shown in Fig. 8.9. In particular, the model of a single cell is shown on the left, while the detailed modeling is shown on the right. The Ri resistance models an insulation for the manufacturing defects, and is generally very large (1000
10,000 Ω) so that the self-discharge time is very long (days or weeks) while the series resistance is of very small value (of the order of tens of milliohms). This allows very small voltage drops at the terminals (compared to the ideal plate voltage, usually 2.7 or 2.85 V) and has a limited overheating when very intense charging and discharging occur. The model described above was implemented in Simplorer with reference to the characteristic quantities reported in the manufacturer’s datasheet and the specific schematic is shown in Fig. 8.10, where the implementation of a cell stack as series composition of the base cell compatibly with the desired voltage values, and the cluster as parallels of several stacks compatibly with the overall capacity of the SC storage are illustrated. It is appropriate to specify that the overall capacity of the SC storage differs from the usable capacity and depends on the minimum voltage allowed by the converter. 8.3.3.7 SC DC/DC converter The literature reports a limited number of circuit topologies for what concerns the modes and the apparatuses of interface and connection of the supercapacitors to the nanogrid [27,28]. These are bidirectional buck-boost converters, Series Resonant Converter (SRC), DHB, and DAB. From a technical-economic evaluation, the most frequently used topological solution is the DAB as shown in Fig. 8.11. The presence

Figure 8.9 Supercapacitor equivalent model.

249

250

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 8.10 Supercapacitor simplorer model.

Figure 8.11 SC DC/DC converter model.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

of a high frequency transformer allows for a high transformation ratio while ensuring very low volumes and losses. 8.3.3.8 Power to hydrogen model The hydrogen-based storage system considered in this chapter refers to a scheme consisting of an electrolyzer which, starting from the electricity, taken from the nanogrid DC bus, produces hydrogen. Hydrogen is stored and then used by the PEM cell to produce electricity to be fed into the DC bus. The reversible PEM is composed of two equivalent electrical models: the model that represents the Fuel-Cell generator (indicated as FC) (see Fig. 8.12) and the one that represents the storage (Electrolyzer) (see Fig. 8.13).

Figure 8.12 FC equivalent model. Eid: Cell theoretical voltage; Ran: anode resistence; Rcat: cathode resistence; Rmem: membrane resistence; C: Capacitor associated with Double Layer charging during the load variation phase; i: cell current; Van: anode voltage; Vcat: cathode voltage; Vreal: output FC voltage; Rcarico: load resistence.

Figure 8.13 Electrolyzer equivalent model. Vreal: input Electrolyzer voltage; Vint: theoretical cell voltage; Ran: anode resistence; Rcat: cathode resistence; Rmemb: membrane resistence; C: Capacitor associated with Double Layer charging during the load variation phase; Gi: current generator; Van: anode voltage; Vcat: cathode voltage.

251

252

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Subsequently, a DC/DC boost converter for the FC generator and a DC/DC buck converter for the electrolyzer was introduced to adapt the output and input voltages at the equal DC bus voltage and to regulate the power flows as shown in Figs. 8.14 and 8.15. 8.3.3.9 Power-to-hydrogen (P-to-H) DC/DC converter model The type of converter used in this application is a DAB (see Fig. 8.16) where the presence of a high frequency transformer allows for a high transformation ratio while ensuring very low volumes and losses. 8.3.3.10 Control The logic of control and management of the nanogrid is based on the concept of distributed control, of the decentralized type, referred to in the literature as DBS

Figure 8.14 Electrolyzer DC/DC buck converter.

Figure 8.15 FC generator DC/DC boost converter.

Figure 8.16 P-to-H DC/DC converter model.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

[29
31]. A distributed control strategy is preferred, since the system becomes independent from a central controller, while still guaranteeing the reliability inherent in the system structure by using the DC bus itself as a communication link. The voltage measured across the converters is the only common information exchanged between the units that make up the nanogrid. At the same time, decentralized management ensures the operation of the NG independently of the other units, avoiding the need to exchange information and making the microgrid stable in all normal and critical operating conditions. Thus, the DBS allows the realization of a decentralized control system with the same advantages and reliability of a distributed control. In addition, the PEI control is designed so that the nanogrid can provide ancillary services to the electricity grid. DBS is an extension of the concept of using the charge/discharge thresholds to program individual sources in a distributed way, which induces changes in the DC bus voltage level to activate communication between the interface converters of the different sources/storage. This control strategy in the literature is generally proposed as a control strategy for modular PV systems connected to the network with or without storage systems or stand-alone systems in the presence of a renewable and nonrenewable microsources. The DBS is an extension of the concept of using charge/discharge thresholds to schedule individual sources in a distributed fashion, which induces DC bus voltage changes to activate communications between different source/storage interface converters. Like distributed control, the DBS control function is distributed among the microsource controllers that are located inside the nanogrid. However, since the communication between the microsource controllers takes place over the DC bus rather than an external communication link, the microsources are effectively controlled using terminal quantities as with a decentralized control. Thus, the DBS, respect to the centralized control strategies proposed in literature, allows the implementation of a distributed control scheme with the same reliability advantages of the decentralized control. With DBS, the source and storage interface converters operate autonomously based on the DC bus voltage level. Each converter has assigned a voltage threshold to trigger the point at which it begins discharging or charging. For an energy community, the DBS control strategy allows the operation of a nanogrid without that each single converter belonging to the nanogrid should be aware of the utility grid state and of any requests sent by the aggregator. Therefore, the power electronic converter that interfaces the nanogrid with the utility grid, namely the PEI, communicates with the aggregator in terms of power references to meet the needs of the utility grid or the results of the local market of the energy community whereas the other converters of the nanogrid operate according to the DBS control strategy answering only to DC bus voltage changes. At any time, only one converter acts as a balance node, namely it keeps the DC bus voltage regulated at the reference value (master converter in voltage control),

253

254

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

while the others supply or absorb chasing the voltage set by the master (control in current). The load converters always operate under voltage control, on the load, to keep the voltage within admissible values. As shown in Fig. 8.17, a converter is master (Master1), when the voltage of the DC bus (VDC_bus) is within a certain range. Consequently, the master converter regulates the DC bus voltage to its reference value (Vref1), whereas the other converters that function as slaves act as current-controlled converters, supplying or absorbing according to their possibilities. The Fig. 8.18 shows the different master/slave functions according to VDC_bus. Those functions have been identified at the control logic level through specific variables (Vdc_SH_H—Vdc_SH_L, Vdc_M_abs_H—Vdc_M_abs_L, Vdc_SC_ H—Vdc_SC_L, Vdc_M_inj_H—Vdc_M_inj_L, Vdc_SL_H—Vdc_SL_L) corresponding to specific threshold voltage levels respect the nominal DC bus voltage (VDC_nombus).

Figure 8.17 Functional scheme of the DBS logic.

Figure 8.18 Master/slave operating functionality.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

To favorite the control and for a stable transition from the Master state to a slave state and vice versa, hysteresis levels around the threshold values have been defined. At this scope, the hysteresis function illustrated as follows has been considered, where:

A specific function is activated or disactivated comparing the DC bus voltage with the above-mentioned threshold voltage and hysteresis levels reaching it to go up or down. Considering, for example, the Master Absorbs state, with its corresponding threshold levels (“Vdc_M_abs_H”, “Vdc_M_abs_L”) it can be reached to go up or down. At the same way, it can be left to go up or down. Then, the value Vdc_M_abs_L is associated with the two hysteresis values of lock-in (Vdc_M_abs_L_lin) and lock-out (Vdc_M_abs_L_lout) if to go up and Vdc_M_abs_H_lin and Vdc_M_abs_L_lout if to go down. So, all the nominal levels are: • Vdc_M_abs_H with Vdc_M_abs_H_lin and Vdc_M_abs_H_lout • Vdc_M_abs_L with Vdc_M_abs_L_lin and Vdc_M_abs_L_lout • Vdc_M_inj_H with Vdc_M_inj_H_lin and Vdc_M_inj_H_lout • Vdc_M_inj_L with Vdc_M_inj_L_lin and Vdc_M_inj_L_lout Moreover, considering a possible case in which the function states are not continues (dead zone), then also for the slave states the hysteresis levels must be defined as follows: • Vdc_SH_H with Vdc_SH_H_lin and Vdc_SH_H_lout • Vdc_SH_L with Vdc_SH_L_lin and Vdc_SH_L_lout • Vdc_SC_H with Vdc_SC_H_lin and Vdc_SC_H_lout • Vdc_SC_L with Vdc_SC_L_lin and Vdc_SC_L_lout • Vdc_SL_H with Vdc_SL_H_lin and Vdc_SH_H_lout • Vdc_SL_L with Vdc_SL_L_lin and Vdc_SH_L_lout Further details are provided below regarding the operational functions. 8.3.3.11 Master With the Master functionality, the Storage System is responsible for adjusting the voltage level of the nanogrid DC bus to a specific reference value defined by the following voltage thresholds: • Master Absorbs: the storage control system adjusts the DC bus voltage to the reference Vref_M_abs. This control is limited both in current respect to the values Imax_M_abs and Imin_M_abs, and the state of charge respect to SOCmax_M_abs

255

256

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

and SOCmin_M_abs. In this operating function the storage system absorbs from the DC bus. Master Injection: the storage control system adjusts the DC bus voltage to the reference value Vref_M_inj. This control is limited both in current respect to the values Imax_M_inj and Imin_M_inj, and the state of charge respect to SOCmax_M_inj and SOCmin_M_inj. In this operating function the storage system delivers to the DC bus.

8.3.3.12 Slave • High Slave: the storage system is controlled by a current control loop respect to the reference value Iref_HS and limited both in current respect to Imax_HS and Imin_HS and the state of charge SOCmax_ HS and SOCmin_ HS; • Central Slave: the storage system is controlled by a current control loop respect to the reference value Iref_CS and limited both in current respect to Imax_CS and Imin_CS and the state of charge SOCmax_CS and SOCmin_ CS; • Low Slave: the storage system is controlled by a current control loop respect to the reference value respect to Iref_LS and limited both in current Imax_LS and Imin_LS and the state of charge SOCmax_LS, SOCmin_ LS. For example, if the demand for loads varies and the sources (PV, microwind, etc) are not able to satisfy this demand, then the DC bus voltage decreases until the reference voltage of the converter of the system is reached. At that point, the DC/DC of the storage system activated, becomes master (Master2) and delivers the power necessary to meet the load demand. Then, the DC bus voltage decreases and once the threshold has been exceeded, Master1 assigns the task of adjusting the DC bus to the Master2 converter, which regulates the DC bus voltage to a new value, Vref2. It can be observed that to activate the DC/DC converter of the storage system, no communication is needed from a centralized control system or other since the converter is activated thanks to the information VDC_bus only. Of course, the DBS must be designed to maintain the energy balance and stable operation of the system in any operating condition, both in grid-connected and stand-alone configuration. For this purpose, the possible operating conditions of the nanogrid, in grid-connected and stand-alone configuration, have been analyzed and classified in operating modes, as illustrated below. Each operating mode described above (master or slave) is associated with a voltage level of the DC bus. These DC bus voltage levels are determined experimentally, as shown below. The voltage difference (ΔV) between the DC bus voltage levels, relating to the different operating modes, is based on the observations reported in [31]: 1. the ΔV must not be too small, otherwise, the nanogrid may find itself changing operating mode due to sampling inaccuracies or an external disturbance that determines the improper variation of the VDC_bus;

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

2. the ΔV must not be too large, otherwise the converters are operating in very different ways, low voltage and high current, and others at low current and high voltages, which could lead to low efficiencies and fault overcurrents. In the proposed nanogrid, microsources and storage systems must be used as a priority to maximize the use of renewable sources. To give priority to renewable sources, then these must be associated with the highest voltage levels, as the voltage threshold lowers, the priority becomes lower and lower. The sources that take the highest levels are called to intervene (deliver) earlier than the sources with lower priority, which are at the lower voltage levels. As regards instead the priorities relating to electrochemical storage systems, such as lithium and flow batteries to the charge, for example of the storage systems, this works in the opposite direction with respect to the discharge; therefore giving priority to energy storage implies assigning higher voltage levels to those who have less priority to load and assigning low voltage levels to those who have higher priority to charge or absorb, as in the case of electrical loads. In grid-connected configuration, four operating modes are available [16]: • Mode I: PEI Inverter mode; • Mode II: Microsource mode; • Mode III: Storage mode; • Mode IV: PEI rectifier mode. It is emphasized that the storage systems are used to store the energy produced by the microsources when the power generated is greater than the power demands of the local loads and this surplus cannot be injected into the network or at least cannot be injected except gradually according to the standard of interconnection to the electricity grid. The stand-alone operation configuration is enabled if the power supply fails due to a blackout or in the case of intentional islanding. Note that in this configuration, the nanogrid management and control strategy disconnects the interruptible loads on the basis of a priority table when the energy generation of the microsources plus the power available from the storage systems is less than the total demand load (local loads and AC loads). In this configuration, there are three operating modes: • Mode V: PEI Inverter mode; • Mode VI: Microsource mode; • Mode VII: Storage mode. It is important to highlight that the modes of the DBS control do not change when a transition from “no service is provided” to “a service is provided” occurs; when a transition occurs then two additional modes are added to previous five modes. The Figs. 8.19
8.21 show, for each voltage range of the DC bus, the converter that performs the function of master converter with the relative voltage imposed on the DC bus and the behavior in each range of the individual converters of the systems integrated in the nanogrid, considering a VDC_nombus equal to 400 V.

257

258

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 8.19 DBS operative mode and DC bus voltage thresholds for NG1.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

Figure 8.20 DBS operative mode and DC bus voltage thresholds for NG2.

8.4 Optimal energy management for virtual nonsumers nanogrid community 8.4.1 Virtual nonsumers community review With the “Clean Energy for all Europeans” package (in the following “Clean Energy Package”) [32], the European Union (EU) introduced new provisions on the energy

259

260

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 8.21 DBS operative mode and DC bus voltage thresholds for NG3.

market design and frameworks for new energy initiatives. Specifically, the recasts of the renewable energy directive (REDII) and the electricity market directive (EMDII) provide basic definitions and requirements for the activities of individual and collective self-consumption as well as for two types of energy communities. Specifically, “renewable energy communities” (defined in the REDII) and “citizen energy communities”

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

(defined in the EMDII), allow citizens to collectively organize their participation in the energy system. These new concepts open the way for new types of energy initiatives aiming at the empowerment of smaller actors in the energy market as well as an increased decentralized renewable energy production and consumption. Regardless the business model, in both cases the basic aim is to obtain an energy autarchy system with a reduced exchange of energy with the wholesale market, maximizing collective self-consumption (CSC), so definitively obtaining a “virtual nonsumer community” from the system point of view. An extensive review of CSC business model for PV plants is reported in [33]. CSC is addressed in article 21 of the REDII. The REDII defines “renewables selfconsumers” as well as “jointly acting renewables self-consumers” as follows: Renewables self-consumer: “a final customer who generates renewable electricity for its own consumption, and who may store or sell self-generated renewable electricity, provided that, for a nonhousehold renewables self-consumer, those activities do not constitute its primary commercial or professional activity.” The key factor in CSC is to trade the power produced by prosumer (or better prosumages) with the consumer which belonging to the community, in such a way the power exchanged outside the community is minimized. To achieve this goal two different market models may be used: negotiated market and peer to peer market. In the first model, a coordinator (third part) exists able to match the buy and sell offers of the community members. In the second market model, the energy is directly traded between buyers and sellers. In the literature, several approaches referring to such market models are reported, [34] and extensive review are available. In any case, whatever market model is suggested, opportune hardware and software enabling technologies are necessary to implement it. IoT (Internet of Things) and ICT (Information and Communication technology) Technologies, power electronics devices, and smart meters are key factors [35].

8.4.2 Mathematical model Negotiated market by the Community Energy Provider (CEP), has demonstrated its effectiveness also in pilot application by using opportune IoT technologies and nanogrids [36]. Starting from the management model proposed in [36] we can formulate a model to obtain a nonsumer community constituted by N nanogrids. The main idea is that CEP handles the community as a whole and determines the j-th nanogrid profile at a time step i (indicated as Sij ), controlling in an opportune manner the storage systems included in it, in order to maximize the community self-consumption computed over a determined time horizon (T). The starting point is Li the load or generated power negotiated by the CEP in the wholesale market, and it is calculated by the sum of the power profile of jth nanogrid (Lij) at time step i, working in autonomous self-consumption.

261

262

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

The main objective of the management model is to determine the power profile of the j-th nanogrid at time step i (Sij) so that Li has to be close as possible to zero considering the energy needs of the community and selling/purchasing price energy outside the community. Obviously, other constraints are given by the nanogrid technological constraints especially for the type of storage unit embedded in it. So, the mathematical model managing the nonsumer community, NonsuComm, may be formulated as follows: ! T X min Ci SLi i51 s.t SLi 2 Li 2

n X

j

Si $ 0 i 5 1; 2. . .T

ð8:1Þ

j51

  j j f j Xj ; Si ; Li 5 0 j 5 1; 2; . . . n j

Sminj # Si # Smaxj j 5 1; 2; . . . n

ð8:2Þ ð8:3Þ

where
Ci is the price of the wholesale market
Li is the initial community power profile at time step i
SLi is the modified community power profile at time step i
Sij is the variation of the intial profile Lij of the nanogrid j at time step i
Xj are the technical parameters of the nanogrid j
f j is the technical constraint of the jth nanogrid due to storage type and electronic devices included in it
Sminj and Smaxj determines the regulation range of jth nanogrid. Solving the NonsuComm, the profile that each nanogrid must track in order to maximize community self-consumption is given by j

j

j

Ri 5 Li 1 Si i 5 1; 2. . .:T

ð8:4Þ

8.4.3 Solution algorithms NonsuComm, in consequence of the typology of Eq. (8.2), is a linear problem or nonlinear problem. If the time step is minutes time order, linear constraints may be considered which consider mainly the charge/discharge constraints of the storage system and, consequently, linear solving optimization techniques may be used. If a very short time step is considered for example sec or lower, f is a nonlinear time dependent

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

function, and in this case more robust and effective algorithms as particle swarm optimization [37] has to be used to solve NonsuComm and consequently determine Rij.

8.5 The energy storage systems for grid ancillary service 8.5.1 The ancillary services market Ancillary services are technical requirements (services) that the electrical system needs to maintain security in its service operations and to ensure the supply of energy in the adequate conditions of security, quality, and reliability (and economic efficiency of the system) required. Currently focus mainly on frequency control, spinning reserve reduction, voltage control, and black start. Electric companies must maintain both frequency and voltage in the electrical system within narrow ranges to ensure reliability and quality of power. In some cases, the variability associated with high penetrations of renewable energy, particularly in the distribution system, can increase the potential for deviations to occur, and therefore increase requirements for load monitoring and regulatory services. Among the causes that give rise to these deviations, at least the following are distinguished: • Forecast errors or variations in consumption demand, in its active and reactive power, or in generation forecast errors, such as renewable generation with variable primary resources, such as wind and solar photovoltaic, and generation of hydroelectric power plants. past, among others. • Unplanned disconnections of generation units and/or Storage Systems, untimely consumption and/or feeders. Unexpected disconnections of elements of the transmission system that cause a detriment to security and quality of service or that cause a Partial Blackout, Total Blackout or a partition of the system in islands. • Demand variation rate (ramps), either the increase and/or decrease, with a sustained trend of system consumption in each time interval, value measured in MW/unit of time and MVAr/unit of time as appropriate. • Natural variations of renewable generation with variable primary resources, such as wind and solar photovoltaic, in time windows of a few minutes to an hour.

8.5.2 The potential benefits of using energy storage to provide ancillary services The installation of energy storage systems can be useful to provide some of these services and since they are remunerated, it is necessary to take them into account to analyze the economic feasibility of this installation [38
43]. 8.5.2.1 Frequency regulation Frequency regulation services are used to help maintain the balance between generation and load in real time.

263

264

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

For the network frequency to remain constant, the demand for energy must always be balanced with its supply. This is highly relevant as changes in frequency can cause some machines to go out of sync and fail. The balancing of continuously shifting supply and demand within a control area under normal conditions is referred to as frequency regulation. Management is frequently done automatically, on a minute-to-minute (or shorter) basis. Frequency control is beginning to pose a challenge for the integration of renewable energies in electricity grids that increasingly require faster control to cope with the disturbances that occur in it. Electric power systems are dynamic and in constant motion. Although in a first approximation it can be considered that the power generated by the primary machines is equal to that consumed plus the losses, this practically never occurs. The difference between what is generated by the primary machines and what is consumed is contributed mainly by the variation of the kinetic energy of the synchronous machines that work as generators. This results in the frequency, directly related to the speed of the generators, constantly varying. Conventional generators have a controller that permanently measures the rotational speed or frequency, and depending on this, the power delivered by the primary machine to the synchronous machine varies. This is generally implemented with a very simple control law that produces power increases proportionally to frequency drops. Furthermore, they are “isolated” electrical systems in the sense that they do not belong to a large system such as the European one. When replacing conventional generation with systems based on renewable sources, these must necessarily contribute to frequency regulation. Frequency control is carried out: 1. by the generating units connected to an electrical system by modifying their state of charge, that is, by varying the active power they inject into the network. That implies working at a suboptimal point, that is, not delivering all the available power from the energy source. In this way, there is the possibility of increasing the power in case the frequency drops. In this way, when the grid frequency is below the nominal frequency, the generating units inject active power into the grid so that the frequency returns to the nominal frequency and when the grid frequency is above the nominal they inject less active power. This mechanism is not a problem for the generation units, but the control speed required by the system operator may be a problem. For bulky generating units with high response time constants due to the mechanical inertia of the generator, rapid frequency control may sometimes require external equipment to assist in this function. Specifically, rapid frequency control does not imply severe modifications in the equipment of the new plants, but it can mean severe modifications in older plants. By means of energy storage devices, it is possible to carry out rapid frequency control in real time without modifying the installations.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

2. using storage, to be able to increase power and at the same time allow the maximum available to be always generated. This option is increasingly interesting as battery costs decrease and battery reliability increases. The fast response time and control of energy storage systems make it an ideal resource for providing regulation. Frequency control through energy storage systems is intended to be a support system for the primary control systems of the generators so that instantaneous variations in the power generated by renewable generators due to variations in the climatic conditions that affect their generation can be compensated or provide faster dynamics to systems with high time constants. Energy storage systems are very effective in reducing instantaneous frequency deviations by performing a quick compensation of the active power. For this reason, the optimal location of this type of solutions is near the generation points so that they can support the primary control systems more effectively. The active control represents exchanges of power pulses so that the Energy System Storage (ESS) absorbs or injects power to the network as required according to the control scheme. In cases where the node frequency is higher than the reference (usually nominal), the device absorbs active power (ESS load) and, if the frequency is lower than the reference frequency, injects power into the network (ESS discharge). The rapid frequency response during large power imbalances can be provided by different storage systems, such as flywheels or supercapacitors, PV plants operating with a certain reserve margin, either by operating outside the MPP (maximum power point) or through storage equipment, and wind power plants operating with or without a reserve margin. Reservoir hydroelectric plants have historically been the most efficient facilities to contribute to frequency control. Other solutions are pumped storage. For both the case of primary control and secondary frequency control services, the response is dynamically superior to the performance of conventional plants subject to a similar requirement. The response times involved are less than one second for the delivery of the committed reserve estimated between 250
500 ms—excluding the time of detection of the event. 8.5.2.2 Spinning reserve reduction The spinning reserve is a margin of power to raise that the generators must include in their generation program. One of the main reasons for the need to have this reserve is to cope with generation losses or unexpected load changes. From the point of view of the generators, the spinning reserve is a problem for various reasons, among which the fact that it forces the generators to work at nonoptimal or nonnominal operating points stands out. The use of storage allows the generators to run at full power, and in the event of having to cope with an increase in power, it is the storage that supplies it. In this way, the storage systems and their

265

266

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

associated converters can take over the spinning reserve, allowing conventional machines to work at their nominal or maximum powers. 8.5.2.3 Inertia emulation Another aspect to take into account is that for a certain imbalance between what is generated and what is consumed, the less the frequency will change the greater the inertia of the generating machines, from which it follows that their inertia is a crucial factor for maintaining the stability of the system. As the generation based on large synchronous machines, with considerable inertia, is replaced by that based on asynchronous machines (wind) or electronic converters (PV), it is expected that the total inertia of the system will decrease and therefore the disturbances in generation-consumption affect the frequency more markedly and over time. The response speed of the converters is also higher than the speed of power variation associated with the inertia of conventional generators. Therefore, with proper control, drives with their batteries can emulate the inertial response of synchronous generators. 8.5.2.4 Voltage regulation The purpose of voltage regulation is to maintain voltage levels in the electrical system by primarily providing or absorbing reactive energy at specific locations. The injection or absorption of reactive power to maintain voltage levels in the transmission and distribution system under normal conditions is known as voltage support. The systems that can participate in the provision of Voltage Control are the following: • Synchronous generating units. • Bypass connected capacitors and reactors, reactive power compensation equipment. • On-load tap changers for transformers. • Power converters equipped to provide reactive power. • Wind and solar photovoltaic generation equipped to provide such control. • Energy Storage Systems, configured to provide this ancillary service. Distributed generation, especially when it only injects power with a unit power factor, is that there are voltage increases in the feeder’s connection in the hours in which the power injection exceeds consumption. Added to this is the typical lack of voltage control means in medium and low voltage distribution networks, where control is normally carried out by modifying the transformer taps at the headend, and eventually connecting or disconnecting capacitor banks in the substation. It is possible to use batteries for voltage regulation. Since they store in direct current, the use of an inverter is necessary to connect them to the grid, so their power factor can be easily regulated using the control capabilities that the connection converters have to inject or absorb the reactive power necessary in every moment.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

The transfer and absorption of reactive power is an application for which distributed storage can be especially attractive, because reactive power cannot be transmitted effectively over long distances. Particularly advantageous is, since voltage regulation is better provided locally, the case in which the storage is distributed along the network, associated with distributed generation facilities or near areas of high consumption, since it would mean that the storage devices are close to the points that require a greater contribution of reactive power. In addition to the above, storage management for applications that are not directly related to stress control, may indirectly lead to an improvement in the stress profile of connection networks. Thus, for example, the fact of leveling the power injected by the distributed generation with the demand, avoids the appearance of the surges mentioned above. Similarly, modulating demand, avoiding hours with very high consumption peaks, helps to avoid undervoltages during those moments in the system. The contribution to voltage control is, therefore, an aspect more technically related to the connection device than to the storage equipment itself. The operating limits of the inverters are determined by the maximum current that can pass through them, that is, by the maximum apparent power. The use of storage for voltage control must be taken into account when sizing the equipment, since it is necessary to give the connection converter extra nominal power to be used in the management of reactive power. 8.5.2.5 Black start This is defined as the capacity of a generating unit to be able to join the electrical system, being initially switched off, without the need for assistance from the electrical network. It becomes important when failures arise, and it is necessary to reconnect some generators to energize the network. Some storage systems (primarily those capable of storing a considerable amount of energy) can provide this type of service. By combining a Voltage Source electronic Converter (VSC) with a battery, a three-phase voltage system can be produced. The synthesized voltages on the AC side of the converter will be able to supply loads and even the auxiliary systems of other conventional generators. In this way, an isolated system, or even a conventional power system, could be started from a storage system after a blackout (blackout) with loss of synchronism and fall of the entire generator park. Compensation for unbalanced loads: the presence of unbalanced loads in the system affects the useful life of electrical machines and reduces the efficiency and transport capacity of electrical installations. The ability of inverters associated with storage systems to balance unbalanced loads as well as the possibility of consuming energy in one phase and generating in others, allows storage systems to act for the benefit of the electrical system by contributing to load balancing, resulting in greater supply quality of the electrical system.

267

268

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

8.6 Case study 8.6.1 Problem formulation A Nonsumer community composed by twentyfour nanogrids, connected downstream to a common MV-LV substation, is considered. Each nanogrid participates to provide a service depending on its own local generation and storage to allow that the entire community can become essentially self-sufficient in terms of net kWhs purchased from the grid, or in some cases, become a net generator of energy. The equipment of each nanogrid is described in Table 8.1. As noted, other two nanogrid configurations are considered, derived from those previously defined. The configuration NG4 is a consumager, that is a consumer equipped with only a storage system without any PV plant; the NG5 is a simple consumer. The community consists of four nanogrids for each configuration, so twenty nanogrids in total. Detailed power profile of PV production and load consumption have been hypothesized for each nanogrid starting from real data monitored by smart meters installed in real-life applications, as shown in Fig. 8.22 in terms of aggregated power profiles of the NonsuComm. The NonsuComm model is therefore solved and the power profile for each 24 hours that each nanogrid (PEI profile) must track to maximize community self-consumption Table 8.1 Technical data for the generation, storage, and load.

NG0 NG1 NG2 NG3 NG4 NG5

Rooftop PV peak P [kW]

Li-Ion P [kW]

SC P [kW]

P-to-H P [kW]

LOAD P [kW]

6 3 6 6 N N

N 3 3 N 3 N

N N 3 3 N N

N N N 2 N N

3 3 6 6 3 3

Figure 8.22 PV and Load power profiles for the NonsuComm.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

is obtained as shown in Fig. 8.23 for the one nanogrid for each nanogrid configuration NG0 [1] (the first nanogrid with NG0 configuration), NG1 [1] (the first nanogrid with NG1 configuration), NG2 [1] (the first nanogrid with NG2 configuration), NG3 [1] (the first nanogrid with NG3 configuration), NG4 [1] (the first nanogrid with NG4 configuration), and NG5 [1] (the first nanogrid with NG5 configuration).

8.6.2 Simulation setup The above-illustrated power profiles have been used as input to the simulation model of each nanogrid implemented in Simplorer environment, using the parameter data specified in the following. For the Li-Ion battery the following model parameters have been considered with a nominal power of 3 kW.  • VOC ðSOCÞ 5 21:031e235 SOC 1 3:685 1 0:2156 SOC 2 0:1178 SOC2 1 0:3201SOC3  • RSeries ðSOCÞ 5 0:1562e224:37 SOC 1 0:07446  • RTransientS ðSOCÞ 5 0:3208e229:14 SOC 1 0:04669  • CTransientS ðSOCÞ 5 2 752:9e213:51 SOC 1 703:6  • RTransientL ðSOCÞ 5 0:3208e229:14 SOC 1 0:04669  • CTransientL ðSOCÞ 5 2 752:9e213:51 SOC 1 703:6 For the SC the following model parameters has been considered with a nominal power of 3 kW and an output voltage of 32
64V. C 5 3000 F Ri 5 1000 ohm Rs 5 0,29 mohm. For the P-to-H the following model parameters (see Figs. 8.24
8.25) has been considered with a nominal power of 1 kW and an output voltage of 24 V for Electrolyzer and 56 V for the FC circuit. Let us consider that at the beginning of the observation time for one of the NG2 and of the NG3 the SC SOC is equal to 90%. Moreover, for each storage system a SOCmin 5 20% and a SOCmax 5 100% have been considered.

8.6.3 Simulation results and discussions In the following the simulation results are illustrated and discussed. The DBS allows the optimal management of several converters of the NG due to the possibility to work with different operational states. This possibility allows to modulate the MS generation and the power flow from/into the grid (PEI_profile), coordinating all the operations with the charge/discharge of the storage systems so to satisfy the request. In the figures below, the PEI (indicated in figures as PCC_pw), the PV (indicated in figures as PV_power) and the Load (indicated in figures as Load_pw) power profiles, and the DC bus voltage are shown for the five nanogrids mentioned in the previous section (Fig. 8.26).

269

270

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 8.23 PV, Load, and PEI power profiles for each nanogrid.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

Figure 8.23 (Continued)

271

272

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 8.24 FC equivalent model parameter.

Figure 8.25 Electrolyzer equivalent model parameter.

Figure 8.26 NG0 simulation results: PEI_profile, PV, load power profile, and VDCbus.

As expected, being the NG0 a prosumer the surplus of PV generation or the deficit respect the load demand is injected or absorbed respectively in and from the grid and the VDC bus is stable at the nominal value of 400 V.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

In Fig. 8.27 the simulation results for the NG1 are shown. In this case, the Li-Ion battery is charged or discharged to match the PEI_profile request and the difference between the PV generation and the Load demand based on the VDCbus variation. When the DC bus voltage increases above the nominal value (403 V) because a surplus of PV generation (1407 W) exists respect to the PEI_profile injection request (237 W) plus the load demand (170 W), the SC converter is activated based on the DBS control to operate in charging phase in order to store the power of 1000 W. In Fig. 8.28 the simulation results for the NG2 are shown. In this case, the Li-Ion battery and the SC storage are charged or discharged to match the PEI_profile request and the difference between the PV generation and the Load demand based on the VDCbus variation. When the DC bus voltage drops below the nominal value (396 V) because a load demand (884 W) plus the PEI_profile injection request (156 W) is

Figure 8.27 NG1 simulation results: PEI_profile, PV, load power profile, and VDCbus.

Figure 8.28 NG2 simulation results: PEI_profile, PV, load power profile, and VDCbus.

273

274

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

higher respect the PV generation that is equal to zero, based on the DBS control, the SC converter is activated to operate in discharging phase at the power of 1000 W. On the contrary at 14:00 a PEI_profile injection request of 2733 W exists but the difference between the PV generation (4275 W) and the load demand (543 W) is higher. So, based on DBS control the storage systems must charge this difference, but the SC SOC is, as illustrated in Fig. 8.29, maximum so only the Li-Ion DC/DC converter is activated to charge a power equal to 1000 W. Also, in the case of NG3 (see Fig. 8.30), at 16:00 when the DC bus voltage increases above the nominal value (420 V) because a load demand (884 W) plus the PEI_profile injection request (1455 W) is lower respect to the PV generation (3942 W) the SC converter is activated based on the DBS control to operate in

Figure 8.29 NG2 SOC of SC.

Figure 8.30 NG3 simulation results: PEI_profile, PV, load power profile, and VDCbus.

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

charging phase. The SC SOC is though, as illustrated in Fig. 8.31, maximum, so only the FC DC/DC converter is activated to charge a power equal to 2000 W. In the case of NG4 (see Fig. 8.32) the simulation results show the contribution of Li-Ion battery to track the PEI_profile request. At 22:00, the DC bus voltage drops below the nominal voltage because the PEI_profile request is to inject 99 W into the grid so the Li-Ion battery must provide both the load power demand and the PEI_profile power. Based on the DBS control logic the Li-Ion DC/Dc converter is activated to discharging the Li-Ion battery at the power of 1000 W. Finally, in Fig. 8.33 the NG5 simulation results are shown. Being the NG5 a simple consumer the PEI_profile overlaps the load power demand exactly.

Figure 8.31 NG3 SOC of SC.

Figure 8.32 NG4 simulation results: PEI_profile, PV, load power profile, and VDCbus.

275

276

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 8.33 NG5 simulation results: PEI_profile, PV, load power profile, and VDCbus.

8.7 Conclusions Achieving growing autonomy, sustainability, and efficiency of energy systems with respect to energy needs is a strategic goal now consolidated. The distributed generation with small photovoltaic systems is a decisive factor for transforming the electrical mix, with a central role entrusted to the end-users. The storage systems, in this contest, are called to become more and more advanced in terms of technology thanks to the “smart grid ready” functions, ready to “work” in harmony with an interconnected, digitalized electrical system, open to the contributions of an increasing number of active end-users. In the chapter, the attention is focused on a nonsumer energy community consisting of several DC nanogrid with the goal to become essentially self-sufficient in terms of net kWhs purchased from the grid using several storage systems integrated in the nanogrids and opportunely managed and controlled implementing an optimization management model and the DBS control logic. The numerical results obtained show the effectiveness to satisfy the power profile valuated as output of the described optimization management model using the illustrated DBS control logic.

References [1] A. Burgio, G. Brusco, D. Menniti, A. Pinnarelli, N. Sorrentino, Energy management system for an energy district with demand response availability, IEEE Trans. Smart Grid 5 (5) (2014) 2385
2393. [2] D. Menniti, A. Pinnarelli, N. Sorrentino, G. Belli, A local market model involving prosumers taking into account distribution network congestions in smart cities, Int. Rev. Electr. Eng. 9 (5) (2014) 976
985. [3] G. Barone, G. Brusco, A. Burgio, F. Certo, D. Menniti, A. Pinnarelli, et al., A local real-time controller to face the problem of power imbalance in a VED consisting of several DC Microgrids, 17th IEEE international conference on environment and electrical engineering, Florence, Italy, (2016).

Optimal management of energy storage systems integrated in nanogrids for virtual “nonsumer” community

[4] P.D. Lund, J. Lindgren, J. Mikkola, J. Salpakari, Review of energy system flexibility measures to enable high levels of variable renewable electricity, Renew. Sustain. Energy Rev. 45 (2015) 785
807. [5] Energy Strategy Report Politecnico di Milano, (2016). [6] M. Aneke, M. Wang, Energy storage technologies and real life applications. A state of the art review, Appl. Energy 179 (2016) 350
377. [7] T.M.I. Mahlia, T.J. Saktisahdan, A. Jannifar, M.H. Hasan, H.S.C. Matselaar, A review of available methods and development on energy storage; technology update, Renew. Sustain. Energy Rev. 33 (2014) 532
545. [8] Fuel Cells and Hydrogen Joint Undertaking (FCH JU), Multi-annual work program 2014
2020. [Online] ,https://www.fch.europa.eu/page/multi-annual-work-plan.. [9] Fuel Cells and Hydrogen 2 Joint Undertaking (FCH 2 JU), Addendum to the multi - annual work plan 2014
2020. [Online] ,https://www.fch.europa.eu/page/multi-annual-work-plan.. [10] G. Guandalini, et al., Long-term power-to-gas potential from wind and solar power: a country analysis for Italy, Int. J. Hydrog. Energy 42 (2017). [11] GKN’s Solid-State Hydrogen Storage System. GKN powder metallurgy website. ,https://www.gknpm. com/globalassets/downloads/powder-metallurgy/2018/gkn-metal-hydride-based-hydrogen-storage.pdf/.. [12] European Commission Directorate-General for Energy, The future role and challenges of energy storage, DG ENER Working Paper. [13] Electricity storage and renewables: costs and markets to 2030, Irena, October 2017. ,www.irena.org.. [14] P. Clerens, Energy storage: a key flexibility resource, proceedings of IRENA South East Europe workshop on grid integration of variable renewable energy sources, Vienna, (2018). [15] O.B. Fosso, The future role of storage in a smart and flexible energy system. [16] G. Barone, G. Brusco, A. Burgio, D. Menniti, A. Pinnarelli, N. Sorrentino, A power management and control strategy with grid-ancillary services for a microgrid based on DC Bus, Int. Rev. Electr. Eng. 9 (4) (2014) 792
802. [17] G. Barone, G. Brusco, A. Burgio, D. Menniti, A. Pinnarelli, N. Sorrentino, Intentional islanding control of a Smart User Network. In Proceedings of the 2015 IEEE 15th International Conference on Environment and Electrical Engineering (EEEIC), Rome, Italy, 10
13 June (2015), 683
688. [18] G. Belli, A. Giordano, C. Mastroianni, D. Menniti, A. Pinnarelli, L. Scarcello, et al., A unified model for the optimal management of electrical and thermal equipment of a prosumer in a DR environment, IEEE Trans. Smart Grid (2017). Available from: https://doi.org/10.1109/TSG.2017.2778021. [19] P. Asmus, M. Lawrence, Making sense of new public power DER business models the business case for energy storage making sense of new public power DER business models Section 1. Navigant Research, (2016). [20] D. Menniti, A. Pinnarelli, N. Sorrentino, A. Burgio, G. Brusco, G. Belli, Design of a gridconnected photovoltaic system with grid ancillary services, Proceedings of IEEE Africon (2013), Mauritius, 9
12 September 2013. [21] S. Chakraborty, M.D. Weiss, M.G. Simoes, Distributed intelligent energy management system for a single-phase high-frequency AC microgrid, IEEE Tran. Ind. Electron. 54 (2007) 97
109. [22] X. Wang, J.M. Guerrero, Z. Chen, Control of grid interactive AC microgrids, IEEE Int. Symposium on Ind. Electronics (ISIE), 2010, (2010), 2211
2216. [23] B. Dong, L. Yongdong, Z. Zhixue, X. Lie, Control strategies of microgrid with Hybrid DC and AC Buses, Proceedings of the 2011-14th European Conference on power electronics and applications (EPE 2011), August
September (2011), 1
8. [24] M. Chen, G.A. Rincon-Mora, Accurate electrical battery model capable of predicting runtime and IV performance, IEEE Trans. Energy Convers. 21 (2) (2006) 504
511. [25] W. Phetphimoon, K. Bhumkittipich, Modeling nd simulation of bidirectional half bridge DC-DC Converter, IEEE J. [26] B. Zhao, Q. Song, W. Liu, Y. Sun, Overview of dual-active-bridge isolated bidirectional DC
DC converter for high-frequency-link power-conversion system, IEEE Trans. Power Electron. 29 (8) (2014). [27] A.G. Hidalgo Oñate, Design of an isolated DC/DC power converter to connect a low voltage supercapacitor string to a DC power system, Project thesis submitted in part fulfilment of the requirements for the degree of Master of Science Electrical and Electronic Engineering, The University of Nottingham, (2016).

277

278

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

ˇ ˇ Jakopovi´c Supercapacitors in power converter DC link a short overview of design and [28] T. Cihak, Z. application issues, IEEE J. [29] J. Bryan, R. Duke, S. Round, Decentralized generator scheduling in a nanogrid using DC bus signaling, in Proc. IEEE Power Eng. Soc. Summer Meet., June (2004), 1, 977
982. [30] J. Schonberger, R. Duke, S.D. Round, DC bus signaling: a distributed control strategy for a hybrid renewable nanogrid, IEEE Trans. Ind. Electron. 53 (5) (2006) 1453
1460. [31] K. Ng, L. Zhang, Y. Xing, J.M. Guerrero, A distributed control strategy based on DC bus signaling for modular photovoltaic generation systems with battery energy storage, IEEE Trans. Power Electron. 26 (10) (2011). [32] EU, Clean energy for all Europeans. Energy, (2019). ,https://ec.europa.eu/energy/en/topics/ energy-strategy-and-energy-union/clean-energy-all-europeans.. (accessed 19.03.2019). [33] C. Nolden, J. Barnes, J. Nicholls, Community energy business model evolution: a review of solar photovoltaic developments in England, Renew. Sustain. Energy Rev. 122 (2020). [34] T. Sousa, T. Soares, P. Pinson, F. Moret, T. Baroche, E. Sorin, Peer-to-peer and community-based markets: a comprehensive review, Renew. Sustain. Energy Rev. 104 (2019) 367
378. [35] A. Burgio, A. Giordano, A.A. Manno, C. Mastroianni, D. Menniti, A. Pinnarelli, et al., An IoT approach for smart energy districts, proceedings of ICNSC (2017), Falerna, Italy, 16
18 May 2017. [36] A. Giordano, C. Mastroianni, N. Sorrentino, D. Menniti, A. Pinnarelli, An energy community implementation: the unical energy cloud, Electron. (Switz.) 8 (12) (2019). [37] Y. del Valle, G.K. Venayagamoorthy, S. Mohagheghi, J.-C. Hernandez, R.G. Harley, Particle swarm optimization: basic concepts, variants and applications in power systems, IEEE Trans. Ev. [38] P. Denholm, et al., The value of energy storage for grid applications, Technical Report, NREL/ TP-6A20-58465 May (2013). [39] Zhou, Z., Levin, T., Conzelmann, G., Survey of U.S. ancillary services markets, United States (2016). [40] S. You, Y. Liu, Y. Liu, A. Till, H. Li, Y. Su, et al., Energy storage for frequency control in high photovoltaic power grids, United States (2019). [41] N. Wade, P. Taylor, P. Lang, J. Svensson, Energy storage for power flow management and voltage control on an 11 kV UK distribution network, 20th international conference on electricity distribution. [42] D. Williams, A. Gole, R. Wachal, Repurposed battery for energy storage in applications of renewable energy for grid applications, 24th Canadian conference on electrical and computer engineering, 446
450. [43] A. Zucker, T. Hinchliffe, A. Spisto, Assessing storage value in electricity markets, JRC scientific and policy reports. ,https://setis.ec.europa.eu/sites/default/files/reports/power-storage-report.pdf..

CHAPTER 9

Demand response role for enhancing the flexibility of local energy systems Seyed Amir Mansouri1, Amir Ahmarinejad2, Mohammad Sadegh Javadi3, Ali Esmaeel Nezhad4, Miadreza Shafie-Khah5 and João P.S. Catalão3,6 1

Department of Electrical Engineering, Yadegar-e-Imam Khomeini (RAH) Shahre Rey Branch, Islamic Azad University, Tehran, Iran 2 Department of Electrical Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran 3 Institute for Systems and Computer Engineering, Technology and Science (INESC TEC), Porto, Portugal 4 Department of Electrical, Electronic, and Information Engineering, University of Bologna, Bologna, Italy 5 School of Technology and Innovations, University of Vaasa, Vaasa, Finland 6 Faculty of Engineering of the University of Porto, Porto, Portugal

Abbreviations AC CHP DRP EES EH EHP GAMS PJM RER TES TOU

absorbed chiller combined heat and power demand response program electrical energy storage electrical heater electrical heat pump general algebraic modeling system Pennsylvania, New Jersey and Maryland renewable energy resources thermal energy storage time of use

Nomenclature Indices t s em k i Scalars ηT ηCHP =ηCHP P H Boiler η ηEH EHP ηEHP H /ηC EES ηEES /η Ch Dis ηAC

time index season index index of emission index of energy hub number index of energy hub type transformer electricity efficiency CHP electrical/ heating efficiency boiler efficiency EH efficiency EHP heating/cooling efficiency EES charging/discharging efficiency AC efficiency

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00011-X

r 2021 Elsevier Inc. All rights reserved.

279

280

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

λem λEES λTES P EES;Ch:;Max /P EES;Dis:;Max P TES;Ch:;Max /P TES;Dis:;Max P Min;CHP /P Max;CHP H Min;CHP /H Max;CHP CapMin;Boiler /CapMax;Boiler CapMin;EH /CapMax;EH CapMin;EHP /CapMax;EHP CapMin;AC /CapMax;AC CapMin;EES /CapMax;EES CapMin;CHP /CapMax;CHP Ga0 PM Max;0 μPmax NOCT T M;0 pr vci /vco /vr LPF e;sh;up / LPF e;do;up LPF ðe;h;cÞ;tr;up /LPF ðe;h;cÞ;tr;do LPF ðe;h;cÞ;cu;up / LPF ðe;h;cÞ;cu;do ϕ1 /ϕ2 /ϕ3 P Flow;max /QFlow;max l l V max i δ max i

emission cost EES operation cost TES operation cost Max charging/discharging rate of EES Max charging/discharging rate of TES Min/Max electrical power of CHP Min/Max heating power of CHP Min/Max capacity of boiler Min/Max EH capacity Min/Max capacity of EHP Min/Max capacity of AC Min/Max capacity of EES Min/Max capacity of CHP irradiation of sun at the standard condition Max power of solar panel at the standard condition thermal sensitivity of the solar panel the normal operating temperature of the solar panel solar panel temperature at the standard condition wind turbine nominal power cut-in/cut-out and rated speed of wind turbine load participation factor of shift-up/shift-down power by shiftable DR load participation factor of shift-up/shift-down power by transferable IDR load participation factor of shift-up/shift-down power by curtailable IDR rebounded load factors maximum rate of active/reactive power flow maximum rate of voltage magnitude at each bus maximum rate of voltage angle at each bus

Parameters

λBuy ki ;s;t λSell ki ;s;t Ref λki ;s;t λDR ki ;s λENS ki ;s λGas ki ;s;t EF G em EF CHP em EF Bem Gasc;s;t αinitial ki T asc;s;t vws;t

electricity purchase price electricity sell price reference tariff operation cost of DR programs penalty price of ENS gas price emission factor for upstream grid emission factor for CHP emission factor for boiler irradiation of sun initial energy factor of EES/TES temperature wind speed

Variables TOC P G-H ki ;s;t H-M P M-H ki ;s;t /P ki ;s;t H-G P ki ;s;t

total operation cost transferred power from upstream grid to energy hub transferred power between energy hubs transferred power from energy hub to upstream grid

Demand response role for enhancing the flexibility of local energy systems

f CHP ki ;s;t f Boiler ki ;s;t f EES ki ;s;t f TES ki ;s;t PENS ki ;sc;s;t PCHP ki ;s;t H CHP ki ;s;t H Boiler ki ;s;t PEES;Ch /PkEES;Dis ki ;s;t i ;s;t TES;Ch Pki ;s;t /P TES;Dis ki ;s;t EH PEH ki ;s;t /H ki ;s;t EHP EHP Pki ;s;t /H EHP ki ;s;t /C ki ;s;t AC AC H ki ;s;t /Cki ;s;t EEES ki ;s;t ETES ki ;s;t PPV ki ;s;t ind PW ki ;s;t e;sh;do Pe;sh;up ki ;s;t /P ki ;s;t e;pb;up Pki ;s;t /P e;pb;do ki ;s;t GLine /BLine l l Flow PFlow s;l;t / Qs;l;t V s;i;t δ s;i;t Pkðe;h;cÞ;tr;up / Pkðe;h;cÞ;tr;do i ;s;t2N x i ;s;t ðe;h;cÞ;cu;up Pki ;s;t /Pkðe;h;cÞ;cu;do i ;s;t

CHP fuel cost boiler fuel cost EES operation cost TES operation cost energy not served electrical power of CHP heating power of CHP heating power of boiler charging/discharging power of EES charging/discharging power of TES electrical /heating power of EH electrical /heating/cooling power of EHP heating /cooling power of AC stored energy in EES stored energy in TES available photovoltaic power available wind power shift-up/shift-down power by shiftable DR program shift-up/shift-down power by TOU DR program susceptance/conductance of network branches active/ reactive power flow of network branches voltage magnitude at each bus voltage angle at each bus shifted-up/shifted-down power by transferable IDR program shifted-up/shifted-down power by curtailable IDR program

Decision variables

I CHP ki ;s;t I Boiler ki ;s;t I EH ki ;s;t EHP;H I EHP;C ki ;s;t /I ki ;s;t AC I ki ;s;t I kEES;Ch /I EES;Dis ki ;s;t i ;s;t TES;Ch TES;Dis I ki ;s;t /I ki ;s;t e;sh;up e;sh;do I ki ;s;t / I ki ;s;t e;pb;do I e;pb;up ki ;s;t / I ki ;s;t ðe;h;cÞ;tr;up ðe;h;cÞ;tr;do I ki ;s;t /I ki ;s;t

binary binary binary binary binary binary binary binary binary binary

variable variable variable variable variable variable variable variable variable variable

of CHP operation of boiler operation of EH operation of EHP operation in cooling/heating mode of AC operation of EES operation in charging/discharging mode of TES operation in charging/discharging mode of shift-up/shift-down power by shiftable DR program of shift-up/shift-down power by TOU DR program of shift-up/shift-down power by transferrable IDR program

This chapter investigates the distributed energy resources (DER) integration to local energy systems and brings new solutions to improve the flexibility of the entire network. The main concept of the local energy system, including diverse multicarrier energy systems, is to supply the consumer’s energy demands, that is, electricity, heating, and cooling loads. In order to enhance the flexibility of the local energy system, an energy management framework is suggested in this chapter to tackle the DER’s uncertainties and enhancing the flexibility of the entire network by adopting the effects of demand response (DR) programs, as well as the effects of electrical energy

281

282

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

storage (EES) devices. The flexibility can be provided by electrical and nonelectrical energy providers. However, the effectiveness of the electrical flexibility provision is much more important than the others. Since the electrical load balance must be met instantaneously, and there is no interruption allowed, the efforts will be concentrated on the electricity flexibility provision. However, considering the flexibility of a nonelectrical system, like thermal loads in a multicarrier energy system, can improve the net flexibility provisions by the electrical system. Therefore, in the model, developed in this chapter, the flexibility provision from the whole energy system would be studied. The main contribution of this chapter is introducing a centralized framework for determining the operating points of the multicarrier energy system and improving the flexibility of the local energy systems, considering the price-based DR programs. The mentioned centralized framework can provide the desired solution for the energy vector system and energy communities, considering the flexibility from the consumer engagement in the DR programs. Moreover, the flexibility can be provided by the EES devices to the local energy systems.

9.1 Introduction The role of consumers has been significantly changed over the past two decades with the power system restructuring. Thus, local energy systems as one of the key components of distribution systems can highly impact the electricity market, system reliability, and policies [1]. Besides, the natural gas (NG) consumption has substantially increased since 2007, specifically after introducing the integrated NG and power systems operation [2,3]. NG is accounted for the cleanest fossil fuel which is easily available in most parts. It is also a fast-response energy carrier for the power plant operation. Accordingly, the integrated operation of NG and power system has captured attention recently. This is important, particularly due to the fact that not only it leads to increased flexibility and efficiency, but also helps to supply different types of load demand, such as cooling and heating loads [4,5]. However, the integrated operation of such systems is associated with severe challenges as the constraints of each system operation would affect the other one. Moreover, converting different energy carriers to other types involves some difficulties as the combined generation of heating, cooling, and electrical power is accompanied by uncertainties. Furthermore, the concern on emission policies and reliability issues has increased [6,7]. One effective solution to system reliability issues is to change the system flexibility. The flexibility can be increased by using the integrated operation of different energy systems, demand response (DR) programs, EES systems, and model predictive control tools along with renewable energies [8]. In this respect, this chapter investigates the role of DR programs and their effects on DERs, EES systems, and the operation of multicarrier energy systems. There are many research works, devoted so far to investigate the impacts of DR programs on the operating costs, emission, and load demand curve of the system. In

Demand response role for enhancing the flexibility of local energy systems

this regard, Ref. [9] presents a comprehensive model for the optimal planning and operation of the energy hub considering the uncertainties of generation and consumption. A price-based demand response program is considered in the model and the impacts of this program on equipment’ capacity and operating costs are fully investigated. Examination of the results proves that applying the DR program reduces the capacity of equipment and thus the cost of investment. The results also show that the DR program modifies the load demand curve by transferring part of the load from peak hours to off-peak hours. The authors present a mixed-integer linear programming (MILP) optimization model with the aim of increasing the flexibility of multienergy communities in [10]. In the mentioned model, all flexible equipment such as combined heat and power (CHP), electric heat pump (EHP), EB, thermal energy storage (TES) and EES are considered. Finally, the simulation results prove the effectiveness of the model. In Ref. [11], in order to improve the flexibility of electromechanical heating systems using demand response programs, a hierarchical optimization algorithm is presented. A new management model for the optimal scheduling of a multicarrier energy hub has been introduced in [12]. In the proposed hub, three types of assets have been considered: dispersed generating systems (DGs) such as micro combined heat and power (mCHP) units, storage devices such as battery-based ESSs, and heating/cooling devices such as electrical heater, heat-pumps and absorption chillers (ACs). The optimal scheduling and management of the examined energy hub assets in line with electrical transactions with distribution network has been modeled as a mixed-integer nonlinear optimization problem. In this regard, optimal operating points of DG units as well as ESSs are calculated based on a cost-effective strategy. Ref. [13] developed a scenario-based stochastic multiobjective framework to minimize the operating cost and emission of three interconnected energy hubs. The impact of price-based DR program has been studied and the epsilonconstraint technique is used to solve the problem. The results derived from the simulation show that the DR program is capable of effectively reducing the operating cost and the dependency on the upstream network. A robust optimization model has been used in Ref. [14] to address the market price uncertainties in the context of optimal scheduling of an energy hub. The studied energy hub is equipped with TES and ESS systems while taking into consideration time of use (TOU)-based and real-time pricing (RTP) based DR programs. The model addresses environmental issues and it is formulated as a MILP problem, and investigated through three different case studies. The simulation results show that RTP mechanism is associated with a more desired performance. Renewable energies are known as the most famous DERs and they have been largely integrated with local energy systems. Although renewable energies involve negligible operating cost and emission, they can bring challenges to power systems due to their intermittent power generation [15]. A conditional value-at-risk model with integrated demand response programs (DRPs) has been proposed in Ref. [16] for the optimal operation of a resourced energy hub with a

283

284

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

wind turbine and compressed-air energy storage (CAES) systems. It is noteworthy that the uncertainties of the problem are characterized using a scenario-based optimization technique and an efficient scenario reduction method is used to alleviate the number of primary scenarios. The simulation results show that the CAES system can effectively compensate for the volatile renewable power generation and improve flexibility. Besides, applying DRPs to electrical and heating load demand has successfully resulted in lowering the operating cost of the system. Ref. [17] introduced a design and operation model for an energy hub, taking into consideration shifting-load DR programs and uncertainties, relating to the load demand and wind power generation. A scenario-based stochastic programming approach along with an effective scenario reduction method in general algebraic modeling system (GAMS)—SCENRED has been employed to tackle the uncertainties. The results obtained revealed that the uncertainties cause the capacity of assets to rise which in turn leads to increased capital costs. On the contrary, applying DR programs effectively reduces the need for higher assets capacities and consequently, decreases the total cost. Ref. [18] develops an energy management model for the microgrids in the presence of renewable energies and DR programs. The studied microgrid is comprised of active loads, besides mCHP unit, an auxiliary boiler and an EES system to supply electrical and heating load demands. The DR program is based on shifting the load demand, and a scenario-based stochastic framework has been deployed to handle the impacts of the uncertainties arisen by the load demand, market price, and renewable power generation. According to the results, it can be deduced that the DR program is capable of mitigating the operating cost both in grid-connected and islanded operation modes. Ref. [19] utilizes a robust optimization framework for the optimal operation of a local energy system, including CHP unit, TES, and a boiler. The mentioned model takes into account the uncertainties and a TOU-based DR program is used which results in mitigating the operating cost of the system by modifying the consumption pattern of the consumer. System flexibility is accounted as one of the most important issues in local energy systems, where it is influenced by several direct and indirect factors. Thus, it has captured attention during recent years. In this regard, Ref. [20] presents a MILP-based scheduling model for an energy hub, considering the uncertainties caused by load demand, renewable power generation, and market price. The energy hub is resourced with a CHP unit, a wind turbine, an EES system, as well as a power-to-gas (P2G) storage system. A shifting-load DR has also been applied. The simulation results show that the P2G storage system, besides the DR program would lead to reducing the operating costs and enhancing the system flexibility. A day-ahead scheduling model has been developed for a multicarrier energy system, aimed at minimizing the energy supply cost. The studied system comprises a photovoltaic (PV) system and a wind turbine along with EES and ETS systems. A shiftable and curtailable (interruptible) loads based DR program has also been considered. The problem has been formatted as a mixed-integer nonlinear programming (MINLP) model, solved using DICOPT solver in GAMS. The

Demand response role for enhancing the flexibility of local energy systems

results obtained indicate that the DR program would impact the elastic loads which in turn leads to increased system flexibility and reduced operating cost. Ref. [21] proposes a DR-oriented operational model for a multicarrier energy system, including EES and TES systems. The objective function of the problem is a quadratic function and the problem is tackled using the genetic algorithm (GA) to optimize the operating cost. The roles of storage systems and DR program in enhancing the system flexibility and operating cost have been studied. Ref. [22] uses a two-stage model to implement the price-based residential DR programs in multicarrier energy systems. The first stage is solved to derive the received price signals. The system operator uses the results obtained from the first stage to minimize energy losses. The consumers are ensured that they would not tolerate a higher cost in the second stage than that obtained in the first stage to motivate them to participate in the DR program. The results reported by the simulation show that the mentioned model can successfully mitigate the energy losses and enhance the operational indexes. Ref. [23] presents a stochastic optimization model for the participation of a local energy system in the electricity market. In this model, flexible loads are also considered. The problem is modeled as a MILP problem and the objective function is to maximize the aggregator’s profit. Finally, the simulation results show that the proposed model is able to find the bidding curves. Ref. [24] proposes a scheduling framework for a prosumer microgrid, taking into account DR programs and an EES system. Besides, the studied microgrid is equipped with solar PV panels. The problem has been modeled as a linear programming (LP) problem solved using MATLAB. The simulation results verify the effective role of the coordinated operation of the DR program and EES system in decreasing the cost and improving the system flexibility. A tri-objective optimization model has been developed in Ref. [25] to minimize the operating cost, the expected energy not supplied, and the mismatch between the load curve and renewable power generation profile. The DR program has been included in the model and the resulted multiobjective problem has been solved using the epsilon-constraint technique. Recently and along with the substantial penetration of multicarrier energy systems in distribution networks, DR programs have also been applied to heating and cooling load demands, known as the integrated demand response (IDR). Ref. [26] provides a comprehensive review of the IDR programs, and future aspects. An integrated model has been used in Ref. [27] for the operation of a multicarrier energy system, including electricity, NG, and heat and equipped with IDR programs, P2G systems, and energy storage systems. A coordinated operation model has been suggested for the flexible loads together with other assets. Finally, the model has been simulated on two test systems and the results show that the operation of storage systems would be useful, but highly constrained by the physical limitations. On the other hand, the IDR programs are not directly impacted by technical constraints but limited by the consumer discomfort

285

286

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

index. Ref. [28] presents a two-stage MILP model for the planning of a multicarrier energy system considering the IDR program. The proposed model uses a matrix structure in which operation constraints are expressed in detail. Also, the effect of using the IDR program on equipment capacity, total cost and load demand curve has been thoroughly investigated. Ref. [29] has carried out optimal day-ahead scheduling of a hydrogen-based smart energy system using a robust optimization model, taking into account IDR programs and market price uncertainty. The IDR programs are applied to both electrical and heating loads. The hydrogen storage system is used to convert the surplus renewable power to hydrogen to supply the required hydrogen of the hydrogen-based assets. The objective function of the problem is the total cost minimization where the integrated operation of IDR programs and the hydrogen storage system would help reduce the operating costs by 7.8% and raise the system robustness against market price uncertainty by 30%. The impacts of pumped-storage units for increasing the operational flexibility of power systems have been examined in [30]. Ref. [31] developed an integrated planning model for multicarrier energy systems where the energy hub’s interconnection and IDR programs have been considered. The problem is solved, aimed at minimizing the load supply cost. The results obtained indicate that the presented model is capable of reducing the installed capacity of the assets and planning cost. A multistage LP model has been used in Ref. [32] for the operation of a multicarrier energy system, taking into consideration renewable energies, storage systems and IDR programs. It is noteworthy that IDR programs have been applied to all types of load demands. In this respect, first, the nodes arrangement and virtual nodes insertion are used to transform the complex energy hub into some simple energy hubs. Then, the coupling matrix of each simple hub is obtained. Such a technique linearizes the primary nonlinear optimization problem and significantly alleviates the computational burden. The simulation results verify that the integrated operation model can reduce the operating cost and enhance the operation robustness. This chapter first reviews the DR programs and their roles in local energy systems, particularly multicarrier energy systems. Afterward, an operation model is proposed to evaluate the impacts of DRPs, DG units, and storage systems on the system scheduling and flexibility. Moreover, the impact of each item on the operating and emission costs would be discussed.

9.2 Demand response programs for local energy systems Local energy systems had always been passive elements in power systems before introducing DR programs. Besides, they were deprived of having the chance to reduce their costs. After the power system restructuring and adding DR programs to such systems, local energy systems can actively participate in electricity markets to mitigate their costs or provide the system with ancillary services. In this respect, they

Demand response role for enhancing the flexibility of local energy systems

would help enhance reliability and mitigate the fluctuations, which in turn results in reduced energy bills [33]. Thus, DR programs have been widely accepted in many countries.

9.2.1 Comprehensive assessment of DR programs DR programs are essential for the sustainable development of electricity markets, as the interaction between the generation and demand can lead to a more competitive environment. In addition, deploying the potential of consumers to change their consumption pattern results in a more efficient market. Therefore, DR-based policies are always supported by decision-making entities. The resulting equilibrium would be required for dynamic markets and providing consumers with diverse power options. In general, DR programs can be defined as the response of end-user to market prices. The response of the end-user would be controlling the asset’s load demand, reducing the load demand, and partially/fully interrupting the lead demand. The entities that may ask for DR programs are independent system operator (ISO), service-providing entities, and distribution companies. Price response also includes RTP, dynamic pricing, critical peak pricing (CPP), TOU pricing. Demand response can be described more accurately as modifying the power consumption with respect to the usual consumption, in response to the market price or incentive to motivate the consumer to change the consumption pattern. This happens during the periods at which the energy prices are high or the system reliability is vulnerable. Overall, DR programs would be interpreted as reducing the consumption over the critical peak periods. The critical periods are those with high wholesale market prices over the day or those at which the system’s reserve is not sufficient due to any failure or extreme weather conditions. A DR program is considered a complementary action to increase system efficiency. Seven major benefits of DR programs are: enhanced system reliability, reduced operating cost, increased market efficiency, risk management, reduced environmental emissions, mitigated market power, and improved services to consumers. According to the research conducted by the electric power research institute, DR programs reduce the peak-load demand of the US by 45,000 MW, that is, 6% of the peak demand. The main challenge in properly implementing DR programs is to select the best program with respect to the type of the load and system conditions. Fig. 9.1 shows different types of DR program. As this figure depicts, DR programs are categorized into pricebased end incentive-based ones. 9.2.1.1 Price-based Demand Response Programs A price-based DRP leads to substantial modifications in the power consumption patterns in response to the market price variations. These DRPs are categorized into TOU mechanism, RTP, and CPP. If the variations in the market price over different

287

288

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Time of use pricing program

Price-based DR programs

Real time pricing program

Critical peak pricing program

DR programs Direct load control program

Interruptible/ curtailable load control program Demand bidding/ buyback program Incentive-based DR programs Emergency program

Ancillary services market program

Capacity market program

Figure 9.1 DR programs classification [34].

time intervals of the day are considerable, consumers would react to price signals and reduce their bill [35]. • Time-of-use pricing TOU pricing is the most used mechanism all over the world. By using this mechanism, the 24 hours of the day are divided into three or four periods as: peak, offpeak, and valley periods. Each of these periods is associated with a fixed price. These prices may vary for different hours of the day, different days of the week, or seasons of the year. The differences in the prices are the incentive for consumers to reduce their consumption or shift their leads to other periods. These programs are mandatory and arbitrary programs. Consumers are able to participate in arbitrary programs and give up after the agreed period. Mandatory programs are designed for all consumers

Demand response role for enhancing the flexibility of local energy systems

and they have to participate. Once the consumer tends to reduce the consumption over the peak periods and shift the peak load to off-peak hours, the load factor improves, and prices would drop in most cases. The main point in implementing such a program is to precisely measure the consumption, issuance of electricity bills, and training consumers. Hence, advanced energy meters are required for each consumer. These meters need smart systems and advanced calculations to issue the electricity bill. Recently, the way priced-based DR programs are implemented has substantially changed with the recent advances in internet technology. Accordingly, advanced digital meters with advanced communication systems, providing the consumers with the several capabilities to observe their consumption and decide on shifting their load demand, will be installed and used [36]. • Real-time pricing RTP is another price-based DRP, with hourly-varying pricing. The type of this DR program is arbitrary. Once the consumers enter this program, they should continue with the contract for a given period. The more substantial the variations of the market prices, the more the load shifting of consumers will be [37,38]. • Critical peak pricing The CPP is a combination of the TOU and RTP mechanisms. CPP is associated with a predetermined high price designed by distribution companies to apply over peak intervals. These tariffs are called for a limited number of days or hours of the day with relatively short cautions. The consumers will receive a price discount over offpeak hours in this mechanism. It is worth mentioning that these tariffs are not yet common and used only in some regions [39]. 9.2.1.2 Incentive-based Demand Response Programs These programs are planned by distribution companies, service-providing entities, and local system operators with respect to the price consideration and the specific features of generators and the system. These programs offer some incentives to consumers to reduce or shift their load demand. Such incentives may be constant or variable in time. Once, the system reliability is vulnerable or market prices are too high, the load demand should be reduced. It is noteworthy that the consumers that are not able to commit to their contracts, would be penalized. Incentive-based DR programs provide the system operator with different options to solve the market problems of a region. These programs would help solve the system reliability issues. For instance, centralized loads would be mitigated to reduce transmission system congestion. These DR programs include direct load control (DLC), interruptible/curtailable services, demand bidding/buyback programs, emergency DRPs, capacity market programs, and ancillary service market programs [40]. • Direct load control The consumers having assets with the capability to turn off or be used for a shorter period of time can participate in DLC programs. Some of these assets are residential central air conditioning systems, boilers, electric pumps and electric heaters (EHs). In

289

290

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

•

•

•

•

this respect, consumers should be equipped with a telecommunication receiver/transmitter to be able to participate in this program. It is noteworthy that once consumers opt to participate in the DLC program, they cannot quit the program. Interruptible/curtailable services Interruptible services can be utilized by ISOs in the cases where there is not sufficient operating reserve in restructured power systems. In these programs, consumers make a contract with the service provider to change the power consumption with previously-provided information. Demand bidding/buyback programs Demand bidding/buyback programs can be used when a consumer decides to give up consuming electricity with a predetermined price. Such programs are arbitrary as the consumer is able to determine the amount and the time at which he/she tends to participate. These programs were introduced in 1993 and they are available as a DR program. By using this mechanism, if consumers make a contract with the service provider and determine the amount of their load to reduce, they can make a higher profit, compared to the case without these programs. From the economic point of view, it can be said that the profit made by reducing the load demand is higher than the cost of energy purchased from generation companies with high prices. Emergency Demand Response programs Once the system’s reserve decreases, some consumers reduce their load demand and receive incentives instead. These consumers are end-users and load aggregators. The end-users usually include large industrial and commercial units that can reduce their load demand for at least 100 kW during emergency cases. These programs are deployed for cases with vulnerable reliability. These programs are similar to DLC programs in terms of the communication systems and actions taken to reduce the load demand. The consumer receives the incentive immediately after verification of the action taken. Capacity market programs Consumers offer the load curtailment as the capacity of the system to replace the conventional generation. In this respect, consumers provide a prespecified interruptible load to deal with contingent events and fluctuations. If they commit to their contracts when it is needed, they receive incentives; otherwise, they will be penalized. Ancillary service market programs As it is obvious, consumers offer their interruptible or shiftable load in the ancillary service market as the operating reserve. In case they have to interrupt or reduce their load demand, they will be paid by the ISO according to the spot market price. In the past, Pennsylvania, New Jersey and Maryland (PJM) ISO and other ISOs relied only upon generation units to provide the required ancillary services. However, today there are numerous reliable sources other than generation units to provide the system with fast response services.

Demand response role for enhancing the flexibility of local energy systems

• IDR programs for multicarrier energy systems IDR programs are used in multicarrier energy systems, planned to simultaneously supply electrical, heating, and cooling load demands [41]. The operation of these systems involves more constraints while some of them are interrelated. Thus, it is of high significance to prevent the coincidence of their peak loads for the sake of having an efficient operation. Accordingly, IDR programs are used and applied to all the three mentioned types of load demands simultaneously. In this respect, they are applied to each load demand separately through the power balance equations. It is worth noting that participating in IDR programs is associated with more limitations compared to the case with DR programs. Thus, different methods are used to implement these programs. Load-based IDR programs would be more desired than price-based IDR programs, as they may cause some other problems. For instance, if the three types of load demands are simultaneously shifted to another time interval, a new peak load would be created. The consumer will receive an incentive using this mechanism from the system operator. Generally, there are three types of load-based IDR programs, utilized according to the application and the consumer’s behavior. These three load-based IDR programs are as follows: • Shiftable IDR Consumers shift their peak-load demand to off-peak hours through this program and receive an incentive. The mechanism of this program is similar to that of shiftable DR while it is used for all load demand types. It should be noted that the total amount of load demand would remain unchanged and only a fraction of the load demand is shifted. • Transferable IDR The load demand is shifted to other intervals using the transferable IDR programs. In this respect, the time interval between these shifts remains fixed, that is, the load demand reduces in a time interval and after a specific period, it increases by the reduced amount (e.g., 8 hours). Consequently, the load demand would be transformed from peak intervals to valley intervals. It is noted that the maximum transferable load is different for each type. • Curtailable IDR The load demand can be curtailed at a specific time using this program and it will be rebounded over the subsequent time intervals (e.g., 3 hours), immediately after the load curtailment. For example, if the load is curtailed at hour 18, it will be rebounded at hours 19, 20, and 21. The percentage of the load rebounded would be logically estimated. For instance, if a consumer curtails a fraction of the load demand at hour 18, the most probable hour to rebound the load demand is hour 19. Then, the remaining load demand will be rebounded over hours 20 and 21. As mentioned above, each of the three IDR programs can be employed. However, shiftable IDR programs are the most prevalent ones.

291

292

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

9.3 Flexibility assessment of local energy systems in the presence of energy storage systems and DR programs System flexibility is accounted as one of the most important issues in local energy systems, impacted by some factors both directly and indirectly. The most influencing factors are the uncertainties in the load demand and generation, DG units and energy storage systems. Moreover, the connection between the local energy systems and IDR programs highly affects the system flexibility. The negative impacts of renewable energies’ uncertainties on system flexibility can be effectively addressed by storage systems, P2G converters, and model predictive control methods. TES, EES, and cooling TES systems can be utilized in local energy systems. These devices would help enhance the system flexibility. CAES system has also been widely used to provide the system with efficient electromechanical energy storage. These devices store the compressed air by consuming electricity and discharge the compressed air to produce electricity when it is needed. In general, the principle of storage systems operation is on the basis of absorbing electricity to charge over off-peak and valley periods and discharge over peak periods to improve the system reliability by load flattening. P2G converters have also been used to use the surplus power generation, particularly surplus renewable power generation, to produce gas and improve the system efficiency and flexibility. DR programs can also be applied to upgrade the system flexibility by shifting the peak load demand to other periods and reducing the total operating cost. Such programs would also help stabilize the voltage at the connection point of consumers and improve the reliability by alleviating the peak load demand. If DR programs are applied to other types of load demand rather than only electrical loads, their impact on the system flexibility would be more highlighted.

9.4 Energy management framework for DER integrated distribution networks This section presents an energy management system for multiple local energy systems, connected to a 33-bus distribution network. These local energy systems include three residential, industrial, and commercial energy hubs with transactive energy trading capability. In this respect, each hub is equipped with a CHP unit to evaluate the impacts of DERs on the system, besides renewable energy sources (RES’s) scheduling and flexibility. Moreover, the residential and commercial hubs include PV panels and the industrial hub includes a wind turbine. Furthermore, TES and EES systems have been used to investigate their roles in the optimal operation of energy hubs. The impacts of DR programs on the system scheduling and flexibility have been studied through considering five different programs. In this respect, two conventional DR

Demand response role for enhancing the flexibility of local energy systems

programs, applied to electrical loads and three IDR programs, applied to heating and cooling loads, have been used in this chapter. It is noteworthy that the network constraints and power flow of lines have been modeled to avoid any unreal power transaction among the energy hubs. Fig. 9.2 illustrates the conceptual model of the energy hubs in this study including industrial, commercial, and residential energy hubs. The presented problem has been modeled as a single-objective optimization problem as follows: • Objective function The objective function of the problem is expressed in Eq. (9.1a), comprised of the operating costs of generation and storage assets. The first and second items of this function state the energy purchase cost and profit of each hub, made by selling energy relating to each hub respectively. fkCHP , fkBoiler , fkEES , and fkTES show the i ;sc;s;t i ;sc;s;t i ;sc;s;t i ;sc;s;t operating costs of CHP units, boiler, EES, and TES systems respectively. PkENS and i ;sc;s;t ENS λki ;s are the amount of energy not supplied (ENS) and its corresponding cost respectively. The last part of the objective function also shows the cost due to emission, propagated by the CHP unit and boiler, and also the cost due to transacting power with the upstream grid. SO2, CO2, and NO2 emissions have been taken into account in this chapter. ! 3   PkG-H 1 PkMi ;s;t-H Buy i ;s;t H-G H-M T Sell η λki ;s;t λki ;s;t 2 Pki ;s;t 1 Pki ;s;t 7 6 ηT 7 Ns NT X 6 X X 7 6 ðe;h;cÞ1 ðe;h;cÞ2 ENS DR 7 6 CHP Boiler EES TES ENS Min:TOC 5 ωs 6 1 fki ;s;t 1 fki ;s;t 1 fki ;s;t 1 fki ;s;t 1 Pki ;sc;s;t λki ;s 1 ðPki ;s;t 1 Pki ;s;t Þλki ;s 7 7 s51 t51 kiAK 6 EM 5 4 X G G-H CHP CHP B 1 λem EFem Pki ;s;t 1 λem EFem Pki ;s;t 1 λem EFem HkBoiler i ;s;t 2

em51

ð9:1aÞ • Operating cost of energy hubs’ assets Eqs. (9.1b)(9.1e) indicate the operating cost of CHP units, boiler, EES, and TES systems respectively [42]. Eq. (9.1b) shows that the operating cost of the CHP units is a function of the heat and power generation, as well as the NG price. It is noted that PkCHP and HkCHP are the electrical power and heat generation of the i ;sc;s;t i ;sc;s;t CHP units respectively, while their associated efficiencies are stated by ηCHP and P CHP ηH respectively. The operating cost of the boiler is expressed as the product of the NG consumption and the NG price. The heat generation of the boiler, its efficiency, and the NG price are denoted by HkBoiler , ηBoiler , and λGas ki ;s;t respectively. As i ;sc;s;t Eqs. (9.1d)(9.1e) indicate, the operating costs of EES and TES systems depend upon the charging and discharging power and their operation duration.  CHP  Pki ;s;t HkCHP CHP i ;s;t ð9:1bÞ fki ;s;t 5 CHP 1 CHP λGas ki ;s;t ηP ηH

293

294

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 9.2 Conceptual model of energy hubs in this study.

Demand response role for enhancing the flexibility of local energy systems

! Boiler H ki ;s;t λGas fkBoiler 5 ki ;s;t i ;s;t ηBoiler

ð9:1cÞ

  EES;Ch EES;Dis EES fkEES 5 λ P 1 P ki ;s;t ki ;s;t i ;s;t

ð9:1dÞ

  TES;Ch TES;Dis TES 5 λ P 1 P fkTES k k ;s;t ;s;t ;s;t i i i

ð9:1eÞ

• CHP model The CHP operation is characterized using the equalities and inequalities presented in Eqs. (9.2a)(9.2e). The CHP generation is limited as expressed in Eq. (9.2a). Besides, the power and heat outputs are also constrained as Eqs. (9.2b) and (9.2c) respectively. It is noted that IkCHP is a binary variable, specifying the turn off/on status of the CHP i ;sc;s;t unit. Expression (9.2d) and (9.2e) show the power and heat flow of this asset. # PkCHP 1 HkCHP # CapMax;CHP IkCHP CapMin;CHP IkCHP i ;s;t i ;s;t i ;s;t i ;s;t

ð9:2aÞ

# PkCHP # P Max;CHP IkCHP P Min;CHP IkCHP i ;s;t i ;s;t i ;s;t

ð9:2bÞ

H Min;CHP IkCHP # HkCHP # H Max;CHP IkCHP i ;s;t i ;s;t i ;s;t

ð9:2cÞ

PkCHP 5 PkCHP-EL 1 PkCHP-EES 1 PkCHP-EHP 1 PkCHP-EH i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t 1 PkCHP-M 1 PkCHP-G i ;s;t i ;s;t 5 HkCHP-HL 1 HkCHP-AC 1 HkCHP-TES HkCHP i ;s;t i ;s;t i ;s;t i ;s;t

ð9:2dÞ

ð9:2eÞ

• Boiler model The boiler model is characterized using Eqs. (9.3a) and (9.3b) where the heat generation is constrained as shown in Eq. (9.3a) and the heat flow is stated in Eq. (9.3b). It should be noted that CapMax;Boiler and CapMin;Boiler are the lower and upper bounds of the heat generation, where the minimum heat generation is assigned to model “zero” in this chapter. Besides, IkBoiler is a binary variable, deteri ;sc;s;t mining the turn/off status of the boiler. As Eq. (9.3b) indicates, the heat generated by the boiler can be directly used by the consumer or indirectly used by the AC to supply the cooling power. Otherwise, it can be stored in the TES system. CapMin;Boiler IkBoiler # HkBoiler # CapMax;Boiler IkBoiler i ;s;t i ;s;t i ;s;t

ð9:3aÞ

295

296

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

HkBoiler 5 HkBoiler-HL 1 HkBoiler-AC 1 HkBoiler-TES i ;s;t i ;s;t i ;s;t i ;s;t •

Electrical heater model The equalities and inequalities presented in Eqs. (9.4a)(9.4d) are used to model the EH operation. The heat generated by the EH is limited as shown in Eq. (9.4a) while Eq. (9.4b) represents the heat generation equation of the EH as the product of electricity consumption and the respective efficiency. The power and heat flow equations of the EH are stated in Eqs. (9.4c) and (9.4d) respectively. CapMin;EH IkEH # HkEH # CapMax;EH IkEH i ;s;t i ;s;t i ;s;t

ð9:4aÞ

HkEH 5 PkEH ηEH i ;s;t i ;s;t

ð9:4bÞ

5 PkG-EH 1 PkCHP-EH 1 PkEES-EH 1 PkRES-EH 1 PkMi ;s;t-EH PkEH i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t HkEH 5 HkEH-HL 1 HkEH-TES i ;s;t i ;s;t i ;s;t •

ð9:3bÞ

ð9:4cÞ ð9:4dÞ

Electric heat pump The heating power generation and cooling power generation equations of the EHP are represented in Eqs. (9.5a) and (9.5b) respectively. It is noteworthy that the heating power and cooling power and their associated binary variables are denoted by HkEHP and CkEHP , IkEHP;H , and IkEHP;C respectively. The EHP is capai ;sc;s;t i ;sc;s;t i ;sc;s;t i ;sc;s;t ble of operating in one of the heating and cooling modes at a time. In this regard, the conflicting conditions have been avoided using Eq. (9.5c). The heating power and cooling power generation equations are the functions of the electricity consumption and respective efficiencies, as indicated in Eqs. (9.5d) and (9.5e) respectively. The electrical power, heating power, and cooling power flow equations are expressed in Eqs. (9.5f)(9.5h) respectively. # HkEHP # CapMax;EHP IkEHP;H CapMin;EHP IkEHP;H i ;s;t i ;s;t i ;s;t

ð9:5aÞ

CapMin;EHP IkEHP;C # CkEHP # CapMax;EHP IkEHP;C i ;s;t i ;s;t i ;s;t

ð9:5bÞ

0 # IkEHP;H 1 IkEHP;C #1 i ;s;t i ;s;t

ð9:5cÞ

HkEHP 5 PkEHP ηEHP H i ;s;t i ;s;t

ð9:5dÞ

CkEHP 5 PkEHP ηEHP C i ;s;t i ;s;t

ð9:5eÞ

Demand response role for enhancing the flexibility of local energy systems

PkEHP 5 PkG-EHP 1 PkCHP-EHP 1 PkEES-EHP 1 PkRES-EHP 1 PkMi ;s;t-EHP i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t

ð9:5f Þ

5 HkEHP-HL 1 HkEHP-TES HkEHP i ;s;t i ;s;t i ;s;t

ð9:5gÞ

CkEHP 5 CkEHP-CL i ;s;t i ;s;t

ð9:5hÞ

• Absorption chiller model The constraints and energy flow of the AC are stated in Eqs. (9.6a)(9.6d). The minimum and maximum limits of the generation of the AC are represented in Eq. (9.6a). Eq. (9.6b) shows the cooling power generation of the AC, depending upon the input heating power and the associated efficiency. The input heating energy flow and the output cooling energy flow are shown in Eqs. (9.6c) and (9.6d) respectively. # CkAC # CapMax;AC IkAC CapMin;AC IkAC i ;s;t i ;s;t i ;s;t

ð9:6aÞ

CkAC 5 HkAC ηAC i ;s;t i ;s;t

ð9:6bÞ

5 HkCHP-AC 1 HkBoiler-AC 1 HkTES-AC HkAC i ;s;t i ;s;t i ;s;t i ;s;t

ð9:6cÞ

CkAC 5 CkAC-CL i ;s;t i ;s;t

ð9:6dÞ

• EES and TES model The EES and TES systems are modeled using the constraints represented in Eqs. (9.7a)(9.7k). The energy stored in the EES system is limited as Eq. (9.7a). The energy balance of this device is expressed in Eq. (9.7b) which is the function of the energy available at time t 2 1 and the charging and discharging power. The charging power and discharging power of the EES system should be in the feasible operating interval of this device as shown in Eqs. (9.7c) and (9.7d) respectively. As constraint Eq. (9.7e) emphasizes, the EES system is capable of operating in either charging or discharging modes at a time. The initial energy and final energy stored in the device are also limited as stated in Eqs. (9.7f) and (9.7g) respectively. Eqs. (9.7h) and (9.7i) indicate the energy flows of the EES system in the charging and discharging modes respectively. The energy flows of the TES system are indicated in Eqs. (9.7j) and (9.7k) respectively. # CapMax;EES CapMin;EES # EkEES i ;s;t

ð9:7aÞ

297

298

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning



EkEES 5 EkEES 1 PkEES;Ch ηEES Ch i ;s;t i ;s;t21 i ;s;t



PkEES;Dis i ;s;t 2 ηEES Dis

!

0 # PkEES;Ch # P EES;Ch;Max IkEES;Ch i ;s;t i ;s;t

ð9:7cÞ

0 # PkEES;Dis # P EES;Dis;Max IkEES;Dis i ;s;t i ;s;t

ð9:7dÞ

0 # IkEES;Ch 1 IkEES;Dis #1 i ;s;t i ;s;t

ð9:7eÞ

EkEES 5 EkEES i ;s;t5T i ;s;t50

ð9:7f Þ

5 αinitial CapMax;EES EkEES ki i ;s;t50

ð9:7gÞ

-EES 5 PkG-EES 1 PkCHP-EES 1 PkPV 1 PkMi ;s;t-EES PkEES;Ch i ;s;t i ;s;t i ;s;t i ;s;t

ð9:7hÞ

PkEES;Dis 5 PkEES-EL 1 PkEES-G 1 PkEES-EHP 1 PkEES-EH 1 PkEES-M i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t

ð9:7iÞ

HkTES;Ch 5 HkCHP-TES 1 HkBoiler-TES 1 HkEH-TES 1 HkEHP-TES i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t

ð9:7jÞ

5 HkTES-HL 1 HkTES-AC HkTES;Dis i ;s;t i ;s;t i ;s;t •

ð9:7bÞ

ð9:7kÞ

Renewable energies model Eq. (9.8a) shows the power generation equation of the solar PV panels, in a M a is the power output of the panel. Besides, Gsc;s;t , G0a , PMax;0 , Tsc;s;t , which PkPV i ;sc;s;t NOCT , and TM ;0 are the hourly solar irradiance, standard solar irradiance, nominal capacity of the PV system, hourly temperature, normal operating cell temperature, and the standard temperature respectively. The power produced by the wind turbine follows the conditional equation presented in Eq. (9.8b). The renewable energy flow equations are shown in Eqs. (9.8c) and (9.8d). PkPV i ;sc;s;t

   a Gsc;s;t NOCT 2 20 M a a 2 TM ;0 5 PMax;0 1 μPmax Tsc;s;t 1 Gsc;s;t 800 G0a

ð9:8aÞ

Demand response role for enhancing the flexibility of local energy systems

8 0 > > >  w 3 > < v 2v pr vs;tr 2vcici PkWind 5 i ;s;t > > > p > : r 0

w # vci vsc;s;t w vci # vs;t # vr w vr # vs;t # vco w vs;t $ vco

5 PkPV 1 PkWind PkRES i ;s;t i ;s;t i ;s;t

ð9:8bÞ

ð9:8cÞ

PkRES 5 PkRES-EL 1 PkRES-EES 1 PkRES-EHP 1 PkRES-EH 1 PkRES-M 1 PkRES-G i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t ð9:8dÞ • Energy transaction between hubs This section includes the power flow equations for the transactive energy trading between the three hubs. As Eqs. (9.9a)(9.9c) show, the energy transaction of each hub would be determined with respect to other hubs. Moreover, the power received by each hub from other hubs can be specified using Eqs. (9.9d)(9.9f). Ind-Com Ind-Res 5 Ps;t 1 Ps;t ; i 5 Industrial PkH-M i ;s;t

ð9:9aÞ

Com-Ind Com-Res PkH-M 5 Ps;t 1 Ps;t ; i 5 Comercial i ;s;t

ð9:9bÞ

Res-Ind Res-Com 5 Ps;t 1 Ps;t ; i 5 Residential PkH-M i ;s;t

ð9:9cÞ

Res-Ind Com-Ind PkMi ;s;t-H 5 Ps;t 1 Ps;t ; i 5 Industrial

ð9:9dÞ

Res-Com Ind-Com PkMi ;s;t-H 5 Ps;t 1 Ps;t ;

i 5 Comercial

ð9:9eÞ

Ind-Res Com-Res 1 Ps;t ; i 5 Residential PkMi ;s;t-H 5 Ps;t

ð9:9f Þ

• Demand response programs models This section provides the mathematical formulation of two conventional DR programs, including a price-based DR program and a transferrable load-based DR program as well as three IDR programs. These IDR programs include shiftable, transferrable, and curtailable programs. It should be noted that the mentioned IDR programs are applied to electrical, heating, and cooling loads. Furthermore, the impact of each DR and IDR program would be individually studied.

299

300

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

Shiftable demand response program By using this mechanism, consumers receive an incentive and agree to shift their peak-load demand to off-peak hours. The mathematical relationships, given in Eqs. (9.10a)(9.10d) show the mechanism of this program. As Eq. (9.10a) indicates the sum of the amount of reduced load, and increased load demands over the scheduling period should be equal. Constraints given in Eqs. (9.10b) and (9.10c) show the maximum hourly increase and decrease in the load demand. Constraint given in Eq. 10(d) state that the simultaneous increase and decrease in the load demand as a result of the DR program is impossible and should be avoided. T X

e;sh;up

Pki ;s;t 5

t51

T X

ð9:10aÞ

Pke;sh;do i ;s;t

t51

e;sh;up

e;sh;up

0 # Pki ;s;t ðsc; s; tÞ # LPF e;sh;up Pks i ;s;t Iki ;s;t 0 # Pke;sh;do # LPF sh;do Pkei ;s;t Ike;sh;do i ;s;t i ;s;t e;sh;up

0 # Ike;sh;do 1 Iki ;s;t # 1 i ;s;t •

ð9:10bÞ ð9:10cÞ ð9:10dÞ

Time-of-use demand response program The mechanism of the TOU program is stated in Eqs. (9.11a)(9.11f). Eq. (9.11a) emphasizes that the sum of the increased load demand and reduced up load demand must be equal over the scheduling period. Dki ;sc;s;t and Dkdoi ;sc;s;t denote the upward and downward load demand respectively, depending upon the hourly electricity price and load elasticity, modeled in Eqs. (9.11b) and up (9.11c). εki and εdo ki indicate the upward and downward load demand elasticities ref respectively. Moreover, λBuy ki ;s;t and λ ki ;s are the hourly electricity price and offpeak electricity price respectively. Constraints given in Eqs. (9.11d) and (9.11e) show the upper and lower bounds of the decrease in the load demand due to the up DR program. Bki and Bdo ki are the maximum upward and downward load variation coefficients, stated in terms of a percentage of the electrical load demand. up Iki ;sc;s;t and Ikdoi ;sc;s;t are upward and downward load demand variation binary variables respectively, where the conflicting situation is avoided in constraints given in Eq. (9.11f). T X t51

e;pb;up

Pki ;s;t 5

T X t51

e;pb;do

Pki ;s;t

ð9:11aÞ

Demand response role for enhancing the flexibility of local energy systems

e;pb;up Pki ;s;t $ εup

e;pb;do Pki ;s;t $ εdo

 Pki ;s;t 1 2

 Pki ;s;t 1 2

πNet ki ;s;t

! ð9:11bÞ

Ref

πki ;s

πNet ki ;s;t

! ð9:11cÞ

Ref

πki ;s

e;pb;up

e;pb;up

ð9:11dÞ

e;pb;do

e;pb;do

ð9:11eÞ

0 # Pki ;s;t # Pkei ;s;t  Bup  Iki ;s;t 0 # Pki ;s;t # Pkei ;s;t  Bdo  Iki ;s;t e;pb;do

e;pb;up

0 # Iki ;s;t 1 Iki ;s;t # 1

ð9:11f Þ

• Shiftable IDR program The shiftable IDR program is modeled using the mathematical formulations presented in Eqs. (9.12a)(9.12d). Eq. (9.12a) states that the load demand should remain constant over the scheduling period, that is, the sum of increases and decreases should be equal. The hourly upward load demand and downward load demand have been characterized through Eqs. (9.12b) and (9.12c) respectively, while constraint given in Eq. (9.12d) removes the conflicting situation. T X

ðe;h;cÞ;sh;up

Pki ;s;t

t51 ðe;h;cÞ;sh;up

5

T X

Pkðe;h;cÞ;sh;do i ;s;t

ð9:12aÞ

t51 ðe;h;cÞ;sh;up

ð9:12bÞ

# LPF ðe;h;cÞ;sh;do Pkðe;h;cÞ Ikðe;h;cÞ;sh;do 0 # Pkðe;h;cÞ;sh;do i ;s;t i ;s;t i ;s;t

ð9:12cÞ

0 # Pki ;s;t

# LPF ðe;h;cÞ;sh;up Pkðe;h;cÞ Iki ;s;t i ;s;t

ðe;h;cÞ;sh;up

1 Iki ;s;t 0 # Ikðe;h;cÞ;sh;do i ;s;t

#1

ð9:12dÞ

• Transferrable IDR program The mathematical relationships, proposed in Eqs. (9.13a)(9.13d) are employed to model the transferrable IDR program. In this respect, Eq. (9.13a) shows that this program should be applied with a determined pace, for example, 8 hours in this chapter. In other words, if the load increases or decreases, the same amount should be compensated after 8 hours. This hourly upward and downward transferrable

301

302

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

load demands are modeled using constraints given in Eqs. (9.13b) and (9.13c). Furthermore, the conflicting situation is avoided using constraint given in Eq. (9.13d). ðe;h;cÞ;tr;up

Pkðe;h;cÞ;tr;do 5 Pki ;s;t1Nx i ;s;t ðe;h;cÞ;tr;up

ðe;h;cÞ;tr;up

ð9:13bÞ

# LPF ðe;h;cÞ;tr;do Pkðe;h;cÞ Ikðe;h;cÞ;tr;do 0 # Pkðe;h;cÞ;tr;do i ;s;t i ;s;t i ;s;t

ð9:13cÞ

0 # Pki ;s;t

# LPF ðe;h;cÞ;tr;up Pkðe;h;cÞ Iki ;s;t i ;s;t

ðe;h;cÞ;tr;up

0 # Ikðe;h;cÞ;tr;do 1 Iki ;s;t i ;s;t •

ð9:13aÞ

#1

ð9:13dÞ

Curtailable IDR program Constraint given in Eq. (9.14a) indicates the mechanism of implementing the curtailable IDR program. The load curtailed at each hour must be rebounded over the immediate subsequent 3 hours. In this regard, ϕ1 , ϕ2 , and ϕ3 are the rebounded load demand in percent and their values are 60%, 30%, and 10% respectively. Constraint Eqs. (9.14b) and (9.14c) indicate the upward and downward load demand at each hour. ðe;h;cÞ;cu;up

Pkðe;h;cÞ;cu;do 5 ϕ1 Pki ;s;t11 i ;s;t ðe;h;cÞ;cu;up

ðe;h;cÞ;cu;up

1 ϕ2 Pki ;s;t12

ðe;h;cÞ;cu;up

1 ϕ3 Pki ;s;t13

ð9:14aÞ

ðe;h;cÞ;cu;up

ð9:14bÞ

# LPF ðe;h;cÞ;cu;do Pkðe;h;cÞ Ikðe;h;cÞ;cu;do 0 # Pkðe;h;cÞ;cu;do i ;s;t i ;s;t i ;s;t

ð9:14cÞ

0 # Pki ;s;t

# LPF ðe;h;cÞ;cu;up Pkðe;h;cÞ Iki ;s;t i ;s;t

As it was mentioned before, there is a limit for DR program implementation at the same time to show the effectiveness of each DR program on operational results. Constraint Eqs. (9.15a)(9.15f) deal with this assumption to avoid multiple integrations of the DR programs at the same time. e;sh;up

e;pb;up

e;tr;up

e;cu;up

Pke;1 5 Pki ;s;t 1 Pki ;s;t 1 Pki ;s;t 1 Pki ;s;t i ;s;t

ð9:15aÞ

h;sh;up

h;tr;up

h;cu;up

ð9:15bÞ

c;sh;up

c;tr;up

c;cu;up

ð9:15cÞ

Pkh;1 5 Pki ;s;t 1 Pki ;s;t 1 Pki ;s;t i ;s;t Pkc;1 5 Pki ;s;t 1 Pki ;s;t 1 Pki ;s;t i ;s;t

Demand response role for enhancing the flexibility of local energy systems

Pke;2 5 Pke;sh;do 1 Pki ;s;t 1 Pke;tr;do 1 Pke;cu;do i ;s;t i ;s;t i ;s;t i ;s;t

e;pb;do

ð9:15dÞ

Pke;2 5 Pkh;sh;do 1 Pki ;s;t 1 Pkh;tr;do 1 Pkh;cu;do i ;s;t i ;s;t i ;s;t i ;s;t

h;pb;do

ð9:15eÞ

c;pb;do

ð9:15f Þ

Pkc;2 5 Pkc;sh;do 1 Pki ;s;t 1 Pkc;tr;do 1 Pkc;cu;do i ;s;t i ;s;t i ;s;t i ;s;t

• Power balance constraints Eqs. (9.16a)(9.16c) show the balance equations for the electrical, heating, and cooling power respectively. As Eq. (9.16a) shows, the electrical load demand is supplied by transacting power with the upstream grid and other energy hubs, and also by the CHP units, EES system, and other RESs. Besides, the heating load demand is supplied using the CHP unit, boiler, EHP, EH, and TES system. The cooling load demand is also supplied using the AC and EHP. PkG-EL 1 PkMi ;s;t-EL 1 PkESS-EL 1 PkRES-EL 1 PkCHP-EL 1 Pke;2 5 PkEL 1 Pke;1 2 PkENS i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t ð9:16aÞ 1 HkBoiler-HL 1 HkEHP-HL 1 HkEH-HL 1 HkTES-HL 1 Pkh;2 5 HkHL 1 Pkh;1 HkCHP-HL i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t ð9:16bÞ CkEHP-CL 1 CkAC-CL 1 Pkc;2 5 CkCL 1 Pkc;1 i ;s;t i ;s;t i ;s;t i ;s;t i ;s;t

ð9:16cÞ

• Power flow constraints Eqs. (9.17a)(9.17i) state the linear formulation of the power flow constraints. Eq. (9.17a) relates to the susceptance and conductance calculations. The active and reactive power flow equations are represented in Eqs. (9.17b) and (9.17c) respectively. Constraint Eqs. (9.17d) and (9.17e) state the maximum active and reactive power flow of lines respectively. The constraints of the voltage magnitude and angle are applied using inequalities given in Eqs. (9.17f) and (9.17g) respectively. The active and reactive power injections of each bus would be determined by employing Eqs. (9.18h) and (9.17i) respectively. GlLine 5

rl x ; BLine 5 2 l 2 rl 2 1 xl 2 l rl 1 xl

ð9:17aÞ

  Flow 5 BLine δs;i;t 2 δs;j;t 1 GlLine Vs;i;t 2 Vs;j;t Ps;l;t l

ð9:17bÞ

  Line Vs;i;t 2 Vs;j;t 2 GlLine δs;i;t 2 δs;j;t QFlow s;l;t 5 Bl

ð9:17cÞ

303

304

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

L X

Gen 1 Pn;s;t

Flow 2PlFlow;max # Ps;l;t # PlFlow;max

ð9:17dÞ

Flow;max 2QFlow;max # QFlow s;l;t # Ql l

ð9:17eÞ

Vimin ðiÞ # Vs;i;t # Vimax

ð9:17f Þ

# δs;i;t # δmax δmin i i

ð9:17gÞ

flow

Pl;s;t 5

l51jm-n

QGen n;s;t 1

L X

flow

Ql;s;t 5

l51jm-n

X ki An

X ki An

! PkG-H 1 PkMi ;s;t-H i ;s;t

1

L X

flow

ð9:17hÞ

Pl;s;t

l51jn-m

  M -H tanðϕki Þ PkG-H 1 P 1 k ;s;t ;s;t i i

L X

flow

Ql;s;t

ð9:17iÞ

l51jn-m

9.5 Simulation results This section is devoted to solving the proposed scheduling problem through simulating five different case studies, and the results, obtained are discussed. Table 9.1 provides the required information of the five case studies. The data of the energy hubs’ assets are available in Ref. [17]. Furthermore, the load demand data of each hub are presented in Table 9.2. Five DR and IDR programs have been simulated to investigate their impacts on the operating cost of the system. The results, obtained are represented in Table 9.3. The results, derived from simulation of the DR programs for electrical loads show that the price-based DR program performs better, as no payment would be made by the system operator. In other words, consumers shift their load demand with respect to the electricity price, leading Table 9.1 The studied five cases for DR program assessment. Case no.

DR

EES

TES

RER’s

Coordinate

Uncoordinated

1 2 3 4 5 6

û ü ü ü ü ü

ü ü ü û ü ü

ü ü ü ü û ü

ü ü ü ü ü û

ü ü û ü ü ü

û û ü û û û

DR, demand response; EES, electrical energy storage; RER, renewable energy resources;TES, thermal energy storage.

Demand response role for enhancing the flexibility of local energy systems

Table 9.2 Load demand data of energy hubs. Hour

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Industrial

Commercial

Residential

Spring/ Fall

Summer

Winter

Spring/ Fall

Summer

Winter

Spring/ Fall

Summer

Winter

450 450 450 450 600 900 1200 1425 1500 1500 1500 1500 1500 1500 1500 1500 1050 900 600 450 450 450 450 450

450 450 450 450 600 750 1050 1350 1500 1500 1500 1500 1500 1500 1500 1500 1350 1050 600 450 450 450 450 450

300 300 300 300 750 1050 1350 1500 1500 1500 1500 1500 1500 1500 1500 1350 1200 750 300 300 300 300 300 300

175 175 175 175 175 175 175 175 280 420 490 490 490 490 542.5 542.5 577.5 595 665 700 700 647.5 595 385

175 175 175 175 175 175 175 175 280 420 490 490 490 490 525 525 560 560 630 700 700 700 700 490

175 175 175 175 175 175 175 175 280 420 490 490 490 490 560 560 595 630 700 700 700 595 490 280

401.46 400.38 384.93 380.90 448.74 587.14 780.00 855.06 899.50 913.75 911.65 931.52 947.14 962.87 976.69 980.40 985.94 1020.5 1019.2 979.17 878.02 729.98 581.45 414.01

584.48 551.92 520.04 499.58 576.18 733.46 893.52 910.00 910.00 910.00 975.00 1105.0 1105.0 975.00 845.00 845.00 910.00 975.00 1040.0 1170.0 1300.0 1248.0 1040.0 780.00

190.37 259.62 273.40 304.33 386.37 551.51 638.60 730.84 729.01 732.25 780.00 780.00 780.00 780.00 845.00 845.00 780.00 715.00 715.00 715.00 624.00 596.03 397.20 192.59

to a modified load demand profile and reduced cost. On the contrary, consumers shift their load demand using the load-based DR program, only if they receive an incentive. It is noteworthy that both DR programs result in modifying the load demand profile while the price-based DR program would be more beneficial to the system operator. The simulation results for the three IDR programs, applied to electrical, heating, and cooling loads verify that the shiftable IDR program leads to a more desired solution. Transferrable and curtailable IDR programs are ranked the second and third in terms of their results desirability. It is noteworthy that the consumer would be paid using each of these three programs. The superior performance of the shiftable IDR program is due to the fact that it is associated with lower operating limitations. The obtained results for the five case studies also verify that the shiftable IDR program is associated with the best performance compared to others. Although the system operator should pay to consumers, it leads to better results as it is applied to heating and cooling loads as well, and it shifts the load demand to off-peak hours. It has been revealed that the

305

306

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 9.3 The simulation results for the five case studies. Operation cost ($/day) Shiftable DR

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3356.36 1006.39 1248.15

4718.62 1401.88 2407.16

2984.14 763.52 806.96

3474.01 904.65 969.64

Price-based DR

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3317.89 987.26 1218.42

4668.27 1403.32 2376.99

2945.05 752.27 790.46

3438.88 886.43 958.61

Shiftable IDR

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3301.43 974.53 1212.43

4475.41 1377.45 2279.39

2941.10 753.64 789.58

3400.73 884.45 955.20

Transferable IDR

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3481.46 990.60 1268.12

4742.99 1381.06 2418.77

3049.75 788.50 819.92

3498.88 904.23 971.51

Curtailable IDR

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3564.38 1002.35 1310.78

4881.90 1411.21 2484.18

3118.69 797.16 833.02

3569.00 909.93 985.26

Without DR or IDR

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3612.40 1010.91 1327.97

4975.28 1425.36 2540.14

3172.39 798.17 840.98

3596.32 919.32 1000.91

DR, demand response; IDR, integrated demand response.

residential energy hub has participated more in the DR programs as it includes more flexible loads compared to the commercial and industrial energy hubs. Figs. 9.39.5 depict the impact of shifting IDR program on the electrical load demand curve in summer, the cooling load demand curve in summer, and the heating load demand curve in winter respectively. As can be observed, this IDR program has successfully and effectively modified the load demand curves by shifting the peak load to off-peak hours.

Demand response role for enhancing the flexibility of local energy systems

Figure 9.3 Electrical load demand in summer.

Figure 9.4 Cooling load demand in summer.

Figure 9.5 Heating load demand in winter.

Table 9.4 represents the results, obtained from simulating the six case studies with and without IDR programs. As the shiftable IDR program has the most desired performance, it is used for further studying the problem. The comparison made between case studies 1 and 2 shows that the operating cost would be much higher without applying the IDR

307

308

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 9.4 The simulation results for the six case studies with and without the shiftable IDR program. Operation cost ($/day) Case 1

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3612.40 1010.91 1327.97

4975.28 1425.36 2540.14

3172.39 798.17 840.98

3596.32 919.32 1000.91

Case 2

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3301.43 974.53 1212.43

4475.41 1377.45 2279.39

2941.10 753.64 789.58

3400.73 884.45 955.20

Case 3

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3401.92 989.80 1365.92

4551.49 1458.23 2614.69

3013.95 791.46 841.72

3435.05 933.80 952.27

Case 4

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3397.53 986.18 1260.39

4593.77 1407.90 2329.01

3001.95 780.61 805.45

3473.09 894.19 952.09

Case 5

Industrial Commercial Residential

Spring

Summer

Fall

Winter

3310.01 978.18 1214.61

4501.24 1379.98 2282.78

2949.61 757.29 791.85

3425.37 888.10 957.47

Case 6

Industrial Commercial Residential

Spring

Summer

Fall

Winter

4525.80 1029.63 1388.68

6450.50 1526.07 2571.90

4349.58 836.76 869.25

5174.02 937.70 983.54

IDR, integrated demand response.

program in Case 1 compared to Case 2. This difference is more considerable in summer and winter due to their higher load demands compared to spring and fall. Moreover, the commercial energy hub is less impacted by the IDR program as its load demand is less flexible compared to other energy hubs. In Case 3, it is assumed that the three energy hubs are only allowed to transact power with the upstream grid and transactive energy

Demand response role for enhancing the flexibility of local energy systems

trading between these hubs is not allowed. Accordingly, the operating cost in this case is substantially higher than Case 2. This is due to the fact that they should pay more for emission costs, associated with the energy transaction with the upstream grid. The commercial energy hub is much more affected by this limitation and it should tolerate a higher cost. In this respect, the peak load demands of the commercial and industrial hubs are not coincident and energy trading between these two hubs could have considerably reduced the operating cost of this hub. The impact of this limitation is more than the absence of the IDR program in residential and commercial energy hubs. In this respect, the residential hub could have purchased its required power over the final hours of the day from the industrial hub and decreased its operating costs by not paying for emission costs. It is noted that the load demand of industrial hub is significantly low during these hours and it can sell power to the commercial and residential hubs. The results indicate that in general, the transactive energy trading between the hubs would substantially enhance the system flexibility. Case 4 is simulated without any EES system, showing that a higher cost should be tolerated which is more tangible in the industrial hub as it owns a larger EES system. It is noted that if the hub is equipped with an EES system, it is charged during the initial hours of the day at low prices through absorbing power from the upstream grid or the surplus power generation of the CHP unit. Thus, it can deliver power to the system over the peak-load periods, resulting in reduced load demand and operating costs. The simulation results, obtained in Case 5 indicate that the impact of the absence of the TES system on the operating cost is less significant compared to the EES system. The comparison with Case 2 shows that the effect of lacking the TES system can be more observed in summer and winter, when it is utilized for providing the required heating power of the AC and supplying the heating load demand. The operating cost considerably increases in Case 6 without any renewable energies, that is, PV panels in residential and commercial hubs and a wind turbine in the industrial energy hub. It is noted that the impact of renewable energies on the operating cost is much more significant compared to other cases. Table 9.5 represents the annual cost due to emissions in each case. The results show that the impacts of the studied cases on the operating cost are different and the cases in which energy hubs pay more for emission are different as well. It is worth mentioning that the lowest cost relates to Case 2 with all assets. Besides, the industrial Table 9.5 Emission costs for the six case studies and for each hub. Energy hub units

Industrial Commercial Residential

Emission cost ($/year) Case 1

Case 2

Case 3

Case 4

Case 5

Case 6

218260.21 103639.36 160766.16

196188.69 102746.42 145154.50

209272.361 115912.99 188975.13

203439.49 104145.05 149009.74

197186.96 106374.95 145238.42

315663.40 110911.37 174660.23

309

310

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

hub tolerates the highest cost in Case 6, where there is no wind turbine as its capacity is relatively high. Accordingly, the industrial hub should purchase more power from the upstream grid, which in turn leads to a higher emission cost. The highest costs of the residential and commercial hubs occur in Case 3, where the transactive energy trading between hubs is not allowed. A substantial fraction of their load demand could have been supplied by the industrial hub, while without any access to the industrial hub, they have to purchase power from the upstream grid and pay more for emission.

9.6 Conclusion remarks This chapter investigated the impacts of five DR and IDR programs on the scheduling of local energy hubs. First, a comprehensive review was carried out on the background of DR programs in local energy systems. Then, the mechanism of local energy systems and IDR programs in such systems were described. In this respect, a MILP framework was developed for the optimal scheduling of multiple energy hubs, connected to a 33-bus distribution network. The studied hubs were industrial, commercial, and residential. After determining the most desired IDR program, that is, shiftable IDR program, showing a better performance compared to other programs, six case studies were simulated and analyzed. In this respect, the impacts of different assets and capabilities on the operating and emission costs were investigated. The results obtained from the simulation showed that renewable energies have the most significant impact on the emission cost of the industrial hub, while the other two hubs were mainly affected by the transactive energy trading between hubs. Besides, in the case without any renewable energies, all the three hubs tolerated the highest cost compared to other cases. This is due to the fact that a significant fraction of their load demand during the initial and final hours of the day is supplied by the industrial hub. Moreover, after renewable energies, the DR programs have the highest impact on the operating cost of the industrial hub. It was also noted that the residential and commercial hubs had to pay more for the emission costs compared to the industrial hub in the absence of transactive energy trading. With respect to the fact that the industrial hub could have supplied a substantial amount of the energy demand of the other two hubs, the residential and commercial hubs have to transact power with the upstream grid and pay for the emission costs. The absence of wind turbine caused the industrial hub to pay the highest amount for the emission as the capacity of the wind turbine was considerable. Five DR programs were tested, besides the energy management program, showing that the best one was the shiftable IDR program. This efficacy was due to the opportunities provided by impacting the heating and cooling load demand as well, particularly in summer and winter. It is also noteworthy that the residential hub is more affected by the DR programs in comparison with the other two hubs which is due to the higher flexibility of residential loads compared to industrial and commercial loads.

Demand response role for enhancing the flexibility of local energy systems

Acknowledgment João P.S. Catalão acknowledges the support by FEDER funds through COMPETE 2020 and by Portuguese funds through FCT under POCI-01-0145-FEDER-029803 (02/SAICT/2017). Miadreza Shafie-Khah acknowledges the support by FLEXIMAR project (Novel marketplace for energy flexibility), which has received funding from Business Finland Smart Energy Program, 20172021.

References [1] S.A. Mansouri, M.S. Javadi, A robust optimisation framework in composite generation and transmission expansion planning considering inherent uncertainties, J Exp Theor Artif Intell (2016) 717730. Available from: https://doi.org/10.1080/0952813X.2016.1259262. [2] C. He, T. Liu, L. Wu, M. Shahidehpour, Robust coordination of interdependent electricity and natural gas systems in day-ahead scheduling for facilitating volatile renewable generations via powerto-gas technology, J. Mod. Power Syst. Clean Energy 5 (2017) 375388. Available from: https:// doi.org/10.1007/s40565-017-0278-z. [3] Z. Yang, C. Gao, M. Zhao, Coordination of integrated natural gas and electrical systems in dayahead scheduling considering a novel flexible energy-use mechanism, Energy Convers. Manag. 196 (2019) 117126. Available from: https://doi.org/10.1016/j.enconman.2019.05.109. [4] P. Dueñas, T. Leung, M. Gil, J. Reneses, Gaselectricity coordination in competitive markets under renewable energy uncertainty, IEEE Trans. Power Syst 30 (2015) 123131. Available from: https://doi.org/10.1109/TPWRS.2014.2319588. [5] A. Zlotnik, L. Roald, S. Backhaus, M. Chertkov, G. Andersson, Control policies for operational coordination of electric power and natural gas transmission systems, Am. Control Conf. (2016) 74787483. Available from: https://doi.org/10.1109/ACC.2016.7526854. [6] F. Böing, A. Regett, Hourly CO2 emission factors and marginal costs of energy carriers in future multi-energy systems, Energies 12 (2019). Available from: https://doi.org/10.3390/en12122260. [7] X. Zhao, F. Liu, B. Fu, N. Fang, Reliability analysis of hybrid multi-carrier energy systems based on entropy-based Markov model, Proc. Inst. Mech. Eng. Part O J. Risk Reliab 230 (2016) 561569. Available from: https://doi.org/10.1177/1748006X16663056. [8] M. Arnold, R.R. Negenborn, G. Andersson, B. De Schutter, Model-based predictive control applied to multi-carrier energy systems, in: 2009 IEEE Power & Energy Society General Meeting, 2009, pp. 18. Available from: https://doi.org/10.1109/PES.2009.5275230. [9] S.A. Mansouri, A. Ahmarinejad, M.S. Javadi, J.P.S. Catalão, Two-stage stochastic framework for energy hubs planning considering demand response programs, Energy (2020) 118124. Available from: https://doi.org/10.1016/j.energy.2020.118124. [10] N. Good, P. Mancarella, Flexibility in multi-energy communities with electrical and thermal storage: a stochastic, robust approach for multi-service demand response, IEEE Trans. Smart Grid 10 (2019) 503513. Available from: https://doi.org/10.1109/TSG.2017.2745559. [11] M. Diekerhof, S. Schwarz, A. Monti, Electrothermal flexibility for demand response using inverse simulation, IEEE Syst. J 13 (2019) 17761785. Available from: https://doi.org/10.1109/ JSYST.2018.2816043. [12] M.S. Javadi, A. Anvari-Moghaddam, J.M. Guerrero, Optimal scheduling of a multi-carrier energy hub supplemented by battery energy storage systems, in: Conference Proceedings17th IEEE International Conference on Environmental and Electrical Engineering and 1st IEEE Industrial and Commercial Power Systems Europe (EEEIC/I&CPS Europe), (2017). Available from: https://doi.org/10.1109/ EEEIC.2017.7977520. [13] J. Salehi, A. Namvar, F.S. Gazijahani, Scenario-based co-optimization of neighboring multi carrier smart buildings under demand response exchange, J. Clean. Prod. 235 (2019) 14831498. Available from: https://doi.org/10.1016/j.jclepro.2019.07.068.

311

312

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[14] A. Najafi-Ghalelou, S. Nojavan, K. Zare, B. Mohammadi-Ivatloo, Robust scheduling of thermal, cooling and electrical hub energy system under market price uncertainty, Appl. Therm. Eng. 149 (2019) 862880. Available from: https://doi.org/10.1016/j.applthermaleng.2018.12.108. [15] F. Sheidaei, A. Ahmarinejad, Multi-stage stochastic framework for energy management of virtual power plants considering electric vehicles and demand response programs, Int. J. Electr. Power Energy Syst. 120 (2020) 106047. Available from: https://doi.org/10.1016/j.ijepes.2020.106047. [16] M. Jadidbonab, E. Babaei, B. Mohammadi-ivatloo, CVaR-constrained scheduling strategy for smart multi carrier energy hub considering demand response and compressed air energy storage, Energy 174 (2019) 12381250. Available from: https://doi.org/10.1016/j.energy.2019.02.048. [17] S.A. Mansouri, A. Ahmarinejad, M. Ansarian, M.S. Javadi, J.P.S. Catalao, Stochastic planning and operation of energy hubs considering demand response programs using Benders decomposition approach, Int. J. Electr. Power Energy Syst. 120 (2020) 106030. Available from: https://doi.org/ 10.1016/j.ijepes.2020.106030. [18] P. Firouzmakan, R.-A. Hooshmand, M. Bornapour, A. Khodabakhshian, A comprehensive stochastic energy management system of micro-CHP units, renewable energy sources and storage systems in microgrids considering demand response programs, Renew. Sustain. Energy Rev. 108 (2019) 355368. Available from: https://doi.org/10.1016/j.rser.2019.04.001. [19] M. Majidi, B. Mohammadi-Ivatloo, A. Anvari-Moghaddam, Optimal robust operation of combined heat and power systems with demand response programs, Appl. Therm. Eng. 149 (2019) 13591369. Available from: https://doi.org/10.1016/j.applthermaleng.2018.12.088. [20] Z. Yuan, S. He, A. Alizadeh, S. Nojavan, K. Jermsittiparsert, Probabilistic scheduling of power-togas storage system in renewable energy hub integrated with demand response program, J. Energy Storage 29 (2020) 101393. Available from: https://doi.org/10.1016/j.est.2020.101393. [21] M. Azimi, A. Salami, Optimal operation of integrated energy systems considering demand response program, J. Oper. Autom. Power Eng (2020). Available from: https://doi.org/10.22098/ joape.2020.6928.1506, In press. [22] F. Jafari, H. Samet, A.R. Seifi, M. Rastegar, Developing a two-step method to implement residential demand response programmes in multi-carrier energy systems, IET Gener. Transm. Distrib 12 (2018) 26142623. Available from: https://doi.org/10.1049/iet-gtd.2017.1557. [23] M. Di Somma, G. Graditi, P. Siano, Optimal bidding strategy for a DER aggregator in the dayahead market in the presence of demand flexibility, IEEE Trans. Ind. Electron. 66 (2019) 15091519. Available from: https://doi.org/10.1109/TIE.2018.2829677. [24] H.A. Muqeet, I.A. Sajjad, A. Ahmad, M.M. Iqbal, S. Ali, J.M. Guerrero, Optimal operation of energy storage system for a prosumer microgrid considering economical and environmental effects, in: 2019 International Symposium on Recent Advances in Electrical Engineering, 2019, pp. 16. Available from: https://doi.org/10.1109/RAEE.2019.8887002. [25] H. Chamandoust, G. Derakhshan, S.M. Hakimi, S. Bahramara, Tri-objective scheduling of residential smart electrical distribution grids with optimal joint of responsive loads with renewable energy sources, J. Energy Storage 27 (2020) 101112. Available from: https://doi.org/10.1016/j. est.2019.101112. [26] W. Huang, N. Zhang, C. Kang, M. Li, M. Huo, From demand response to integrated demand response: review and prospect of research and application, Prot. Control Mod. Power Syst. 4 (2019) 12. Available from: https://doi.org/10.1186/s41601-019-0126-4. [27] L. Ni, W. Liu, F. Wen, Y. Xue, Z. Dong, Y. Zheng, et al., Optimal operation of electricity, natural gas and heat systems considering integrated demand responses and diversified storage devices, J. Mod. Power Syst. Clean Energy 6 (2018) 423437. Available from: https://doi.org/10.1007/ s40565-017-0360-6. [28] W. Huang, N. Zhang, J. Yang, Y. Wang, C. Kang, Optimal configuration planning of multienergy systems considering distributed renewable energy, IEEE Trans. Smart Grid 10 (2019) 14521464. Available from: https://doi.org/10.1109/TSG.2017.2767860. [29] A. Mansour-Saatloo, M. Agabalaye-Rahvar, M.A. Mirzaei, B. Mohammadi-Ivatloo, M. Abapour, K. Zare, Robust scheduling of hydrogen based smart micro energy hub with integrated demand response, J. Clean. Prod. 267 (2020) 122041. Available from: https://doi.org/10.1016/j.jclepro.2020.122041.

Demand response role for enhancing the flexibility of local energy systems

[30] 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 (2018) 854866. Available from: https://doi.org/10.1016/j.energy.2017.09.144. [31] T. Song, Y. Li, X.-P. Zhang, C. Wu, J. Li, Y. Guo, et al., Integrated port energy system considering integrated demand response and energy interconnection, Int. J. Electr. Power Energy Syst. 117 (2020) 105654. Available from: https://doi.org/10.1016/j.ijepes.2019.105654. [32] T. Liu, D. Zhang, S. Wang, T. Wu, Standardized modelling and economic optimization of multicarrier energy systems considering energy storage and demand response, Energy Convers. Manag. 182 (2019) 126142. Available from: https://doi.org/10.1016/j.enconman.2018.12.073. [33] L.A. Hurtado, E. Mocanu, P.H. Nguyen, M. Gibescu, R.I.G. Kamphuis, Enabling cooperative behavior for building demand response based on extended joint action learning, IEEE Trans. Ind. Informatics 14 (2018) 127136. Available from: https://doi.org/10.1109/TII.2017.2753408. [34] P. Siano, Demand response and smart grids—a survey, Renew. Sustain. Energy Rev. 30 (2014) 461478. Available from: https://doi.org/10.1016/J.RSER.2013.10.022. [35] V.S.K.M. Balijepalli, V. Pradhan, S.A. Khaparde, R.M. Shereef, Review of demand response under smart grid paradigm, in: ISGT2011-India, 2011, pp. 236243. Available from: https://doi.org/ 10.1109/ISET-India.2011.6145388. [36] G.G. Dranka, P. Ferreira, Review and assessment of the different categories of demand response potentials, Energy 179 (2019) 280294. Available from: https://doi.org/10.1016/j.energy.2019.05.009. [37] P. Faria, Z. Vale, Demand response in electrical energy supply: an optimal real time pricing approach, Energy 36 (2011) 53745384. Available from: https://doi.org/10.1016/j.energy.2011.06.049. [38] P. Mancarella, G. Chicco, Real-time demand response from energy shifting in distributed multigeneration, IEEE Trans. Smart Grid 4 (2013) 19281938. Available from: https://doi.org/10.1109/ TSG.2013.2258413. [39] K. Herter, S. Wayland, Residential response to critical-peak pricing of electricity: California evidence, Energy 35 (2010) 15611567. Available from: https://doi.org/10.1016/j.energy.2009.07.022. [40] F. Wang, B. Xiang, K. Li, X. Ge, H. Lu, J. Lai, et al., Smart households’ aggregated capacity forecasting for load aggregators under incentive-based demand response programs, IEEE Trans. Ind. Appl. 56 (2020) 10861097. Available from: https://doi.org/10.1109/TIA.2020.2966426. [41] J. Wang, H. Zhong, Z. Ma, Q. Xia, C. Kang, Review and prospect of integrated demand response in the multi-energy system, Appl. Energy 202 (2017) 772782. Available from: https://doi.org/ 10.1016/j.apenergy.2017.05.150. [42] M.S. Javadi, M. Lotfi, A.E. Nezhad, A. Anvari-Moghaddam, J.M. Guerrero, J.P.S. Catalao, Optimal operation of energy hubs considering uncertainties and different time resolutions, IEEE Trans Ind Appl 56 (2020) 55435552. Available from: https://doi.org/10.1109/tia.2020.3000707.

313

CHAPTER 10

The integration of electric vehicles in smart distribution grids with other distributed resources Morris Brenna1, Federica Foiadelli1, Dario Zaninelli1, Giorgio Graditi2 and Marialaura Di Somma2 1

Department of Energy, Politecnico Di Milano, Milan, Italy Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy 2

Abbreviations AC BEV DC DR DSO ESS EV EVSE GEV GIS HD HEV ICE ICT LV MV PHEV RES SOC STLF TSO V2G V2H V2V

Alternating Current Battery-Electric Vehicle Direct Current Distributed Resources Distribution System Operator Electric Storage System Electric Vehicle Electric Vehicle Supply Equipment Gridable Electric Vehicle Geographic Information System Hybridization Degree Hybrid Electric Vehicle Internal Combustion Engine Information and Communications Technology Low Voltage Medium Voltage Plug-in Hybrid Electric Vehicle Renewable Energy Source State Of Charge Short-Term Load Forecast Transmission Systems Operator Vehicle-to-Grid Vehicle-to-Home Vehicle-to-Vehicle

Nomenclature PSS Ptot

power of the secondary source of a PHEV total power of a PHEV

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00006-6

r 2021 Elsevier Inc. All rights reserved.

315

316

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

P EV ;0 P EV ;max t max α QEV :DEM P P0 Q Q0 pvs pf s qvs qf s k1 to k8

initial status of battery system while charging maximum power capacity of the electric vehicle maximum charging time. constant parameter declared by the EV manufacturers reactive power demand of an EV active power at any voltage and frequency active power at the rated voltage and frequency reactive power at any voltage and frequency reactive power at the rated voltage and frequency voltage sensitivity of the active power frequency sensitivity of active power voltage sensitivity of the reactive power frequency sensitivity of reactive power constants for the different types of the loads

10.1 Introduction to electric vehicles and charging infrastructures Electric vehicles (EVs) are spreading rapidly thanks to the continuous researches and innovations of the automotive sector in response to the need to find solutions for the decarbonization of the transport sector. Both EVs and plug-in hybrid electric vehicle (PHEV) need to be charged by an external source through a distribution grid. There are several charging types according to vehicle type. According to customers’ facilities that they have, these charging stations can be changed. For instance, the basic version of charging station is the residential type which can be fed by household power inlet 220 V AC. Another way is to charge the EV at a charging station. These charging stations are mostly operated by private companies, and they can be found on common areas inside of the city, universities, shopping malls, workplaces, airports, etc., and also on highways as well. Another way to say that, charging stations are parking lots to charge EVs. Thus, they contain one or more EV charging equipment, radio-frequency identification reader, buttons, displays and light emitting diodes to make the process much easy for users. For the different power levels, there are several charging modes for EVs. In particular, European Standard ISO 61851 identifies four main modes. The main parameters to divide into groups of these charging modes are the current type, alternating current (AC) or direct current (DC), and power and protection levels.

10.1.1 Characteristics of electric vehicles Electric or electrified vehicles can be classified as follows: • Battery Electric Vehicle (BEV), is an electric traction vehicle exclusively powered by rechargeable batteries. The battery is recharged by connecting the vehicle to the electric grid and/or through a regenerative braking system. The battery EV does not produce local exhausts, contrary as it occurs in ICE (Internal Combustion Engine) vehicles.

The integration of electric vehicles in smart distribution grids with other distributed resources

• Hybrid Electric Vehicle (HEV), is a vehicle formed by the integration of an ICE with an electric drive system; both systems contribute, simultaneously or separately, to drive the vehicle. The batteries are not rechargeable from mains, but only by the regenerative braking system and from the ICE equipped with a suitable generator. The main purpose of the electric motor is not the short section in which can support individual driving, but the considerable help in terms of torque that can be given to the vehicle during its acceleration or braking. • PHEV, is a type of hybrid electric vehicle that combines a gasoline or diesel engine with an electric motor and a large battery that can be recharged by plugging into an electrical outlet or EV charging station. PHEVs typically can run in at least two modes: “all-electric”, in which the motor and battery provide all the car’s energy; and “hybrid”, in which both electricity and gasoline are employed. Some PHEVs can travel more than 70 miles on electricity alone and under typical driving conditions, store enough electricity to cut their gasoline use. PHEVs provide the fueland cost-efficiency of hybrid models along with the all-electric capabilities of battery-electric or fuel-cell vehicles. PHEVs use approximately 30%
60% less gasoline than conventional vehicles, potentially saving the hundreds of euros per year. In addition to plugging into an outside electric power source, PHEV batteries can be charged by an ICE or regenerative braking. The electric motor supplements the engine’s power and, as a result, smaller engines can be used, increasing the car’s fuel efficiency without compromising performance. PHEVs typically start up in allelectric mode and operate on electricity until their battery pack is depleted. Some models shift to hybrid mode when they reach highway cruising speed. Once the battery is empty, the engine takes over and the vehicle operates as a conventional, nonplug-in hybrid. • Extended Range Electric Vehicle, is a vehicle that runs only in electric mode until the battery supplies energy; when the charge drops below a certain level a small size ICE charges the battery to extend the driving range. • Fuel-Cell Electric Vehicle, is an electric vehicle in which the electric energy is produced on-board by a fuel-cell generator, supplied from hydrogen or other fuels. • Starting from a conventional ICE, a vehicle tends toward a pure BEV through different hybridization degrees (HD), according to the following relation: HD 5

Pss Ptot

ð10:1Þ

in which PSS is the power of the secondary source and Ptot is the total power. The different solutions proposed for hybrid vehicles are reported in Fig. 10.1 in which the various functions for the different categories are highlighted.

317

318

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Conventional ICE Micro HEV

Start-stop

Mild HEV

Start-stop Regenerative brake Electric support drive

Full HEV

Start-stop Regenerative brake Exclusive electric mode driving

PHEV

Electric mode driving (greater autonomy) Internal or external power charge

ICE

Electric motor

Storage Secondary source

PSS

0 Hybridization Degree 1

BEV

Figure 10.1 Different categories of electrified vehicles according to their hybridization degree.

ICE

G

AC DC DC

M AC

Storage

DC DC

Figure 10.2 Series PHEV configuration.

Hybrid and plug-in hybrid vehicles can have different configurations depending on how the internal combustion engine participates in the traction of the vehicle itself. In particular, three categories can be identified: • Series PHEV • Parallel PHEV • Series-parallel PHEV 10.1.1.1 Series PHEV In this configuration, the vehicle has only one or more electric traction motors. The electric energy is provided by a generator connected to a prime mover ICE and to a storage system as reported in Fig. 10.2. This configuration is suitable for extended range EVs, since the traction is entirely provided by electric motors. 10.1.1.2 Parallel PHEV There are various parallel hybrid configurations depending on the HD of the vehicle. In any case, the traction power is provided both from ICE and an electric motor supplied from a storage system as depicted in Fig. 10.3. This solution is suitable for micro and mild hybrid vehicles.

The integration of electric vehicles in smart distribution grids with other distributed resources

ICE M

ICE

Storage

DC

M

Storage

AC

DC AC

(A)

(B)

ICE

DC

Storage

M

AC

(C)

Figure 10.3 Different parallel PHEV configurations: electric motor and ICE are parallelized at the transmission level (A), the electric motor is placed on the same axel of the ICE (B), ICE moves the front axle and the electric motor moves the rear axle (C).

M

ICE

G

AC DC DC AC

Storage

DC

DC

Figure 10.4 Series-parallel PHEV configuration.

10.1.1.3 Series-parallel PHEV This configuration is exploited by the full hybrid vehicles. Depending on the working mode, the vehicle can have series configuration or parallel one or both at the same time. The powertrain scheme is shown in Fig. 10.4. Series hybrid vehicles have a simpler mechanical transmission system and allow a greater flexibility in the choice of the layout of various devices. In particular, ICE can be placed in any part of the vehicle allowing a full low-floor, that is an advantage especially for public transport vehicles. Besides, series hybrid vehicles have a simpler control system as the traction is provided by only one or more electric motors. Moreover, in series hybrid vehicles, ICE has usually a smaller size compared to the

319

320

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

traction motor. This means that ICE works in optimal conditions even if the vehicle has to do many start&stop cycles. However, depending on the vehicle, the total efficiency of a series hybrid can be lower than a parallel configuration. Indeed, in series hybrid, the electric motor is the only propulsion system, therefore it has to be sized for the maximum traction power, even when the vehicle works at a reduced power rate for most part of the time. Instead, for long travel distance, in a series hybrid the ICE has to be sized for the maximum power due to the reduced capacity of the storage system.

10.1.2 Low power AC charging infrastructures AC charging infrastructures allow EVs to be connected directly to a single-phase or three-phase AC grid, therefore they are used for a massive installation of charging points, and so they permit a wide spread of EVs. However, AC charging infrastructures are characterized by low charging power (usually up to 22 kW), so the charging time is long. They are suitable for domestic applications or whenever the vehicle can stay stopped for long time, at least a few hours. Charging modes are classified in standard IEC 61851-1:2017 [1]. 10.1.2.1 Mode 1 Charging Mode 1 is the basic level of charging stations. In this mode, an EV is directly connected to a standard socket-outlet of an AC supply grid, utilizing a cable and plug, that are not equipped with any supplementary pilot or auxiliary contacts. The basic scheme is represented in Fig. 10.5. The rated values for current and voltage shall not exceed: • 16 A and 250 V AC, single-phase for a maximum power of 3.6 kW • 16 A and 480 V AC, three-phase for a maximum power of 11 kW EV Supply Equipment (EVSE) applied to Mode 1 charging uses the same protection of systems at which it is connected; moreover, it has to provide a protective earthing conductor from the standard plug to the vehicle connector. 10.1.2.2 Mode 2 Mode 2 charging is a bit developed model of Mode 1. It is a method for the connection of an EV to a standard socket-outlet of an AC supply distribution grid utilizing Mode 1 AC

Figure 10.5 Basic scheme of charging Mode 1.

The integration of electric vehicles in smart distribution grids with other distributed resources

Mode 2 AC

PWM

Figure 10.6 Basic scheme of charging Mode 2.

Mode 3 AC

PWM

Figure 10.7 Basic scheme of charging Mode 3.

an AC EVSE with a cable and plug, in which the control pilot function and system for personal protection against electric shock are placed inside the connecting cable, according to scheme represented in Fig. 10.6. The rated values for current and voltage shall not exceed: • 32 A and 250 V, single-phase for a maximum power of 7.2 kW • 32 A and 480 V, three-phase for a maximum power of 22 kW As for Mode 1, the EVSE has to provide a protective earthing conductor from the standard plug to the vehicle connector. 10.1.2.3 Mode 3 The latest version of the AC charging mode is Mode 3. In charging mode 3 an EV is supplied from an AC EVSE that is permanently connected to the AC mains. Control pilot and protection function extend from the AC EVSE to the vehicle. Charging mode 3 requires the installation of a wall box or individual charging station. Charging cable can be separated from the EVSE or it can be permanently connected. The basic scheme is shown in Fig. 10.7. Charging mode 3 is able to handle higher power up to 43 kW in three-phase systems, but the actual applications are limited to 22 kW since it is difficult for the installation of a high-power battery charger onboard the vehicle. Nowadays, typical onboard battery chargers are in the range of 7.2 kW single-phase to 22 kW threephase, with a preferred power of 11 kW three-phase. This is the reason why the 43 kW version of the charging stations has been practically abandoned. Mode 3 is mandatory for public charging stations, but it is recommended also for private applications because it guarantees the highest protection degree.

321

322

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Mode 4 DC

Figure 10.8 Basic scheme of charging Mode 4.

10.1.3 High power DC charging infrastructures DC charging infrastructures are characterized by a high charging power because the battery charger is now placed in charging station itself. In this way it is possible to increase the power of the electronic converter since there are less constraints in terms of volume and weight. The basic scheme of a DC charging infrastructure is depicted in Fig. 10.8. Now the EVSE includes not only the control and protection functions, but also all the power appliances needed to convert and regulate the power from an AC threephase system to the onboard batteries. The cost for these infrastructures is much higher than the AC ones, so they are mostly used when it is necessary to reduce the charging time. Indeed, they are mostly installed along the highways to realize the so-called e-corridors, that is, electrified roads that allow an EV to travel long distances. Typical power for a DC charging station nowadays is around 100
150 kW, that permit to extend by 100 km the driving range of an EV in less than 10 minutes. Besides, new companies are proposing ultrafast chargers that can reach 350 kW, so that an EV can add 100 km of driving range in about 3 minutes. New standards from the United States, Europe, China, and Japan are moving towards 500 kW. In this way it is possible to extend the driving range by 100 km in less than 2 minutes. Once the EVs will be able to accept such amount of power, they can be used as traditional vehicles with internal combustion engine. DC charging Mode 4 can be used for a wide type of vehicles such as cars, buses, and motorbikes. In the United States and Europe, DC Mode 4 is combined with AC Mode 3, hence the vehicle is equipped with only one inlet that is able to accept AC or DC plugs, even if other standards are present.

10.2 Integration of electric vehicles in smart distribution grids The traditional grid, also called passive grid, presents a simple scheme in which electric power is transferred in unidirectional way that is being generated in traditional power plants, dispatched through the transmission network and absorbed by passive load. This is an inefficient way to manage the grid since it is a very inelastic system, fault sensible with the subsequential low quality of services due to interruptions.

The integration of electric vehicles in smart distribution grids with other distributed resources

Therefore, in traditional electrical systems, energy production is centralized in big power plant and transported by extra high voltage and high voltage [HV] network (132
150220
380 kV), which is then transformed in medium voltage [MV] in primary station where the voltage is brought to 15
20-23 kV for the distribution network which then, thanks to secondary stations, provides low voltage [LV] current to final user (230
400 V). Although this structure, the traditional grid worked well, but nowadays with renewable energy sources (RES) contributes in power generation it is becoming too rigid and inefficient. This is due to the fact that the power production curve and the power consumption are no longer a match. It is mandatory to have a more flexible and dynamic grid in order to make production to follow the energy demand avoiding any energy to be wasted. This is the reason why nowadays modern distribution networks have to evolve towards the so-called smart grids [2]. A definition of smart grid is: “an electricity network that can efficiently integrate the behavior and actions of all users connected to it, generators, consumers and those that do both, in order to ensure economically efficient, sustainable power system with low losses and high levels of quality and security of supply and safety.” The transition from the traditional to the smart grid requires time and specific action aimed to improve all the actors involved in the electric grid. Generation, transmission and distribution networks, storage systems and active users must be optimized and coordinated. This is possible only if an efficient communication system is in place. The first element that characterizes the smart grid is the presence of the active user, which is a very important figure appeared during the last decades. An active user, also called Prosumer, is a private who has the possibility of both producing and consuming energy. The most widespread and well-known example are the photovoltaic plants for residential application. The presence of this distributed generation must not be underestimated since it affects the grid in multiple ways, that is, islanding issues. Analyzing other terms is possible to highlight the smart features characterizing the smart grid as represented in Fig. 10.9 [3]. In this context, charging infrastructures for EVs vehicles are becoming an active part inside a smart grid, that is, EVs have not been considered as simple loads, but they can participate in the power regulation. Smart grid structure is a bit more sophisticated and developed with respect to the traditional grid, and many levels of smartness can be recognized: usually first improvement for safety and reliability, then for integration of diffuse generation and lastly to add complementary services to grid (storage, smart metering, etc. . .). To realize all these functions, a smart grid could not exist without a dedicated and optimized communication system. The information collected from the final subjects, that include users, distributed generators and storage systems, and charging infrastructures for EVs, has to be aggregated when it is transmitted towards to the control center.

323

Smart generaon

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Communicaon between producers and consumers

Increasing RES forecasng reliability

Automaon in fault detecon and grid riprisnaon

Communicaon between distributed generaon and tradional generators

Load control opmizaon

Smart network

324

Distributed generaon interface protecon [intertrip and transfer trip]

Voltage profile control

Automac detecon of faulty branches [logical selecvity]

Minimizaon of transmission congeson

Interface with recharging infrastructure for electric vehicle

Figure 10.9 Smart features that characterize a smart grid.

Therefore, the communication system has a hierarchical structure in which each element of the grid has its own communication device which is in contact with both the upper and lower level. As previously seen, an EV is a vehicle that is powered, at least in part, by electricity. Apart from the traditional electrified vehicles (e.g., rail rolling stocks), road EVs are now a real choice. To power up these vehicles it is required to access the energy from the grid and it represents a new sophisticated mobile load to the system (in terms of mobility and high charging current requirement). As smart loads, EVs are often

The integration of electric vehicles in smart distribution grids with other distributed resources

connected to communication networks using any of several different communication technologies to communicate with the outside world using on-board sensors and internet connectivity. This allows the vehicle to enhance the in-car experience, optimize its own operation and maintenance as well as the convenience and comfort of passengers. It works in different communications such as Vehicle-to-driver, Vehicle-to-vehicle, Vehicle-to-infrastructure, Vehicle-to-internet. Taking initiative to create modern structures is a demand to underpin the smart mobility and a good deal of things are considering raising the infrastructure. The major initiatives are EV charging infrastructures, smart roads, E-highways, etc. Among all of them, the EV charging infrastructures are the first step to boost up the EV uses.

10.2.1 Impact of the charging infrastructures on distribution grids One of the major challenges from the angle of distribution grid operators remains to foresee location and time of EV charging events to guarantee a continuous supply of energy and install enough power capacity within each service area of the distribution grid. This is especially important as EV adoption patterns might result in spatialtemporal clustering of charging demand which could cause unexpected overloads at secondary distribution levels. The above details become even more crucial due to the fact that it is necessary to take into account the impact of the mass adoption of EVs on the existing electric distribution grids so that the measures taken can fully work to cope with these effects. The facts and the norms set forth clearly state that in the coming future there will be a strong shift from the conventional/fossil fuel driven vehicles, and they will be completely eradicated from the roads. On the brighter aspects, this change will help combat the alarming carbon footprint increment in a much broader sense. Although the diffusion of EVs at a fast pace will benefit the society and the environment greatly, it is also important to consider the effect this huge diffusion of EVs will have on the existing grid infrastructure: network infrastructure, grid capacity, increasing demand for electricity as a result of the constant growth of its consumption. The deployment of EVs in huge numbers will create a need for equally available charging facilities. This wide increment in the EVs due to a complete ban of the conventional fossil fuel driven vehicles will create a huge impact on the traditional grid. It will lead to an urgent requirement for the modifications in the traditional grid planning. As a matter of fact, energy can be stored in different forms but electricity itself can’t be stored. On the contrary, the production and consumption of electricity are needed to be balanced continuously in every instance. The gap between production and consumption of electricity makes an effect on the electricity system parameters like voltage and frequency. Maintaining system frequency is one of the major

325

326

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

fundamental drivers of power system reliability. Similarly, the variation of voltage is one of the key elements characterizing the quality of service. Even the frequency of the power system must be kept within the nominal values, that is, 50/60 Hz. Automatic control systems are necessary for power systems to respond to the mismatch between the power generation and loads, otherwise, it can hinder the devices of the power system. For the same reasons, the frequency control is applied in three layers as Primary, Secondary and Tertiary. On the other hand, consumers feel satisfied when the system operates at its best in terms of performance & continuity with energy being supplied at the nominal voltage. Fluctuations beyond the set limits can have an adverse effect, thus hindering the system performance. This voltage variation is caused by the disturbances and control actions of reactive power sources in the grid. The disturbances may be the variation of load and variation of the system structure. So, to respond to these variations, voltage regulation is done, through which voltage is controlled in the grid by HV/MV transformers with On Load Tap Changer and MV/LV transformers merely with No Load Tap Changer. In recent days, the increase of nonprogrammable renewable energy-based power plants are significant. Load varies throughout the day; conventional generation can often deviate the schedule, besides the nonprogrammable power plants’ outputs vary on different time scales based on the weather. So, it is not possible to predict the nonprogrammable power plants generation with perfect accuracy. Similarly, the contingency of programmable power plants is unexpected, even the load forecast errors are unexpected. The traditional grid planning is already focused on coping with the tremendous increase in the consumption of electricity and meeting these needs, that are a result of the internal factors on the country-wise basis. Analyzing the effect of coupling the huge number of EVs to the grid for charging purposes, thus becomes more and more important. In order to carry out this analysis precisely, it is important to understand the load profile of these EVs that have charging requirements based upon their spatialtemporal locations. This is important because the diffusion of EV results in an increase in charging demand. As a result, the increasing charging demand will lead to the overloads at the distribution system levels. Before focusing on the load forecasting techniques, it is necessary to classify the load characteristics to understand the relationship between real or reactive power and system parameters such as voltage and frequency. The load characteristics influence is significant for both steady-state and dynamic state. Thus, appropriate load modeling is one of the most important tasks to represent the performance of those loads, which leads to suggest the planning and operation in power systems. The actual load modeling is finalized from the number of historical events which lead to component outages, voltage collapse, and system instability. We want to describe these events in terms of EVs, and it will lead us to respond adequately to the Energy Demand Management next.

The integration of electric vehicles in smart distribution grids with other distributed resources

• Outage: A power outage (also called a power cut, a power out, a power blackout, power failure or a blackout) is a short-term or a long-term loss of the electric power to an area. Typically, the power generation must follow the dynamic load demand; when they do not match, an outage occurs. Most of the time the event happened for the most optimized situation. A mass number of EV integration in the grid may make the grid more vulnerable and outage can take place frequently if the simulation results with the recorded response do not match. Therefore, regular load modeling is a must to avoid interruption. • Low frequency oscillation: Different load models such as a composite load model, the dynamic load model, and the exponential load model are required to handle low frequency oscillation and damping. Many studies show that the effects are different on the power system stability for the different load models. • System overloading: Charging procedure of many electric vehicles can create an overloading of distribution transformers and distribution lines. This is a sensible aspect especially for low voltage systems that have to embed many uncontrolled domestic charges in Mode 1 or 2. Fast and ultrafast charging stations usually require the construction of new lines or a suitable reinforcing of the existing distribution infrastructures. In order to face these issues, it is necessary to model the load with its characteristics. The load models are described as the following classifications and all types of loads make the composite load for a substation and by studying these models, we can control and monitor the voltage and frequency precisely of the busbars in a substation, as many loads are dependent on voltage (lighting, heating, etc.) and many of them have dependency of frequency (induction motors). Different load typologies are represented in Fig. 10.10. For analyzing the load profile, it is necessary to have a deeper insight into the traditional forecasting techniques. Considering from the broader perspective, load forecasting without a hint of doubt is crucial for utilities and other participants in electric energy generation, transmission, distribution and markets. Thus, it is necessary to have an overview of the load forecasting techniques in detail to have a better understanding as to whether they will cater sufficiently to the tremendous charging demand of the fast-growing EVs. By the definition of the most commercial battery types, it is well known that those are chemical storage devices and the charging and discharging characteristics depend on the types of chemical characteristics. Most of the time they are exponential functions over time. Considering the EVs battery we can analyze the following exponential formula for the battery systems for the instantaneous charging [4]:   αt ð10:2Þ PEV ðt Þ 5 PEV ;max : 1 2 e2tmax 1 PEV ;0

327

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Exxpoteenall Po olyno omial Lineear

Stattic

Inducon motor

Load models

328

Power electronics interfaced Exponenal dynamic Dyynam mic IM M Transfer funcon IM Compo osite

Dynamic

Distribuon n Bulk supply point DESS model

Figure 10.10 Different typologies of load models.

where, • PEV ;0 is the initial status of battery system while charging • PEV ;max is the maximum power capacity of the electric vehicle • tmax is the maximum charging time • α is the constant parameter declared by the EV manufacturers Additionally, considering the unity constant power factor of the EV’s battery system, the reactive power demand is zero, which implies QEV :DEM 5 0

ð10:3Þ

The equations prove that EV is an active load. Considering the EVs as an active load, now it is easier to form a composite model by aggregating all the loads under a busbar. As the utility service provider categorized the loads as residential, commercial, and industrial; the new EV loads will be included in the load table. In this way, it is easier for a utility service provider to find the ratio of different categories of the load in a busbar or in substation. Following the ratio, they may create the composite load models for the busbar too. As different types of

The integration of electric vehicles in smart distribution grids with other distributed resources

loads are connected to the busbar, the aggregation of such loads create a composite load model for the busbar considering the characteristics of the voltage and frequency dependency of the composite load. In usual cases of steady-state analysis, that is, the load flow analysis, it can be either the constant power loads, that is, P & Q remain constant, or it can be constant apparent power of the load. On the other hand, for the transient stability analysis, the load models usually are modeled as constant impedance load, which is connected as constant impedance at the bus. So, with the change in voltage the power drawn by the load changes; also, the power drawn by the load changes because of the change in frequency. In addition, the induction machine loads are modeled as constant current loads in transient stability analysis. At the end when we sum up all the loads as composite, we may find those sensitive with respect to the voltage and the frequency. So, the general equations for the load can be considered as below: P5 Q5 pvs 5 pfs 5 qvs 5 qfs 5

P0 3 pvs 3 pfs Q0 3 qvs 3 qfs k1 1 k2 V 1 k3 V 2 1 1 k4 f k5 1 k6 V 1 k7 V 2 1 1 k8 f

ð10:4Þ

where: • P is the active power at any voltage and frequency • P0 is active power at the rated voltage and frequency • Q is the reactive power at any voltage and frequency • Q0 is the reactive power at the rated voltage and frequency • pvs is the voltage sensitivity of the active power • pfs is the frequency sensitivity of active power • qvs is the voltage sensitivity of the reactive power • qfs is the frequency sensitivity of reactive power Higher-order frequency terms are neglected as the frequency band is very small. The load is considered to be a constant impedance, so the real power varies with the square of the voltage. Here, k1 to k8 are the constants for the different types of the loads and these values come from the percentage of various kinds of load in a busbar. By choosing different values of k1 to k8, we can model the loads as a composite load model. Again, decoupling the EV loads from the composite model to analyze them separately, it is just needed to solve the load flows for the EVs actually connected, but the percentage of the different loads on the busbar will roll back to the past values without the EV. So, it is certain that the new load is needed to be handled properly before it creates new issues in the power system.

329

330

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

As location wise most of the loads are static; but the EV changes its position dynamically considering the psychology and movement of the users. Conventional loads are always connected to the same busbar, while EVs have to be considered in probabilistic terms. This is one of the reasons to implement the Time-Space Theory concept to plot the graph of the stationary time of the EVs in certain places. It helps to predict the possible consumption or dispatch the power [Vehicle-to-Grid (V2G)] to a busbar. To secure the load model in more effective way it is now needed to integrate the geographic information systems (GIS) to the busbar. This integration leads the busbar to specify the location with proper data, that is, line/cable data, load/generation data and GPS coordinates/schematic diagram information. In order to analyze the proper implementation of the above-mentioned, it is important to first study the existing load forecasting techniques and evaluate their pros and the challenges. This in turn will lead the path for the updated Spatial-Temporal based load forecasting that forms one of the prime focus of the whole research. A prediction of electrical power required to meet the short term, medium-term or long-term demand is load forecasting. Looking at the forecasting methods, the most typical way of characterizing them is whether they are quantitative methods, or they are qualitative methods. Quantitative forecasting methods focus on collecting the historical data or time series and correlation data, and analyzing these data to obtain the forecasts as to what predictably can be the load profile in the future. On the other hand, there are the qualitative forecasting techniques, which focus on gathering opinions of the experts, policy makers, customers as to see that how the demand will be in the coming future in a specific situation. There is a further classification of the load forecasting techniques based upon the duration as short-term forecasting, medium-term forecasting, and long-term forecasting. Keeping at par with carrying out the forecasting in the respective durations, as a matter of fact, will, in turn, achieve maximum savings. This will eventually be the interest of both the generator, distributor and consumer of electricity, because if the forecasting is not carried out adequately then the resulting forecasting errors will increase the operation cost for power generation. Because of the operation cost increasing for the power generation, the other sectors of electricity transmission and distribution will also have to bear the consequences. The forecasting leads the utility companies to maintain the desired generation, expected operation & maintenance, and meet the demand of their customers. Electrical load forecasting is one of the most necessary process to the utility service providers which can increase the efficiency and revenues of the transmission system operator (TSO) and Distributed Systems Operator (DSO). It assists them to plan on their capability and operations to supply all consumers with the utmost reliability. To get the benefit of load forecasting, it is always needed to predict the real-time demand utmost. For this reason, different techniques have been adopted till now. Among many factors, the weather itself has big impact in load forecasting, and weather

The integration of electric vehicles in smart distribution grids with other distributed resources

Temporal sequences

specialists’ efforts are significant to forecast it closer to the real one despite sometimes the weather is unpredictable and depends on many other factors. In addition, the past data of the load consumptions is also a major consideration for the load forecasting but in recent days, the emergence of large number of EVs stressed the past load consumption data a lot. The vehicle was never been a part of the electric grid earlier, but the scenario is changing dramatically. The traditional load forecasting techniques are also getting shaped from the manual technique to database system. Specially the smart meter is also getting to play a major role for future load forecasting technique. Despite of the continuous improvement of load forecasting technique, there are still challenges and connecting EV in the grid is increasing the challenges more. So, we need to build a solution by discussing the evaluation of the classical and modern approach of the load forecasting techniques. The solution should minimize the complexity to add the new loads in the current load forecasting process. As mentioned before, load forecasts can be split into three categories: short-term load forecasts (STLFs) which are usually from 1 hour to 1 week, medium-term load forecasts which are usually from a week to a year, and long-term load forecasts which are longer than 12 months. The load forecasts for different time horizons are significantly important for variety of operations within a utility service provider. The integration of EVs requires a STLF as the movement of the EVs vary by the needs and day to day activities of the users. A large variety of statistical and artificial intelligence techniques have been developed for STLF. In the case of the EV scenario these days, GIS-based forecasting will provide the desired results with utmost accuracy. Because data needed for forecasting is updated on a real-time basis, this will ensure a real-time forecasting for EV charging loads. In Fig. 10.11, the model for implementation of GIS-based Spatial-Temporal forecasting is shown. The proposed charging scheme is based upon the parameters of an EV; mileage and battery capacity, and the data related to the battery charging; charging power and efficiency. The Spatial-Temporal model is obtained through the information about the activities, their duration and time, and their location. Time-based recording Change-based recording Locaon-based recording Event-based recording

Figure 10.11 Types of methods for obtaining the temporal sequences.

331

332

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

One of the main parameters is the Spatial-Temporal activity model. In order to obtain the model for the Spatial-Temporal activity, it is required to analyze the concept of Space-Time paths. Based upon the information and communication technology devices, there are a handful of ways to generate these temporal sequences, such as Event-based recording, Time-based recording, Location-based recording, Change-based recording.

10.2.2 Planning of the charging infrastructures An EV can be seen as a small storage system compared to the typical power flows that characterize a distribution grid, hence it is possible to identify two types of battery models for EVs: aggregate and individual model. In the aggregate battery modeling approach, batteries of all vehicles within a fleet are modeled as a single, so-called aggregate battery with a single State of Charge (SoC) state variable and a single charging power input. In the aggregate battery models a charging optimization for delivery EV fleets is based on dynamic programming method. Each individual vehicle is optimized separately within the fleet to provide a global optimal solution regarding the charging status. Considering an upper limit for the grid, the total optimizations are coupled together on the fleet level in a suboptimal way. Repeating this method for different levels in the network can lead to a global optimization. In this way the sensitivity of optimization results with respect to ordering of charging optimizations can be analyzed and a solution closer to global optimum can be found. Another method of charging considers the additional demand of EVs on LV distribution system. Problems such as thermal overload of transformers and lines, voltage deviation, harmonics, and phase unbalance are arising due to uncontrolled charging. The charging algorithms can be implemented in centralized and distributed mode. In the centralized method, it is used only when available power is measured at the distribution transformer and it allots equal shares to all connected vehicles. Instead, the distributed method uses local voltage measurements to determine whether the present network load is low (in which case the vehicle can be charged) or high (in which case the vehicle should not be charged). The distributed method, also known as voltage adaptive method, is mainly based on the SoC of EV and the voltage at the connection point to determine its charging power. Analyzing the scientific literature of various simulations compared with real data in a LV distribution system, it is possible to state that the location of vehicles in the networks has an undeniable impact on predicting the adverse effects. The more vehicles are connected to the same transformer, the more problems appear with regard to voltage drops, current unbalances, power factor, etc. Furthermore, the equal share (centralized) and voltage adaptive methods are slower in terms of speed of charging in comparison with the uncontrolled method, but the risk of above-mentioned problems is significantly lower.

The integration of electric vehicles in smart distribution grids with other distributed resources

In the case of charging stations equipped with photovoltaic generation, there is an algorithm that reduces the effect of intermittency of electricity supply and the cost of energy trading of the charging station. The vehicles are divided into three main categories considering their charging behavior: (1) premium (2) conservative and (3) green. Premium and conservative EVs consider interested only in charging their batteries, with noticeably higher rates of charging for premium EVs. Green vehicles are more environmentally friendly and thus assist the charging station to reduce its cost of energy trading by allowing the station itself to use their batteries as distributed storage. The proposed classification scheme is facilitated using the mixed-integer programming (MIP). The first step is the evaluation of the solar radiation data. Premium vehicles charge their batteries at maximum available charging rate and they do not allow any discharge of their batteries. Conservative vehicles have a lower rate of charging with respect to the premium ones, but also for this type of vehicles the discharge is not considered. The green vehicles have the lowest charging rate among all the types and they can be discharged at a certain rate. They get a discount on their charging price and are also paid a price per unit of the energy they sell to the grid. Formulating a basic MIP scheme captures the benefits of the proposed classification in reducing the charging station costs. The results determine that as the number of green EVs increases in the system, the total cost tends to converge more towards the baseline. Furthermore, as the percentage of green vehicles in the charging station increases from 10% to 100%, the difference for energy trading, between the total cost of the proposed scheme and the baseline approach, has been quantified as reducing by 85% in summer, and by 82% in winter. The reduction in their average total cost of a day charging during summer and winter is another benefit for EVs that choose the green charging behavior. The need for a recharge planning of EVs lies in the fact that user’s behavior, without any optimization factor, is such as to concentrate the recharges in the same hours, thus creating power peaks on the electric distribution grids. Indeed, starting from various researches in the scientific literature, the gathered data of the charging behavior of many EV users indicates that the charging plug- in time for 80% of the users is between 1 and 2 hours. The lowest average charging energy distribution is during weekends and the peak is recorded on the working days between 1 pm and 5 pm. A comparison among charging energy distribution in different seasons shows that in summer and winter the charging energy consumed is higher than the other two seasons. The reason comes from the fact that in winter and summer time, drivers tend to use heating and air conditioning system in a regular basis.

10.3 Vehicle-to-Grid The increasing penetration of EVs into the grid and development of smart grids has opened a door for the emergence of additional technologies based on how EVs

333

334

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

interact with the grid. V2G is one of the concepts which describes how a plug-in EV interacts with the grid by providing ancillary services, storing and transferring electrical energy produced by renewable sources like solar, wind which produce power depending on environmental conditions. These EVs are called gridable EVs (GEVs), that means that their interaction with the grid is bidirectional as they allow user to either buy or sell electricity from the grid.

10.3.1 The use of EVs for grid support Traditional power plants with fossil fuel have a very low efficiency from sources to end users, approaching an overall efficiency of about 30%, whereas RES have a high efficiency from generation to grid connection, approaching an overall value of about 70%. However, the intermittent nature of RES (such as wind power and solar power) adversely affects the grid voltage, frequency, reactive power, and so on. Hence, the power grid needs to be compensated or regulated. Studies have shown that a car remains parked on average for 95% of time either at home or office, therefore the batteries in the EVs can be used as energy storage systems (ESS) to improve the conditions of the grid and decrease the burden of DSO. Based on charging/discharge capabilities of electrical vehicles and energy requirements of the grid Vehicle-to-Home (V2H), Vehicle-to-Vehicle (V2V), and Vehicleto-Grid (V2G), in general V2X, enable GEVs to not only serve as a transportation mean, but also to act as controllable loads and distributed sources for the power grid. So, GEVs can play positive roles in the home grid, the community grid, and even the distribution grid during the charging and/or discharging period. Gridable electric vehicles usually allow V2X functions in charging Mode 4, i.e. with DC charging infrastructures, since it is easier to realize an off-board bidirectional charger than an onboard one, even if international standards allow V2X functions also in AC charging Mode 3. When talking about connecting GEV to the grid, there are two modes in which they can be used to provide benefits for the grid: V2G mode, which includes Vehicleto-Vehicle and Vehicle-to-Home, and Grid to Vehicle (G2V) mode. The G2V mode is used when supply is more than demand like during off-peak hours when power systems is operating on base load to charge the vehicles. G2V is sometimes also called as unidirectional V2G. V2G mode is used during peak demand to function as peaking power plants using aggregators to also provide economic benefits to the users. It can also be used to provide various ancillary services. A GEV can be used to improve power quality and stability of the electrical grid. GEV can use only its bidirectional charger which has a DC link capacitor to provide harmonic compensation without involving the batteries of the electric vehicle thus reducing number of charge/discharge cycles of the battery pack.

The integration of electric vehicles in smart distribution grids with other distributed resources

The term Vehicle-to-Grid (V2G) is a generalized term which covers different modes of operations. As previously discussed, it can be further divided into two modes: Unidirectional V2G or G2V, and Bidirectional V2G. Under Bidirectional V2G mode there are three main frameworks through which EVs and grid interact. This is represented in Fig. 10.12. Unidirectional V2G or G2V is the mode in which EV charging time is controlled. EV can contribute to the grid by charging only when capacity is available thus reducing the load on grid with uncontrolled charging. Studies have shown that controlled charging can level the loading of the LV transformer by utilizing extra capacity at night leading to delaying of network upgrade. This can be seen through the Fig. 10.13 [1]. Moreover, delayed charging leads to decrease in average SoC. This decrease in average SoC is predicted to increase battery life by about 9%, assuming all other conditions remain same. Since average daily car journey is usually less than 40 miles, with scheduled charging an EV is less frequently charged thus decreasing average SoC and leading increased battery life for the same number of miles driven. 10.3.1.1 Vehicle-to-Home V2H is a concept in which GEV Charge/Discharge at home using on-board of offboard bidirectional chargers for a single EV/home. V2H framework includes small scale renewable energy source, generally photovoltaic generators and/or small wind turbines, home appliances, and a bidirectional charger which includes a controller to which the GEV is connected. The battery of GEV can be utilized for active power exchange through the bidirectional charger which can use its DC link for reactive power compensation to the grid.

Figure 10.12 Vehicle-to-Grid classification.

335

336

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 10.13 Transformer loading (a) and voltage of the feeder (b) with an optimized charging strategy (solid line) and without any optimization (dotted line).

The V2H framework can achieve many features in its residential grid operations. V2H usually involves a single EV in a single home, so it has straightforward implementation, hence it is easy to implement in practice. A useful function of the V2H is to sell excess energy to the grid at high priced peak times and charge GEV batteries at less expensive off-peak times. Besides, it is able to act as a back-up generator and a controllable load, cooperate with domestic devices for load shift thus smoothing out daily load profile. V2H is able to provide reactive power compensation to home grid or community microgrid. Reactive power compensation for residential grid can be done without involving the batteries of GEV using only the capacitor present in bidirectional charger to decrease the number of charge/discharge cycles of the battery. Therefore, the information flow among the various elements during V2H, V2V and V2G operations has to be multidirectional. V2H has high efficiency and can also improve the performance of installed RES using GEV as energy storage. Installation of V2G can be done without change of existing home grid; its implementation can make smart home more attractive and improve development of smart grid. 10.3.1.2 Vehicle-to-Vehicle Charging EVs through the grid is accepted as conventional method. But with increasing penetration of EVs since charging infrastructure is limited due to variety of reasons it creates a burden on distribution systems with sudden increase in charging demand. Various studies have shown that users request for charging only after 25
30 miles of commute with more than 50% of battery capacity available. V2V framework allows users with available power to help those in need at suitable tariffs at charging stations. The use of V2V functions allows power reserve to be kept in the community and decrease the burden on the power grid.

The integration of electric vehicles in smart distribution grids with other distributed resources

Figure 10.14 V2V system schematic.

Fig. 10.14 shows V2V framework along with V2H and V2G frameworks [5]. V2V charging requires the aggregator to effectively combine the supplies with those requested and must also take into account the exchange of energy with different charging technologies. V2V framework offers the following features for community grid: • V2V involves multiple GEVs • Smart homes and public charging stations are used for energy transfers • V2H framework is included in V2V framework • Unlike in V2H, charging and discharging is controlled by the aggregator • Energy transfer is first done between community EVs and if necessary, power is drawn from the grid • V2V framework is comparatively less simple and flexible • V2V can be further integrated with renewable sources and ESS to establish an isolated community microgrid There are different ways to implement V2V framework and its viability in EV energy exchange through charging infrastructure. Such infrastructure will decrease market reliance on installation of additional charging stations to help with EV penetration. 10.3.1.3 Vehicle-to-Grid Due to small battery capacity of EVs, a single GEV or a small number of GEV present in a community does not affect the power grid. V2G framework aims to include large number of GEVs and their interaction with the power grid. Fig. 10.15 shows the structure of V2G network in which V2H and V2V along with smart building and charging stations and EV aggregator as its elements [5]. According to grid voltage level

337

338

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 10.15 V2G system structure.

The integration of electric vehicles in smart distribution grids with other distributed resources

there are two zones in which V2G can be divided. LV network consists of V2H and V2V units along with smart micro grids. MV network consists of charging stations characterized by DC fast charging stations. V2G has the following features in power grid: • V2G framework involves a large number of EVs • V2V and V2H frameworks are included in V2G • Smart home, public and fast charging stations are used for power transfer • V2G provides ancillary services to grid in both MV and LV levels • V2G is the least simple and most flexible framework Aggregators used for V2G take into account the additional constraints such as placement of charging station, generation and transmission parameters, and network parameters. Besides V2G can also be used to temporarily stabilize the grid. Research and development on GEV and its interactions with grid is mainly based on the economic aspect. Some works prove the economic gain that is possible by implementing a smart charging algorithm that takes into account time varying tariffs to charge/discharge the battery and decreasing household load during peak demand [6]. GEVs can be used as offline uninterruptable power supply to improve grid power stability as an interesting V2H implementation [7,8].

10.3.2 V2G functions for frequency regulation The way to keep constant the power frequency is to balance instant by instant the power generation with power demand. In the presence of high penetration level of renewable sources, due to their unpredictability, this balancing requires a high level of spinning reserve or the curtailment of renewable sources. In any case, the cost for this service increases as the RES increases, besides, and the global efficiency decreases. A solution for this problem can be the installation of large ESS, but up to now they are still not economically profitable with the only exception of pumping hydro storage systems. However, thanks to the automotive technologies, the costs for the Li-ion batteries continuously decreases, so they are under consideration also for frequency regulation in fixed installations. EVs can be considered, when connected to the grid through their charging stations, as buffers which can provide to the grid a certain amount of power or energy according to their SoC, the demand on DSO networks and EV-owner willing to participate as an active user inside the grid. The rising number of local devices connected to the LV grid producing, consuming and storing electricity drives the need for integration and control. Especially, the complementary character of storage capacities and renewable energy supply calls for an

339

340

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

intelligent integration as the benefits can only fully be realized if they are managed jointly within a network. This bears new challenges for the TSO and DSO. EVs with the capability of bidirectional charging are not only energy storage units but also controllable energy consumers within a grid system. To activate this potential the grid needs to get smarter and include a power management system by incentivizing energy consumption and allocating energy reserves where they are needed most. Adding Advanced Metering Infrastructure to the distribution system, such as digital sensors and remote controls, increases information exchange and thus transparency among all involved players. In summary, a smart grid allows a more efficient coordination between power generation and power consumption. Adapting the consumption to the supply of distributed generation sources has its merits for the DSO as transmission lines do not need to be expanded but can also pose a threat. Increasing the consumption share of self-generated solar power within a microgrid leads people on the path of independence. As the proportion of variable RES increases, the fluctuations in the total supply also rises. The ability to accurately forecast the generation capacity—along with loads—would improve the efficiency. However, forecasting the state of such a complex system, with the vast number of dependencies and interdependencies, from weather to energy prices, calls for new levels of intelligence. EVs and, more in general, batteries can provide several support services to grid systems, both power intensive and energy intensive. The benefits attainable are both technical and economical. The main services are listed here: • Peak shaving: this occurs when a temporary increase in power on the network is requested and some power plants are temporarily put into service. With V2G technique, EV batteries can be used for this function in a cheaper and faster way with respect to putting a traditional plant in service. The duration of this kind of service is on average 3
5 hours, coinciding with the hours of peak demand. In V2G mode, this power request can only be provided by a large number of EVs connected at the same time to the network in a limited area. The batteries of EVs can supply energy (through the discharge) during the phases of high demand and absorb it (through the charge) during the phases of low demand, in order to decrease the difference between the maximum and minimum; • Spinning reserve: this term refers to devices that are able to provide power peaks quickly (max 10 minutes) at the request of the network operator. Typically, these generators are called to produce power 20
50 times a year and are paid based on their ability to produce energy during an unplanned event and in relation to the power delivered. This is a perfect condition for the exploitation of EV batteries, which constitute a spinning reserve just because they are connected to the grid and have great ramping characteristics. The user also benefits from this condition because the energy delivered is paid based on the spinning time;

The integration of electric vehicles in smart distribution grids with other distributed resources

• Frequency regulation: the grid frequency must be maintained at 50 or 60 Hz to guarantee the correct functioning of the loads connected to the network. If the frequency rises, it means that the loads connected to the network are not sufficient to absorb all the power generated, therefore it is necessary to increase the load on the network or decrease the power produced. If the frequency drops, it means that the load connected to the network requires a higher power than the one supplied, therefore it is necessary to reduce the load or to increase the power injected. The frequency regulation must take place under the direct control of the grid operator, which sends signals to the generators: they must respond within 1 minute by increasing (adjusting upwards) or decreasing (adjusting downwards) the power generated. In V2G mode the batteries offer a valid aid for frequency regulation because through the charging process they can absorb power from the grid in case of over-generation, while with the discharge process they can operate as a generator in the case of subgeneration. This process is also called active regulation; • RES support: one of the tasks of V2G is to support RES in order to optimize network balancing operations. Wind power and photovoltaics are two nonprogrammable RES and often their production peaks do not coincide with the demand of the network. EV batteries allow to absorb the renewable energy produced during periods of low demand and to re-inject it into the grid at times of higher demand. Therefore, there is a temporal decoupling between the production of energy and its use; • Coordinated charging: not strictly a support service, but fundamental to avoid criticalities. Charging of EVs add an extra burden on the current distribution grid and, if not coordinated properly, it can cause expedited aging of the equipment and tripping of the relays under severe overload conditions. Thus, grid improvements are necessary to alleviate the tension on the lines in some cases. However, in the majority of cases, this problem can be solved by intelligent charging methods such as coordinated charging, which means that in nodes that are far from the slack bus, vehicle charging is scheduled to occur at moments of low demand. Coordinated charging does not require upgrading the distribution infrastructure. However, the grid operator/aggregator should reach the charger of EVs to perform demand response actions for coordinated charging (e.g., shedding the EVs chargers from the grid in case of overload). As previously mentioned, other services, on a smaller scale, typically V2H, include: • Back-up: in case of a disconnection from the network of the load it is possible to use the energy stored in the EV battery; • Time-of-use bill reduction: the EV battery allows to store energy at off-peak times (low price) and use it at peak times (high price) in order to reduce the cost of electricity for a load. Typical times for these services are summarized in Table 10.1.

341

342

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 10.1 Grid service overview. Service

Duration

Peak shaving Spinning reserve Frequency regulation RES support

30 min
5 h 15
20 min 1
5 min Seconds
30 min

10.3.3 Synergies between electric vehicles and renewable energy sources The synergies between RES and EVs lie in the fact that usually both of them are close to the end users, so their vicinity decreases the part of distribution grids involved in the energy exchange, and in the flexibility of the EVs charging procedure allowed by V2G that compensates the variability and unpredictability of the RES. Analyzing the daily energy demand, it is easy to notice why RES brought instability to the system. In fact, during night, the demand is lower with respect to the daily hours, between 9 and 11 AM there is a first demand peak and between 17 and 19 there is the maximum peak. This load profile has the same shape of the mean daily profile during the year. Some consideration must be made about these peculiarities of the load curve. If wind power generation is considered, during night we have an increase of power generation, keeping all the other factors constant, due to the higher density of the air ρ. Being the power generated by a wind turbine P 5 12 UρUv3 , the higher the fluid density, the higher is the power that can be extracted from its kinetic energy. So, the extra power production generated during the night cannot be exploited since the energy demand is very low. During day, instead, considering constant the share of power generated by wind generators, the problem is shifted on photovoltaic generation. Since photovoltaic generation is proportional to solar radiation incoming the Earth’s surface, it is trivial that the photovoltaic power peak generation will be distributed around midday. Also, in this case there is a power peak generation, shown in Fig. 10.16 [9], which does not match with the energy demand curve. It is clear that in order to fully exploit RES production, which would lead to reduction in primary sources consumption and CO2 emission, it is necessary to manage the grid to shift power production and power consumption in a way to maximize RES energy production [10]. Analyzing, for example, the EV charging behavior and the photovoltaic generation over a year, as depicted in Fig. 10.17, a strong mismatching between them in all seasons can be clearly seen. However, acting on the flexibility of the EVs charging and implementing the V2G functions with a bidirectional power flow as previously described, it is possible to

The integration of electric vehicles in smart distribution grids with other distributed resources

Figure 10.16 Typical generation profile of a photovoltaic plant.

Figure 10.17 Comparison between EV charging behavior without any optimization and the generation profile of a photovoltaic generator.

343

344

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 10.18 Optimization of EV charging behavior through V2G to maximize the photovoltaic generation.

improve the exploitation of the photovoltaic generation minimizing, at the same time, the power exchange with the grid as shown in Fig. 10.18.

10.4 Conclusions With the increase of the share of EVs in the global fleet, smart integration in the electrical ecosystem is both an opportunity and a necessity. The support that V2X systems can provide to distributed generation and RES could help the transition to a decentralized, green and smart energy network. This integration requires, first of all, the standardization of charging protocols and the further adoption of ICT devices both on EVs and charging equipment. Reduction of the costs of EVs and batteries, along with improved performances, will progressively water down barriers for the electrification of the fleet. Widespread charging infrastructure will tackle psychological factors, such as range anxiety. Smart grid development is the main enabler of this change of paradigm: investments are needed in order to have the possibility of managing all the resources connected to the network, both producing and consuming electricity. Regulation is another critical factor: heterogeneous framework is not helping the transition, with many countries still behind the adjustment of markets to let complex products and unconventional resources, such as V2G and DR, participate in the trading and balancing mechanisms.

The integration of electric vehicles in smart distribution grids with other distributed resources

On a smaller scale, V2H paradigm is the closest to widespread implementation, as the investment in terms of power electronics and ICT is affordable. Therefore, wide diffusion of EVs represents a great opportunity, not only to make sustainable the mobility sector, but also to work in synergy with distribution grids to improve the quality of the power and increase the percentage of RES that they can embed. Ultimately, EVs are not to be seen as another burden for the electric power system, but they are a way to move towards the decarbonization of this sector.

References [1] European Standard IEC 61851-1:2017, Electric vehicle conductive charging system - Part 1: General requirements, by CENELC European Committee for Electrotechnical Standardization. [2] R. Miceli, Energy management and smart grids, Energies 6 (4.) (2013). [3] T. Morgan, Smart grids and electric vehicles: made for each other? Discussion Paper 2012-02 Menecon Consulting UK. [4] R. Garcia-Valle, J.G. Vlachogiannis, Letter to the editor: electric vehicle demand model for load flow studies, Electr. Power Compon. Syst. 37 (5) (2009) 577
582. [5] C. Liu, K.T. Chau, D. Wu, S. Gao, Opportunities and challenges of Vehicle-to-Home, Vehicle-toVehicle, and Vehicle-to-Grid technologies, Proc. IEEE 2013 101 (11) (2013). [6] A. Alahyari, M. Fotuhi-Firuzabad, M. Rastegar, Incorporating customer reliability cost in PEV charge scheduling schemes considering Vehicle-to-Home capability, IEEE Trans. Vehicular Technol. 64 (7) (2015) 2783
2791. [7] H. Shin, R. Baldick, Plug-in electric Vehicle-to-Home (V2H) operation under a grid outage, IEEE Trans. Smart Grid 8 (4) (2017) 2032
2041. [8] V. Monteiro, B. Exposto, J.C. Ferreira, J.L. Afonso, Improved Vehicle-to-Home (iV2H) operation mode: experimental analysis of the electric vehicle as off-line UPS, IEEE Trans. Smart Grid 8 (6) (2017) 2702
2711. [9] T. Gerke, Sunday, solar sunday: Germany’s recent solar energy record in-depth, ,http://theenergy collective.com/thomas-gerke/248721/sunday-solar-sunday-germany-s-july-7-solar-power-recorddepth., 2013 (accessed 17.06.20). [10] G. Lacey, G. Putrus, E. Bentley, Smart EV charging schedules: supporting the grid and protecting battery life, IET Electr. Syst. Transportation 7 (1) (2017) 84
91.

345

CHAPTER 11

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER C.N. Papadimitriou1, M. Patsalides1, V. Venizelos1, P. Therapontos2 and V. Efthymiou1 1 FOSS Research Centre for Sustainable Energy of University of Cyprus, Nicosia, Cyprus Electricity Authority of Cyprus (EAC), Nicosia, Cyprus

2

Abbreviations aFRR DC DER DSO DR EC ESSs EVs FCR FRR GC HILP KPIs LC LTC LIHP LV mFRR MV NPV RACDS RC REP RES RoCoF SAIDI SAIFI TR UAS

Automatic Frequency Restoration Reserves Direct Current Distributed Energy Resources Distribution System Operator Demand Response European Commission Energy Storage Systems Electric Vehicles Frequency Containment Reserves Frequency Restoration Reserves Generation Cost High-Impact Low-Probability Key Performance Indices Labor Cost Load Tap Changing Low-Impact High-Probability Low Voltage Manual Frequency Restoration Reserves Medium Voltage Net Present Value Resilient AC Distribution Systems Replacement Cost Reactive Energy Provided Renewable Energy Sources Rate of Change of Frequency System Average Interruption Duration Index System Average Frequency Index Restoration Cost Utilization of DER for Ancillary Services

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00010-8

r 2021 Elsevier Inc. All rights reserved.

347

348

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Nomenclature CC Net load Pconventional PRES AENS ENS LPENSi Ci frestored fnom or fn f df/dt Pg Pl Hsys UAS EAS Etotal TR LCi RCi GCi R Rpdo Pshed

capacity credit reserve margins [MW] generation from conventional resources [MW] generation from renewables resources [MW] average energy not supplied [MW] energy not supplied [MW] load point i energy not supplied [MW] customer i frequency restoration control effectivity [Hz] nominal frequency [Hz] system frequency [Hz] rate of change of frequency [Hz/s] generators’ active power [pu] demand active power [pu] system inertia [s] utilization of DER for ancillary services [%] energy used for ancillary services [MVA] total energy produced [MVA] restoration cost [h] labor cost [h] replacement cost [h] generation cost [h] resilience indicator resilience post disturbance power that was shed [MW]

11.1 Introduction This chapter’s main objective is to describe the renewable energy sources (RES) impact that mainly stems from the uncertainties inherent in their stochastic nature. This chapter analyses how these characteristics affect power system’s reliability and resilience under the operational challenges point of view and more specifically under the day- ahead operational challenges for a system operator. Among others, a new approach for assessing operational resilience of a system is described and a new index is formulated. Under this prism, the impact of high RES penetration is studied using extensive numerical studies on a real grid of Cyprus with projection of the future scenarios. The reader of this chapter can find gathered all the operational impacts of RES and link them with mathematical equations and Key Performance Indicators (KPIs) under the reliability and resilience point of view to use them as a basis for his own research endeavors. The rest of the chapter is organized as follows: Introduction provides the objectives, the motivation and the scope of the chapter providing general information about the

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

present and future status of the power system. Section 11.2 analyzes the impact of the RES penetration on the reliability of the system and more specific on the linked daily operational challenges of the system. Section 11.3 analyzes the impact of the RES penetration on the operational resilience of the system and interlinks them with the flexibility of the system. These sections provide the theoretical background for establishing KPIs for systematically studying a grid with high RES penetration in next section. So, Section 11.5 provides a methodology for projecting the present real Cyprus grid to its future version and study it thoroughly based on the KPIs of the previous sections. Last section provides the conclusions and main outcomes of this chapter study.

11.1.1 Present and future energy landscape Traditionally, distribution networks have been mostly passive with a relatively small percentage of distributed energy resources (DER). As electricity grids are slowly getting smarter, the integration of distributed generation is rapidly increasing. In addition, the environmental issues, that play a dominant role on the political agenda, have favored the significant growth of renewable energy-based DER. As the percentage of renewable generation capacity is rising, new challenges are introduced in the smart grid. Power flows in distribution networks are no longer unidirectional, that is from the point of connection with the transmission network down to customers. In many cases the flows may reverse direction when the wind or solar irradiation is strong, and therefore renewable generation exceeds load, hence under these circumstances, the distribution networks become net exporters of power. Due to their variable output, these new technologies are quite different from traditional generation resources. This inherent characteristic of RES creates many technical problems with respect to settings of protection systems, voltage drops, congestion management and other quality attributes of power systems. As distributed renewable resources increase, their intermittent and variable nature, if not compensated, adds more uncertainty to the system, which adversely affects the reliability of the interconnected systems. RES production has been promoted in the past years through different incentive schemes throughout Europe. The process of their integration in the distribution system and their mode of operation was standardized in many ways. In the present grid, RES production aims at the maximum power point of operation without providing any kind of support to the grid through their electronic interfaces. Yet, this is about to change in order to secure the efficient operation of the system, mainly for the following reasons: • Higher RES penetration is expected to grow in order to fulfill the EC commitments.

349

350

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

Electrification of other sectors, for example, transport, heating will grow aligned with higher efficiency and sustainability objectives. • Emerging technologies needs for seamless integration. So, in addition to the familiar wind power and rooftop solar generators that are proliferating across every continent, electric vehicles (EVs), Energy Storage Systems (ESSs) distributed or acting through controlled areas like microgrids, are starting to grow complimenting the needs of the grid with a promise to become a DER power source in the foreseeable future. The electrification of vehicles and buildings (e.g., heating, cooling, water heating), as the efficient way to decarbonize other sectors of the economy, will render society even more dependent on electricity, and therefore even more vulnerable to potential power disruptions if RES are not complimented through other supporting technologies. With the growing share of RES penetration in the power grid, the increasing electrical load due to the electrification of heating and transport, and the exploitation of flexibility resources at distribution level, a number of network operation issues may arise. The difficulty associated with integrating variable sources of electricity stems from the fact that the power grid was designed around the concept of large, controllable electric generators. Intermittent renewables are challenging because they disrupt the conventional methods for planning the daily operation of the electric grid. Their power fluctuates over multiple time horizons, forcing the grid operator to adjust its day-ahead, hour-ahead, and real-time operating regimes. Although the practical capacity of these systems is relatively small, their cumulative effect significantly alters the behavior of the feeder where they are installed. In all distribution networks, challenges arise from the technical characteristics, the end-user operation of electrical loads, and the network assets. For the broader power system, the variability of renewable generation can also complicate the procurement and dispatch of other generators and increase their ramping requirements, among other operational challenges [1]. To ensure that the end-user equipment and infrastructure can operate safely and effectively, electricity grids must have standard conditions of supply that are commonly referred to as power quality requirements and are defined in standards or in operational codes. The active management of the distribution system with DER, microgrid, and smart technologies is expected to address such challenges shifting the traditional passive distribution network to an active/smart distribution network.

11.1.2 Future system grid projection While numerous existing studies address how to improve the reliability and resilience of the grid of today, which contains a mix of fossil and nonfossil power generation, little research exists on the resilience and reliability of a decarbonized grid in the light of

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

dominance of renewable DER. Recently, smart methods attracted researchers’ attention as much more efficient actions to make system sufficient resilience. There are various resilience-boosting actions in the context of smart techniques such as DER and Demand Response (DR). Also, the emergence of the concept of microgrids and web of cell architectures, that are based on the capability of DER’s fast response and advanced control schemes are enhancing further system’s resilience. Especially enabling these control schemes -either in interconnected or islanded mode- is considered one of the major resilience-improving approaches. Network microgrids, which are interconnected microgrids, are expected to be even more effective than single microgrids as they can support each other during the event [2]. In this manner, local DER’s can be optimally dispatched and operated in order to ensure that at least their critical loads would remain operational [3]. While the Direct Current (DC) loads and generation are increasing, the concept of hybrid networked microgrids has emerged. In hybrid microgrids, Alternating Current (AC) resources and loads are connected to an AC microgrid which is connected, via an interlinking converter, to a DC microgrid. It has been found that also these type of microgrids can significantly reduce the amount of load shedding during major events in a relatively economical operation [2] (Fig. 11.1). The continuing modernization of distribution systems can be characterized as resilient AC distribution systems (RACDS). The RACDS concept is to improve resilience of distributed systems systematically by deploying power electronics for DER integration, voltage/ var/frequency control, and efficient consumption of electricity by loads, which includes microgrids, the use of RACDS devices (Flexible AC Transmission system (FACTS)-like control devices at distribution levels), and meshed distribution systems (looped feeders and local generations). The “ideal” interconnected distribution network within a smart grid (the potential locations for installation of RACDS are marked) is presented in [4]. Unlike the centralized generation mechanism, the local generation requires power electronics to interface with this uprising “ideal” distribution system, thus establishing the high proliferation of power-electronic (PE)-DER. This large-scale deployment introduced new Microgrid 2 DER Central interface Intercoupling Intracoupling

Microgrid 1

……………….. Microgrid n

Figure 11.1 Concept of microgrids.

351

352

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

opportunities and advanced solutions for power distribution system resilience against the high-impact low- probability (HILP) incidents as well as unpredictable faults and disturbances. In abnormal operation scenarios where either there is a fault and voltage sag or when a sudden variation of the unbalanced load happens, the voltage at the grid-edge connection of the PE-interfaced DER will experience distortions and the voltage source inverter (VSI) adjusts the DER output power to regulate the voltage accordingly. The VSI has the ability to restrict the DER power injection to the grid to handle the over-voltage conditions, or increase the DER output power to mitigate the voltage dips [5]. Another key enabler for effectively monitoring and managing the distribution grid of the future, thus improving its robustness, lies with the utility’s ability to ingest, route, process and act upon the increased levels of data from DER and the larger grid. With the increasing deployment of information technology solutions such as enterprise service buses (ESBs) and stream processing tools, the time interval between “sensory” and “actionable” information has the ability to be dramatically compressed, consequently reducing the time between grid critical event and effective utility response. Essentially, real-time information fed from DER to the utilities can be exploited by the new generation of Distribution Management System power flow applications which can improve observability, reliability and ultimately resilience of the power grid. In this context, concepts of smart district electricity networks (e.g., microgrids, web-of-cells, etc.) with proper enabling technologies may accelerate the paradigm shift in delivering resilience and security of supply from redundancy in network assets and preventive control to more intelligent operation at the distribution level through corrective control actions supported by a range of enabling technologies and Information and Communication Technology (ICT). Smart district electricity networks may be able to alleviate grid disturbances, serve as a grid resource for faster system response and recovery, and strengthen the overall supply resilience to end consumers [6].

11.2 RES uncertainties description and assessment Within this section, the impact of the RES uncertainties on the reliability of the power systems is discussed. The focus will be given on the impacts related to the dayahead operational scheduling whereas other impacts on reliability will be mentioned. These impacts will result to a table of KPIs that can be used for the analysis and study of the power grid systems in a systematic way.

11.2.1 Impact of RES on power system grids 11.2.1.1 Impact of variability in the secure and efficient operation of the power system A critical issue in power system operation with a large capacity of intermittent production is the volume of operating reserves that will be needed to preserve the secure and efficient operation of the power system as well as to maintain the load-generation

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

balance considering the highly uncertain production of RES. From a system perspective, integrating nonmanageable stochastic generation constitutes a challenging task. Aspects such as low availability, lack of correlation between RES generation and energy consumption, as well as absence of firmness in generation programs, among others, impose new power balance challenges given that, electricity systems should be constantly adjusting to fluctuations in demand and generation. The value of a variable generation resource towards providing reliable capacity to meet planning reserve margins is measured by its “capacity credit,” which is a ratio of the power output during peak demand periods and the rated capacity of the intermittent resource [7]. CC 5 1 2

Pwith 2 Pwithout PRES

ð11:1Þ

Studies of capacity credit show wide variation for different RES types. The capacity credit for a fossil fuel or nuclear plant is typically 90%95% (the capacity credit less than 100% is due to unplanned plant outages and summer derated capacity). Because wind resources are not typically well-correlated with peak demand, wind capacity credits are generally low—in the range of 5%40%, while capacity credits for PV range from approximately 30%75% [8]. The impact of variable renewable generation to reserve margins is also often measured by the “net load” which is equal to the conventional way generated load minus the load generated by renewable sources. Net Load 5 Pconventional 2 PRES

ð11:2Þ

The impact of large penetration levels of RES is very evident by looking the progress of the electricity net-load requirements of California [9]. As more solar power was generated during the sunny middle of the day, the net load on the network decreased (duck’s belly). This is followed by a steep ramp-up to peak demand in the evening (duck’s neck and beak), putting enormous pressure on a grid that was not designed to accommodate efficiently such variation (Fig. 11.2). The increased variability and uncertainty allied with integrating large shares of renewable resources into the power grid will consequently have implications for the capacity required to maintain the robustness of the grid. It is important to stress at the outset, as we look towards 2030, that the uncertainty as well as mismatch between demand and supply will be further enhanced as a range of factors will be at play— including possible improvements to energy efficiency, changing behaviors and social norms, electrification of heat and transport, and the establishment of ESS in all electricity sectors—all of which make predicting future demand with any certainty challenging. To efficiently meet the variation in the electric load demand with large amounts of intermittent renewable resources at all times, with the same level of robustness as the current

353

354

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 11.2 Electricity load requirements highlighting the duck curve phenomenon (past and future estimations for the case of California) [9].

system, the integrated grid requires re-alignment in making available the appropriate sources that can manage effectively and efficiently the anticipated fluctuations. In order to carry out an accurate reliability analysis within a short horizon (e.g., a day), it is important to represent network operation in detail. At the operational stage, reliability requirement is typically assessed as a deterministic function or rule, which is typically related to the Average Energy not Supplied and Energy not Supplied. The former represents the average amount of energy not supplied, for all customers in units of [MWh/Ca], while the latter is the total amount of energy on average not delivered to the system loads during a certain period when comparing a reference case to a study case, in units of [MWh/a]. ENS AENS 5 P i Ci ENS 5

X

LPENSi ½MWh=a

i

where: AENS: Average Energy Not Supplied ENS: Energy Not Supplied LPENSi: Load Point i Energy Not Supplied Ci: Customer i

ð11:3Þ ð11:4Þ

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

This equation can stand as a KPI and be applied to energy supplied from a certain source only, for example, distributed generators in the Distribution System Operator (DSO) network. The reason for energy not supplied may be that active power injection by distributed generators is getting limited, for example, by grid congestion avoidance or reactive power infeed for voltage control. In this case, the KPI is an important means of comparing different grid operation approaches or an index of RES uncertainty [10]. 11.2.1.2 Impact on overall inertia The most vital factors for the stability of a power system are its mechanical inertia, delivered by the rotating masses of all the conventional turbines and the electricity generators, and its capability to damp any perturbation. In principle, the inertial response of wind farms and solar PV plants to the overall power system is nearly negligible. Hence, in systems with a high penetration ratio of RES systems, the operational inertia of the system may be abridged and the system response to large disturbances could be drastically affected. This is likely to occur for system conditions with a high level of renewable generation and low demand levels. Of course, the high uncertainty of the RES production can induce great unbalances in the system resulting in instability. Especially, small standalone interconnected systems, such as the Irish or the Hawaiian power systems, are more exposed to contingencies like the sudden loss of generation [11]. Consequently, system inertia drops during times of high wind and PV generation leading to faster frequency drops (or increases) in the case of a sudden generation shortfall (respectively surplus), as it may occur in the case that a large generator is suddenly disconnected from the system or that an area of the system that operates under heavy import or export conditions is suddenly disconnected and goes into island operation. Frequency Containment Reserves (FCR), also known as primary reserve power, is the power that is activated under the control of the primary regulation. The amount of primary control reserve that needs to be preserved is typically related to the largest units of a system. If the outage of the largest generating unit must not result to any load disconnection, adequate primary control reserve power must be allocated in the system for recovering this event. Technically, it would be possible that renewable generation contributes to primary frequency control. However, as for every power plant, it is required to limit the power output of a Power Park Module in order to enable primary control reserves. Contribution of renewable generation to primary reserve power can be possible in cases where storage is available as the required reserve power could be made available by the associated storage device. FCR must be activated within a few seconds and must be made available to the system until the secondary control reserve has been fully activated (typically 5 minutes). Frequency Restoration Reserves (FRR), or secondary control reserve, is responsible for restoring the nominal frequency after the occurrence of

355

356

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

an imbalance and can be activated by an automatic control process (aFRR) or manually (mFRR). In the time frame above 15 minutes (regulatory reserve, load following reserve), the variable nature and the high unpredictability of renewable generation has significant impact on the required active power reserve. Frequency stability can be assessed through the following proposed KPIs: x Frequency restoration control effectivity Frestored 2 Fnom # ε; ε-0

ð11:5Þ

x Rate of Change of Frequency (RoCoF): This quantity was traditionally of minor relevance for systems with generation mainly based on synchronous generators, because of the inertia of these generators, which inherently counteract to load imbalances and thus limit RoCoF in these cases. It however becomes relevant now during significant load-generation imbalances (caused by disconnection of either large loads or generators, or by system splits), when larger RoCoF values may be observed because of low system inertia caused by (amongst others) disposal of synchronous generation in case of high instantaneous penetration of nonsynchronously connected generation facilities [12]. Large df/dt values may endanger secure system operation because of mechanical limitations of individual synchronous machines (inherent capability), protection devices triggered by a particular RoCoF threshold value or timing issues related to load shedding schemes. Pg 2 Pl df 5 dt 2Hsys

ð11:6Þ

where, df dt : rate of change of frequency ½Hz=s Pg : generators0 active power ½ pu Pl : demand active power ½ pu Hsys : system inertia ½s The withstand capability requirements can be defined as a set of frequencyagainst- time profiles. The frequency-against-time-profiles shall specify lower limits for under-frequency and upper limits for over-frequency of the actual course of the frequency deviation as a function of time before, during and after the frequency event. Such profile should include frequency nadir/zenith, steady state frequency offset, the duration that the user has to stay connected and consecutive df/ dt ramps [13].

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

Frequency nadir: The most important aspect of frequency behavior following the sudden loss of generation is the point at which frequency is arrested or the frequency nadir. If frequency nadir is greater than (i.e., frequency is arrested above) the highest set point for under-frequency load shedding, then the primary frequency control reserves that were in place at the time generation was lost were adequate. If, however, frequency decline is not arrested and frequency crosses below the highest set point, firm customer loads will be dropped through the actions of under-frequency load shedding. This means the primary frequency control reserves that were in place were inadequate [14] (Fig. 11.3). maxðfn 2 f Þ½Hz

ð11:7Þ

fn : nominal frequency ½Hz f : system frequency ½Hz x Frequency zenith: Similar to the frequency nadir, the frequency zenith is also a direct measure of how close a system has come to interrupting the delivery of electricity to customers in case over-voltage event surpasses a threshold frequency value. minðfn 2 f Þ½Hz fn : nominal frequency ½Hz f : system frequency ½Hz x and Indication of Stability, which is a Boolean variable (YES/NO)

Figure 11.3 Frequency nadir representation [12].

ð11:8Þ

357

358

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

11.2.1.3 Impact on voltage regulation Radial distribution systems regulate the voltage by the aid of Load Tap Changing (LTC) transformers at substations, by line regulators on distribution feeders and shunt capacitors on feeders or along the line. Voltage regulation is based on one-way power flow where regulators are equipped with line drop compensation (Fig. 11.4). Large integration of RES may result in fluctuations in voltage profile along a feeder by altering the direction and magnitude of real and reactive power flows. This impact on voltage regulation can be positive or negative depending on distribution system and distributed generator characteristics as well as the RES location. Additionally, the high levels of injected active and reactive power due to the large penetration of RES systems along with the power distribution feeders may lead to over-voltage issues. Of course, the uncertainty of RES production and their stochastic nature can also induce the exact opposite issue of undervoltage, for example, due to temporary cloudiness. For instance, a small RES system sharing a common distribution transformer with several loads may raise the voltage on the secondary side, which is sufficient to cause high voltage at end-user side. This is possible if the physical location of the distribution transformer is at a point on the feeder where the primary voltage is near the threshold limits. The conducted case study assesses the voltage quality by investigating the voltage magnitude variations. Voltage variation (over-voltage or undervoltage) is the deviation of voltage in a certain range and can be identified by monitoring the bus bar voltages of the grid substations. • For low voltage (LV): - 95% of the 10-minute mean r.m.s values for 1 week (610% of nominal voltage). - 100% of the 10-minute mean r.m.s values for 1 week (110%/215% of nominal voltage) • For medium voltage (MV): - 99% of the 10-minute mean r.m.s values for 1 week below 110% of reference voltage and 99% of the 10-minute mean r.m.s values for 1 week above 210% of reference voltage.

Figure 11.4 Frequency zenith representation [12].

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

- 100% of the 10-minute mean r.m.s values for 1 week (615% of reference voltage) According to the defined EN 50160 Standards [15], bus bar voltage magnitudes must comply with following allowed range of variation. LV: (610% of nominal voltage) MV: (65% of nominal voltage) Voltage deviation indices can be defined to find the frequency or duration that the bus bar voltages violate the allowed voltage range. x Number of voltage excursions exceeded n minutes per year x Percentage of time that the transmission voltage exceeds the permissible limits. 11.2.1.4 Other impacts of RES on the system As already mentioned, this sub section summarizes -for reasons of completeness- the different impacts of RES on the system that do not affect directly the daily operation and scheduling of the system and that are based on the technical specifications of the RES integration. Technical failure: Another variability factor that can potentially affect the robustness of the power grid is operation failure of the components and configuration systems that are associated with the renewable generating units. Of course, the maturity of these technologies is high and these failures aren’t common but for reasons of completeness, we mention them. For example, the first issue is about wind turbines themselves which consist of many moving and rotating subassemblies installed at a high elevation. This equipment consists blades, rotor, gears and generator which bear more tension and wear during operation compared with conventional generation. In the case that large shares of PV systems are integrated in the power grid, then its robustness is affected by unpredictable failures such as hot spot, diode failure, ethylene-vinyl acetate (EVA) discoloration, glass breakage, delamination with breaks in the ribbons and solder bonds, light induced degradation, low irradiance losses, potential induced degradation, shading effect, soiling effect, sun tracking system misalignments, wiring losses, mismatching effect in solar array, and other failures such as ground faults, line-to-line faults, and arc faults [16]. Impact on Harmonics: Harmonics are always present in power systems to some extent and can be caused by factors such as the nonlinearity in transformer exciting impedance or loads, AC to DC conversion equipment, variable-speed drives, switch mode power equipment, etc. Power electronics equipment such as inverters, combined with RES systems, can also be a source of harmonics to the network. In this case, their contribution to the harmonics currents is in part due to the Silicon Controlled-Rectifier (SCR) type power inverters that produce high levels of harmonic currents. Most inverters are designed with Insulated Gate Bipolar Transistor (IGBT) technology that utilize pulse width modulation to

359

360

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

generate the injected “pure” sinusoidal wave. Although the impacts of RES systems on the power system are not significant, in some instances complications may arise and levels can exceed the IEC 61000-3-6: Assessment of harmonic emission limits for the connection of distorting installations to medium voltage (MV), high voltage (HV), and extra high voltage (EHV) power systems standard or equally the IEEE-519: Standard for Harmonic Control in Electric Power System. These problems are usually caused by resonance with capacitor banks, or problems with equipment that are sensitive to harmonics. In the worst case, inverters may need to be disconnected as a consequence of the extra heating caused by the harmonics. Impact on Short Circuit Levels of the Network: The presence of RES systems in a network affects also the short circuit levels of the network and thus the fault currents when compared to normal conditions at which no RES system is installed. The fault contribution from a single small RES system is not large, but in the case of many small units, or few large units, the short circuit levels can be altered enough to cause a miss in coordination between protective devices, like fuses or relays. The influence of RES systems to faults depends on some factors such as the generating size, its distance from the fault location and the type of the RES system. This could affect the reliability and safety of the distribution system. • Generating size: many small units or a few large RES units can alter the short circuit levels sufficient to cause fuse-breaker miss-coordination. This could affect the reliability and safety of the distribution system. In this case if the fault current is large enough, the fuse may no longer coordinate with the feeder circuit breaker during a fault leading to excessive fuse operations and decreased reliability. • Location: If the RES system is located between the utility substation and the fault, a decrease in fault current from the utility substation may be realized. This decrease needs to be investigated for minimum tripping or coordination problems. Alternatively, if the RES, or total combination of RES units, is resilient compared to the utility substation source, it could have a substantial impact on the fault current emerging from the utility substation. This may cause fail to trip, sequential tripping, or coordination problems. • Type: Although the impact is relatively low, the cycles of the connected inverters can contribute to faults. Even though few cycles are a short time, in some cases the duration may be long enough to impact fuse-breaker coordination and breaker duties.

11.2.2 Benefits of DER on power system grids As already seen in previous section, RES inherent intermittency, impacts the system jeopardizing its security and normal operation. This has been given clear in terms of how RES affects the reliability of the system. The main objective of this section is to

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

highlight how DER can support a system with high RES penetration neutralizing their impact through different operation strategies. The task of maintaining security and reliability of the electrical system falls on the system operators who are responsible for ensuring the required degree of quality and security of supply. The emergence of RES in the liberated electricity markets complicates this task as the provision of ancillary services is regularly required. These services include frequency control services (like regulation, load following, operating reserves), voltage control services (through reactive power support), and emergency services (by black-start services). However, at the same time DER and renewable generation combined with storage can significantly contribute on reactive power reserve and voltage control. The orchestration of DER operation towards the above objective is a task of the operators. To measure the utilization of DER for ancillary services (UAS), a KPI that expresses the ratio between the energy used for ancillary services and the total energy produced is proposed. UAS% 5

EAS  100 Etotal

ð11:9Þ

As already stated, voltage control through reactive power can be provided by renewable sources connected to the power system via solar photovoltaic and battery storage. To assess the impact of DER on voltage control, a KPI that calculates the reactive energy provided (REP) for voltage control by RES and DG at the DSO level networks during a predefined time is also proposed. REP 5

t2  XX  qg ðtÞ

ð11:10Þ

gAG t5t1

where G is the set of RES and DG generators producing reactive power, and qg(t) is the reactive power generated by generator g at a given time t. The main objective is to compare to which extent generators are taking part in maintaining voltage stability. Another necessity of the upcoming distribution network evolution, that can be considered as a DER, is DR. Traditionally, the goal of the DR is to alter the load curve through a variety of programs such as peak clipping, load shifting, valley filling, energy conservation, etc. by providing financial incentives to the electricity end-users. However, DR is little more than a way of financially motivating customers to reduce their energy use when electricity is particularly scarce and expensive or when the wires are overburdened. DR can be a great resiliency tool by working to reduce stress on our electricity grid during times of prolonged disaster or shifts in weather. Critical conditions in which energy demand is higher than anticipated and supply is less than expected due to unpredicted disasters or extreme weather events as well as conditions

361

362

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

where, distributed generation is higher than the demand due to the volatilities of renewable generation, can both lead to black-outs. DR is one approach to increase flexibility during those critical conditions by contracting customers to provide load reductions in rewards of incentive payments. Electricity storage has also great potential to be a game changer in terms of balancing electricity supply and demand if it can be brought forward as a cost-effective proposition. Distributed storage systems can store electricity when there is excess electricity available at lower price and supply electricity at the time of deficit. Therefore, they act like both generation and load at different situations. Additionally, energy storage has the potential to take excess generation such as on a windy or sunny summer day, and store it in multiple places from large pumped hydro stations down to batteries within homes and everything in-between. In this way, electricity storage can improve resilience, as it would remove some of the challenges related to managing intermittency and reduce the requirement for backup generation and use of the network. Even though electricity storage is ready and available today, we do need advances in market conditions to encourage further deployment of storage technologies. As owners of EV fleets are emerging as active nonutility actors, the ability of integrated storage from EVs to enhance grid resilience is becoming more important. Each mode of EV integration (Grid-to-Vehicle, Vehicle-to-Building and Vehicle-to-Grid) comes with a unique set of grid resilience attributes and possibilities, and the need for these grid services will vary across states and regions. The members of the Department of Energy’s Electricity Advisory Committee have identified 10 types of resilience and reliability services related to EV integration, including DR, frequency regulation, emergency backup, capacity firming and voltage control, etc. [17]. The ESS together with other DER can supply a certain portion of the distribution network in the form of a microgrid during planned or unplanned utility supply interruptions, thus improving resilience. This can be accomplished in two ways. First, island-capable microgrids can disconnect from the central grid during major weather events to allow energy to be diverted to critical loads. This enables utilities to utilize available flexibility in restoring generation stations, responding to critical outages, and shutting down systems before a major event to prevent damage. Second, during times of mismatch between supply and demand, island-capable microgrids and web-of-cells can disconnect from the grid and use self-generated energy. This delivers grid-side management of electrical supply and frees centrally generated energy for other loads. In either case, the capability to divert energy away from the microgrid/web of cell and towards other loads provides utilities with more flexibility in generation and distribution of power. The enormous technical potential of microgrids/web-of-cells to make electric service more robust by coping with outages efficiently, lessening their impact and

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

rebounding quickly renders these approaches the most promising option for bolstering the grid resiliency.

11.3 Uncertainties affecting system resilience As already mentioned, the uncertainties in generation can significantly affect the resilience of distribution networks. Resilience is defined as “the ability of a network to withstand and reduce the magnitude and duration of disruptive events, which includes the capacity to anticipate, absorb, adapt and rapidly recover from such an event” [18]. Resilience also applies to the operation of the network during HILP events such as extreme weather events, and cyber-attacks. There are two types of resilience: infrastructure and operational. The former is related mainly to network physical strength and geographic characteristics, and the latter to the ability of the system to supply the consumers [19]. On the other hand, reliability defines how the system operates during low-impact high-probability events (LIHP) and the key features that distinguish resilience and reliability are listed in Table 11.1 [20]. While the penetration of intermittent weather-depended energy sources increases, the effect of unpredicted extreme weather variability in the system’s operational resilience increases dramatically. Unexpected heat or cold streams can create a significant imbalance between generation and demand and can have a significant effect on the system frequency. At the same time, as the inertia of the system is reduced with the decommissioning of large centralized conventional power plants, the rate of change of frequency (ROCOF) of the network is becoming more vulnerable. Furthermore, most of the installed capacity of RES are connected to distribution networks, where redundancy is limited, and a single outage event can result in a loss of electricity production. Also, the digitalization of the power systems using other critical infrastructures such as telecommunications, enhances the vulnerability of the network to cyber-attacks. Therefore, transition to decarbonization and digitalization of distribution systems, through “greener” variable generation and the increased utilization of ICTsdepending on the circumstancesleave a significant number of consumers without electricity even when the infrastructure remains fully functional. Consequently, assessing and enhancing network resilience is essential for optimal operation of distribution networks before and during extreme events.

11.3.1 Metrics for assessing distribution system resilience The concept of power system’s resilience is relatively new and complex, and thus metrics for assessing network resilience during HILP events are not readily established. Currently, there are visual, mathematical and economical methods for estimating distribution system resiliency. This section looks at how the resilience is affected by the

363

364

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 11.1 Comparison between reliability and resilience. Reliability

Resilience

High-probability, low-impact Based on average indicators Short-term, typically static Evaluates power system state Concerned mainly with customer interruption time

Low-probability, high-impact Based on risk profile Longer term, adaptive, ongoing Evaluates power system states and transitions Concerned with customer interruption time and infrastructure recovery time

(Source: Reproduced from P. Maria Luisa, D. Michele, E. All, Resilience of distribution grids, CIRED Working Group, 2018).

intermittent nature of the high RES grids and how the DER operating strategies can support resilience by unlocking existing flexibility. 11.3.1.1 Signs of vulnerability Measuring resilience is not a straightforward process; therefore, it is much easier to measure system performance during LIHP events in order to identify the vulnerability of the system. In the case where the system performance during such events is poor, during HILP the operation of the system will be further reduced, jeopardizing overall system stability. It should be noted that if the system performance is high during LIHP events, it does not necessarily mean that the system is resilient. The performance during LIHP events of networks can be estimated using the average reliability indicators System Average Frequency Index (SAIFI) and System Average Interruption Duration Index (SAIDI) presented in Eqs. 11.11 and 11.12 respectively [21]. P Total number of customers interrupted ð1Þ SAIFI 5 ð11:11Þ Total number of customers served P ð2Þ SAIDI 5

Customer minutes of interruption Total number of customers served

ð11:12Þ

A more resilient oriented index can be evaluated by calculating SAIDI only for major event days (MED). When the daily SAIDI exceeds the T MED threshold value then that day is categorized as MED. T MED can be calculated using Eq. 11.13 [21]. For this index, SAIDI is evaluated only for the days where the event impact was significant. ð3Þ T MED 5 eðα12:5βÞ where α is the log-average of all daily SAIDI values β is the log-standard deviation of all daily SAIDI values

ð11:13Þ

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

11.3.1.2 Total restoration cost Higher restoration cost (TR) indicates a lower resilient grid. Therefore, calculation of the monetary cost of an event, including labor cost (LC), infrastructure replacement cost (RC) and ancillary (or additional) generation cost (GC), can be considered as an index of resilience [22]. If this index is employed during planning or preevent stages, it can identify network areas where resilience enhancement methods can significantly reduce monetary cost for the system operator. We note that for positive net present value for the infrastructure enhancement the possibility of a disruptive event also has to be taken into consideration. X ð4Þ TR 5 LC i 1 RC i 1 GC i ð11:14Þ i51

11.3.2 Resilience trapezoid The resilience trapezoid can be used as a visualization tool to reflect the performance of the network during any event. It presents the different states of the network, predisturbance, postdisturbance, its restorative state and the postrestoration state. It can be considered as a more comprehensive version of the resilience triangle, with the main difference manifested in the fact that in the latter, restoration time is considered zero [23]. This tool, though, does not reflect the DER impact on addressing such events under a grid with high RES penetration. RES uncertainties will vastly impact the future grid especially under major events and that is why the DER impact should show in such a vizualization tool. The authors propose in the next section how this tool can be enhanced. For operational resilience, the resilience indicator (R) can be considered as the percentage of the load demand satisfied (or system frequency f). Therefore, at the predisturbance state, it is expected that the electricity demand will be satisfied by the available generation capacity. • During the disturbance progress (Phase I), resilience drops to Rpdo which reflects the amount of demand being satisfied (after load shedding) over the initial demand. Rpdo 5

P initial 2 P shed P initial

ð11:15Þ

• During Phase II, the system remains stable until tor, • when Phase III (restorative state) begins. It should be noted that postrestoration state can be equal or lower than the initial resilient state but for a disruptive event, due to operational reasons, it is expected that eventually the system will return to its initial state. In addition, it should be noted that the restoration time for operational resilience is lower compared to restoration time of

365

366

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

the infrastructure resilience, as operational measures can be performed remotely, while infrastructure restoration will probably require local and time-consuming interventions as stated in [24] (Fig. 11.5).

11.3.3 ΦΛEΠ Resilience quantitative framework The ΦΛEΠ framework captures the different states presented in the resilience trapezoid. This metric evaluates system resilience by considering how fast resilience drops (Φ), the lower value of resilience Rp (Λ), the duration of postdisturbance degraded state (E) and how fast the system returns to its initial state (Π). Table 11.2 presents the individual equations and measuring units for each individual metric of the ΦΛEΠ quantitative framework. The overall system operational resilience during an event can be simply evaluated by the area of the trapezoid (Eq. 11.16). A lower value of ΦΛEΠ indicates a resilient network since the grid is less affected by the event [23]. ΦΛEΠoperational 5

ð tor

Ri ðt Þ 2 R0 ðt Þdt

ð11:16Þ

t oe

As seen previously, this metric does not mirror the DER operating strategies when evaluating the resilience of the system.

11.3.4 Flexibility and resilience matrix The aforementioned metrics do not represent how flexibility resources can enhance the network resilience during the event, as they capture only the overall performance of the system. Knowing how the contribution of the available flexibility resources would neutralize RES uncertainties is very useful for the system operators during day-ahead scheduling and real-time operation. For instance, if the system operators know that a specific amount of flexibility can alleviate or reduce the effects of a HILP event, they will have the ability to deploy the necessary flexible resources to prevent the event consequences or reduce the impact of the event with

Figure 11.5 Resilience trapezoid.

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

Table 11.2 ΦΛEΠ Resilience metrics equations [24]. Metric

Equation

Unit

Φ

Rpdo 2 R0o tee 2 toe

MW/hours

Λ E Π

R0o 2 Rpdo tor 2 tee

R0o 2 Rpdo Tor 2 tor

MW Hours MW/hours

real-time operations. On the contrary, if the network resources (including flexibility resources) are considered adequate, frequency restoration and conservation reserves can be reduced, resulting in an economically optimal unit commitment and economic dispatch. Therefore, an adaptation of the ΦΛEΠ operational metric has been defined that compares the operational performance of the grid during the event with and without the available flexibility resources. The flexible resources that are considered within the new metric are DR, ESS and EVs. These resources can be controlled and utilized in short-term scheduling as preventive measurements or during the event as restorative operations. The new resilience metric (R) at each point in time can be evaluated using Eq. 11.17. Rt n 5

P L 2 P loss 2 P implDR    ðP G 2 max ðP Gloss ; P inf Þ 1 P spin 1 P explDR 1 P ESS 1 P EV Þ

ð11:17Þ

where, P L System load P loss Load Loss P implDR Implicit DR P G Central or distributed generation P Gloss Constrained generation P inf Infrastructure capacity constrained P spin Spinning reserve P explDR Explicit DR P EV Electric vehicles In order to evaluate the system resilience for the whole event duration, the boundaries of the network under investigation have to be initially defined. These boundaries can include the whole transmission substation or a single distribution feeder. Additionally, a value for every time slot after the event for generation, load, spinning reserves and flexibility resources must be measured. During an event, the lowest value between the available generation connected to the loads (within the boundaries), and the physical network capacity should be considered in the evaluation.

367

368

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

This new metric can take values from 0 to 1, where 1 indicates an ideal fully resilient grid that can satisfy the total demand of the connected consumers under the influence of the extreme event. R tn 5

PLðtÞ ðPGðtÞ 2 maxjðPGloss ; Pinf Þj 1 PspinðtÞ 1 Pflex

demandðtÞ Þ

ð11:18Þ

where, PLðt Þ 5 PL 2 Ploss 2 PimplDR Pflexdemand 5 PexplDR 1 PESS 1 PEV The resilience trapezoid can also be adapted in order to graphically represent how the system resilience is affected by the utilization of the available flexible resources. Initially the metric R is evaluated using Eq. 11.18 without the flexible resources enabled, thus Pflex demandðtÞ is set to zero. Afterwards, R is revaluated with the flexible resources enabled and both characteristics are plotted on the same graph. In this manner, the quantified flexibility that enhances the system performance during the event can be evaluated by calculating the in-between area of the two characteristics.

11.3.5 Increasing resilience of a high RES system with flexible resources When the weak points of the network have been identified, preventive measures should be taken to make the system less vulnerable to HILP events. Traditionally, in critical infrastructures like transmission systems, systems were designed to be more robust and by design equipped with redundant circuits. Redundancy can be increased by commissioning additional equipment in parallel with the minimum equipment to meet demand. In distribution grids, redundancy is not an economical option, as the lengths of overhead lines and underground cables are tremendous. Reinforcement and hardening, which are infrastructure actions that can make the grid less vulnerable, are very effective during major catastrophic events, thus smarter, more flexible operational solutions are more affordable options for optimal performance of future distribution grids (Fig. 11.6).

11.3.6 Operational measurements Several operational measures can boost distribution system resilience. Initially, improving load and generation forecasting methods can increase resilience, as they will significantly reduce uncertainties. If sudden and significant weather variations are accurately predicted, system operators can deploy preventive measures to maintain system stability. These measures usually include synchronization of flexible gas turbines or the activation of Pumped Hydropower plants. On the other hand, system balance can be achieved by demand management. Therefore, DR strategies and respective remuneration schemes have to be developed. At the same time, as advanced metering infrastructures are being

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

Effecveness

High

Hardening reinforcement measures

Hybrid

Smart operaonal measures

Low Less affordable

Cost

More affordable

Figure 11.6 Cost and effectiveness of resilience enhancements measures [20]. Source: Reproduced by authors.

massively deployed, smart load shedding can be performed rather than the common under-frequency schemes. In this manner when load shedding is unavoidable, critical loads like hospitals and telecommunication infrastructures can maintain supply during the event and most importantly, distributed generation and storage can be smartly excluded from mass load shedding. DER can have a beneficial effect on system resilience due to their ability to control, generate and store energy. ESS can export energy when available generation is constrained, and store excess energy during low load conditions. In this manner, fluctuations and balancing deviations during events can be mitigated. Similarly, controllable loads, like EVs can provide balancing services when needed. Distributed energy sources, conventional and renewables, if controlled correctly, can under certain circumstances also increase system performance during HILP. One main drawback of RES penetration is that during critical events, when ROCOF exceeds a predefined threshold, antiislanding protection schemes are activated and disconnect them from the main grid. Consequently, generation from RES becomes unavailable and cannot contribute in tackling the critical event. In order to avoid this phenomenon, the ROCOF threshold value has to be defined as high as possible, maintaining RES connected during critical under-frequency events, while at the same time ensuring that RES will be disconnected during an unintended island operation. Of course, this cannot handle the uncertain nature of RES that needs to be supported by DER. Since microgrids contain DER and can operate either connected or disconnected to the main grid, they intrinsically enhance grid resilience. Microgrids can be inherently developed or can be formed dynamically during events by system operators. For optimal performance of the microgrids, proper control strategies have to be created and defined. Local controllers should be connected to the DER of the network and headed by a central controller installed at the system boundary [25]. For optimal performance of

369

370

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

microgrids during the events, the expected event duration has to be considered. In cases where the event duration is expected to last long, load reduction techniques can be applied in order to ensure that the grid would be able to supply energy until normal conditions are restored. Intentional islanding methods must be well developed and verify that islanded areas have sufficient generation for critical identified loads.

11.4 Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER The challenges when trying to analyze a power grid in the presence of dynamic/varying power production and distributed power generation are quite complex and multidimensional. The cost of building a real system in an attempt to test and evaluate new technologies can be extremely high and prohibitive. Therefore, the utilization of simulation software/tools is inevitable and unavoidable. In such a case, the accuracy of simulation models is always crucial in producing appropriate outcomes that can be used even as a driver for policy and development decisions. Another important factor is the adoption of the right controllers and control schemes which can be used to achieve operational targets and solve expected problems. In this specific section, a methodology for analyzing the grid under performance indexes of both reliability and operational resilience of the system is presented. The DER control strategies that are employed are presented along with indicative results of the system analysis for different operational scenarios.

11.4.1 Methodology For the study of resilience and reliability towards flexibility and thus DER supporting RES inherent uncertainties, it was firstly required to define a representative power grid based on real case characteristics. For this reason, a power grid was selected composed of five different areas. Each area is defined to have specific power characteristics based on load and power generation relevant to Cyprus. More information on the selected system is provided in Section 11.4.2. In support of the operational needs of operators, specific controllers are developed, which are responsible for containing frequency and voltage variations using appropriate attributes of the interconnected system. Details about the controller’s structure and their functionality are presented in Section 11.4.3. Subsequently, the scenarios under investigation are defined and presented in Section 11.4.4. After performing the investigations, the generated simulation results are analyzed, using graphical representations to help the process and these are presented in Section 11.4.5. The general outcomes of the work are summarized in Section 11.4.6. The adapted process is represented via a flowchart shown in Fig. 11.7.

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

Figure 11.7 Flowchart for the simulation/analysis process.

Time-series data are used to facilitate real system analysis and these are suitably linked to the simulation case(s). Multiple simulations are conducted for each time step and for each simulation scenario/fault. Both a load flow and an RMS simulation are needed to be performed for each individual case study to generate system performance results. The load flow is needed to define the initial data for the RMS simulation. During the RMS simulation, a fault is introduced and the time-series data produced by the simulation are noted. All the simulations are performed with and without controllers to evaluate frequency regulation and the simulation results are related to the key performance indices (KPIs) shown in Table 11.3. The proposed KPIs are considered a valuable tool for evaluating frequency regulation studies with meaningful outcomes.

11.4.2 Grid system under investigation The power grid system formulated and used in the simulation studies is a suitably selected section of the Cyprus transmission and distribution grid extrapolated to the adapted scenario expected to be in operation by 2030 [26]. In more detail, the power grid model represents a synthetic benchmark grid which comprises of transmission substations with terminals of 132 kV voltage level and distribution substations operating at voltage levels of 11 and 400 V phase to phase. For completeness and appropriate use of the results achieved, the details of the interconnected system used for building the required model for the conducted analysis

371

372

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 11.3 Key performance indices (KPIs). Key performance index

ID

Name

Formula

1

Frestoerd 2 Fnom # ε; ε-0

2

Frequency restoration control effectivity Frequency nadir

3

Frequency zenith

4

Rate of change of frequency (RoCoF)

5 6 7 8

Indication of stability Voltage RMS Resilience Utilization of DER

maxðfn 2 f Þ ½Hz fn 2 nominal frequency ½Hz f 2 system frequency ½Hz minðfn 2 f Þ ½Hz fn 2 nominal frequency ½Hz f 2 system frequency ½Hz Pg 2 Pl df dt 5 2Hsys 2 rate of change of frequency ½Hz=s Pg 2 generators0 active power ½pu Pl 2 demand active power ½pu Hsys 2 system inertia ½s Boolean variable (YES/NO) Vn 2 Voltage of Bus n ðV Þ Eq. (11.17) Eq. (11.9) df dt

are as follows: 1721 lines, 3009 busbars (4891 terminals), 1006 transformers, 1925 loads (including 962 electric vehicle loads), and 1931 generators (962 PV systems, two synchronous machines, 962 battery ESSs, two hydro systems, two wind farm systems and one biomass unit). Furthermore, it provides 2284 protection devices (993 fuses) and 1291 breakers/switches. In general, the Cyprus grid model includes the distribution network of three transmission substations of Cyprus: ALAMBRA-Area1, PROTARAS-Area 2 and DISTRICT OFFICE-Area 3. Each transmission substation constitutes a different control area with distributed generation, storage, EVs, and loads. Fig. 11.8 shows a part of the ALAMBRA distribution network as modeled in DIgSilENT PowerFactory which includes the transmission substation (marked with S symbol) and a distribution substation. The model of the distribution substation is shown in Fig. 11.9. The distribution system of Cyprus has a radial architecture and the transmission system operates in a ring / loop architecture. More details about the model are as follows: • The distribution substation model is composed of a distribution transformer and aggregated elements for load, EVs, photovoltaic systems, and storage units. • The main grid model also incorporates a conventional power generation and a wind power generation area shown in Figs. 11.10 and 11.11 respectively.

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

Figure 11.8 Transmission substation and distribution substation models for ALAMBRA distribution network.

Figure 11.9 Composition of distribution substation model.

373

374

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 11.10 Model of wind power generation area.

• •

The wind power generation area is composed of two transmission transformers and two aggregated areas of wind turbines. The conventional power generation area includes two transmission transformers and two synchronous machines.

11.4.3 DER operational strategies Due to the increased utilization of RES in the power grid and the application of smart load and storage strategies, major operational challenges are expected to arise through such system synthesis. In such a situation, the frequency and voltage regulation via traditional methods is not going to be sufficient. A distributed approach for voltage/frequency regulation is a more feasible and a more appropriate solution for the future power grid. Advanced real-time monitoring of the power system status is required to locate, predict and handle the operational issues. Another important core element of the future power grid is the local controller attached on DER which can be responsible for the grid support needs. Still, a

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

Figure 11.11 Model of conventional power generation area.

transition to local smart controllers is not yet implemented in the real systems and a lot needs to be done towards this direction. The controller adopted in this chapter aims to solve the power grid frequency instability issues using the distributed resources present in the grid [27,28]. Specifically, the adopted controllers are attached on storage systems in each power grid area and have the responsibility to solve frequency issues in a local way. In Fig. 11.12, the operational strategy is shown with the main functions highlighted. The main local controller functions are: • Event location • fFRCfast frequency restoration control • Predefined power-frequency curve calculation • Local resources control The diagram of Fig. 11.13 is implemented into the control circuit shown in Fig. 11.13 which is responsible for controlling the available storage sources of the area that the controller itself is monitoring. During the occurrence of a power event, the controller sends the appropriate control signal to the storage elements of its area to achieve effective frequency regulation. The core functions and their interconnections are also shown in Fig. 11.13. Specifically, the following core functions are collaborating to achieve the desired local control of storage system in the way described:

375

376

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 11.12 Control logic scheme.

Figure 11.13 Functions and interconnections of area controller [29].

•

The “SecControl” and “AGCControl” functions that represent respectively a secondary control and the traditional Automatic Gain Control. For the purposes of the analysis presented in this chapter, these functions have been deactivated but their presence gives more flexibility to the controller for other applications.

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

• The “dPBorder” function can also be active and it evaluates the total tie-lines active power variation which is used as an input to “fFRC” and “Event Location” functions. • The “Event Location” function locates the possible instability event. The instability event is external to the area if the signs are concordant, internal to it in case of discordant signs. In case of internal instability, a trigger signal activates both the fFRC function and power-frequency pre-defined response of the resources at local resources level. • The fFRC, in particular, deals with the total tie-lines active power variation with the aim to counteract it using the assets in the own control area while the powerfrequency curve predefined is calculated for power resources at control area level considering the available active power flexibility.

11.4.4 Scenario under study In the investigated scenario, it is assumed that strategies for climate mitigation are adopted lowering the fossil fuel consumption worldwide by the year 2030. Therefore, due to the low dependence on conventional fuels, fuel costs are relatively low. Similarly, it is considered that storage solutions are going to be utilized in a wide scale. Decentralized storage solutions are also considered of adequate size to help in deploying higher penetrations of RES throughout the electricity grid. The nominal capacity per power source for the scenario under investigation is shown in Table 11.4. Four power events were investigated that simulate realistic conditions and challenges of the grid operator shown in Table 11.5.

11.4.5 Simulation case results A significant loss of load/power generation can cause high frequency variation and/or unstable conditions. The specific phenomenon is expected to be more intense in power systems with high penetration of RES which have low inertia. Based on the above, the stability and the frequency levels will need to be controlled within the desirable and expected range by utilizing DER and emerging technologies that provide flexibility and increase the security and resilience of the system. The use of smart controllers is unavoidable and can help significantly in restoring frequency within Table 11.4 Scenario case for 2030nominal power production per type. Scenarionominal capacity per power source type (MVA) Scenario cases

Solar

Wind

Hydro

Biomass

Conventional

Pump storage

2030

42.3

42.3

48.9

9.2

22.5

—

377

378

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Table 11.5 Power events under investigation for 2030 scenario. Stability analysis scenarios for 2030loss of power generation—nominal capacity (affected source type marked with bold color) Source type

Fault in Area 1

Fault in Area 2

Fault in wind station

Fault in gen. station

Solar Wind Hydro Biomass Conventional

42.3 42.3 34.2 6.4 15.8

42.3 42.3 34.2 6.4 15.8

42.3 42.3 34.2 6.4 15.8

42.3 42.3 34.2 6.4 15.8

Figure 11.14 Frequency vs time having the controllers deactivated/activated fault in area no. 1.

desirable range as can be seen in Fig. 11.15. Without the use of smart controllers, the frequency can reach quite low levels in a quite short time and eventually force the system to become unstable (Fig. 11.14). The abbreviations shown in next figures refer to the following case studies: • A1—CD: Area No 1—controllers deactivated • A1—CD: Area No 1—controllers activated • A2—CD: Area No 2—controllers deactivated

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

• • • • •

A2—CD: Area No 2—controllers activated PSA—CD: power station area—controllers deactivated PSA—CD: power station area—controllers activated WSA—CD: wind station area—controllers deactivated WSA—CD: wind station area—controllers activated The simulation results shown in Figs. 11.1511.16 depict that the local control approach manages to minimize the frequency variation quite effectively. Boxplot In descriptive statistics a box plot or boxplot is a method for graphically depicting groups of numerical data through their quartiles. Box plots may also have lines extending from the boxes (whiskers) indicating variability outside the upper and lower quartiles, hence the terms box-and-whisker plot. Outliers may be plotted as individual points. Box plots are nonparametric: they display variation in samples of a statistical population without making any assumptions of the underlying statistical distribution. The spacings between the different parts of the box indicate the degree of dispersion (spread) and skewness in the data, and show outliers. Hence, a boxplot is a standardized way of displaying the dataset based on a five number summary: the minimum, the maximum, the sample median, and the first and third quartiles. Where two sets of results are analysed / depicted the extended range for the second lot is depicted in red. In the case of this chapter, the extended range depicts the results of the system without controllers.

Figure 11.15 Frequency variation during the fault events with and without the controllers enabled (box plot format)—all study cases.

379

380

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 11.16 RoCoF during the fault events with and without the controllers enabled (box plot ormat)—all study cases.

Additionally, the frequency nadir shown in Fig. 11.17 is reduced adequately by using local controllers. Furthermore, in Fig. 11.18, it can be observed that by enabling the controllers the resilience of the whole system improves in all cases. This means that the uncertainties of the RES are highly alleviated via specific DER operation strategies. In Fig. 11.19, it is justified that the controllers utilized energy from the storage systems in order to cope with the power events under consideration. Depending on the loss of power generation the controllers allocate different amount of power/energy in an attempt to improve the frequency levels. Also, the power events can cause severe voltage variation as can be seen in Fig. 11.20, which can be reduced via regulating frequency variation within smaller range close to nominal. From the results of the worst-case scenario, it is seen that the controllers are quite effective in controlling frequency according to the target range (Fig. 11.21). In general, the overall frequency stability is improved by enabling the controllers for frequency regulation as depicted in Fig. 11.22. Fig. 11.23 shows the matrix introduced by the authors in order to capture the correlation between the resilience of a high intermittent grid (RES dominant) and the flexibility delivered by DER operation strategies. The results, captured in the figure, reveal that the distributed nature of resources in the emerging grid of 2030 and

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

Figure 11.17 Frequency nadir calculated with and without the controllers (box plot format)—all study cases.

Figure 11.18 Resilience index variation during the fault events with and without the controllers enabled (box plot format)—all study cases.

381

382

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 11.19 Utilization of DER index variation during the fault events with and without the controllers enabled (box plot format)—all study cases.

Figure 11.20 Voltage variation during the fault events with and without the controllers enabled (box plot format)—all study cases.

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

Figure 11.21 Control effectiveness index with and without the controllers enabled (dark red is the combination of the two cases: with and without).

Figure 11.22 Stability index calculated with and without the controllers enabled (dark red is the combination of the two cases: with and without).

383

384

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 11.23 Resilience/flexibility matrix.

beyond, act positively towards the resilience of the system, since energy resources are physically distributed where energy is consumed. Moreover, in the cases where no controllers are activated, the linear relation between resilience and flexibility has been verified as anticipated. The more flexibility the system can deliver, for example, through storage, the more resilient the grid becomes as the RES uncertainty can be handled more effectively. In the case where the controllers are activated, the resilience approaches unity for most of the cases. This does not imply by any means that the resilience is noncorrelated with the flexibility. It actually highlights the controllers’ quality to activate the appropriate flexibility measures, that is, storage flexibility, power curtailment, load shifting, etc. to improve the resilience of the grid by appropriately utilizing to the maximum, all available resources. An important conclusion that can be reached through the recorded results, is the mere fact that distributed energy systems in situ where the load is connected, make the connected grid more resilient. In addition, through the use of appropriate distributed controllers, this can further enhance the resilience of the interconnected grid since it captures the available flexibilities coming from connected storage systems, EVs, Demand Side controllable loads and the advance features of the controllers of RES.

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

11.5 Discussion and conclusions Within this chapter, the impact of the uncertainties of RES in the day-ahead operation of the system is discussed. They are categorized under the resilience and reliability aspect of the system and performances indexes are summarized. Another important attribute that is highlighted in this chapter is the development of a new metric that interrelates resilience with flexibility that can be provided by DER. Under this prism, a thorough analysis of the real distribution grid of Cyprus was performed and DER operating strategies that support the system are presented. They are presented using the performance indexes of the previous sections under real challenging scenarios. The power system under test is evaluated using the developed hierarchical controllers under different fault events revealing acceptable results: • Frequency variation results are depicting that the frequency controllers succeed to minimize the frequency reduction adequately. • Justification/verification that the controllers are quite effective in controlling frequency within the acceptable range. • The overall stability is improved when applying hierarchical controllers for frequency regulation. • From the reliability metrics, it is verified that controllers can be effective in improving the reliability of the power system. • From the resilience metrics it is seen that by adopting a distributed control approach utilizing directly the distributed flexibility resources such as storage systems, it is possible to make the power system more resilient and less susceptible to power events and RES uncertainties. • Flexibility of the system is highly dependent on the storage operating strategies that are complimented by other flexibility resources such as flexible demand, EVs etc. It is shown that storage will play an important stabilizing resource for future power grids where renewable energy resources will be the dominant type of power source. • The resulting high resilient grid with controllers enabled, highlights the importance of developing effective operating strategies using the advanced features of dispersed generation in handling the operational needs of the active interconnected grid for effectively containing system interruptions and faults.

References [1] P. Denholm, J. Novacheck, J. Jorgenson, M.O. Connell, P. Denholm, J. Novacheck, et al., Impact of flexibility options on grid economic carrying capacity of solar and wind: three case studies, Nrel (2016) 1125.

385

386

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[2] A. Hussain, V.H. Bui, H.M. Kim, Resilience-oriented optimal operation of networked hybrid microgrids, IEEE Trans. Smart Grid 10 (1) (2019) 204215. [3] Q. Zhou, M. Shahidehpour, A. Alabdulwahab, A. Abusorrah, Flexible division and unification control strategies for resilience enhancement in networked microgrids, IEEE Trans. Power Syst. 35 (1) (2020) 474486. [4] F.Z. Peng, J. Wang, Flexible transmission and resilient distribution systems enabled by power electronics, Power Electron. Renew. Energy Syst. Smart Grid (2019) 271314. [5] S. Wang, P. Dehghanian, M. Alhazmi, M. Nazemi, Advanced control solutions for enhanced resilience of modern power-electronic-interfaced distribution systems, J. Mod. Power Syst. Clean. Energy 7 (4) (2019) 716730. [6] European Technology Platform SmartGrids, The need for a fundamental review of electricity networks reliability standards  ETP SG Security And Resilience Task Force’s white paper (2016). [7] S.H. Madaeni, R. Sioshansi, P. Denholm, Comparison of Capacity Value Methods for Photovoltaics in the Western United States,Technical Report NREL/TP-6A20-54704. July 2012, Comparison Capacity Value Methods Photovolt. West. US (2012) 138. [8] U.S. Department of, Energy,Office of Energy Policy and Systems Analysis, Maintaining Reliability in the Modern Power System 139 (2016). Available from: https://www.energy.gov/sites/prod/ files/2017/01/f34/Maintaining%20Reliability%20in%20the%20Modern%20Power%20System.pdf. [9] P. Denholm, M. O’Connell, G. Brinkman, J. Jorgenson, Overgeneration from solar energy in California: a field guide to the Duck Chart, National Renewable Energy Lab.(NREL), Golden, CO (United States), 2015. 309, 2015. [10] INTERPLAN consortium, D3.1:INTERPLAN use cases,2018, (https://interplan-project.eu/ resources/). [11] L. Xie, P.M.S. Carvalho, L.A.F.M. Ferreira, J. Liu, B.H. Krogh, N. Popli, et al., Wind Integration in power systems: operational challenges and possible solutions, Proc. IEEE 99 (1) (2010) 214232. [12] European Network of Transmission System Operators for Elecricity, Rate of change of frequency withstand capability, 2012. [13] DNV GL Energy AdvisoryEirGrid, RoCoF alternative solutions technology assessment: high level assessment of frequency measurement and FFR type technologies and the relation with the present status for the reliable detection of high RoCoF events in a adequate time frame, 2015. [14] J. Eto, J. Undrill, P. Mackin, R. Daschmans, B. Williams, B. Haney, et al., Use of frequency response metrics to assess the planning and operating requirements for reliable integration of variable renewable generation, December 2010, p. LBNL4142E, 2010. [15] H. Markiewicz, A. Klajn, Voltage disturbancesstandard EN 50160, voltage characteristics in public distribution systems, Power Qual. Appl. Guid., vol. 5.4.2, pp. 411, 2004. [16] I. Lillo-Bravo, P. González-Martínez, M. Larrañeta, J. Guasumba-Codena, Impact of energy losses due to failures on photovoltaic plant energy balance, Energies 11 (2) (2018). [17] U.S. Department of Energy, Enhancing grid resilience with integrated storage from electric vehicles, 2018. [18] A. Stankovic, The definition and quantification of resilience, IEEE PES. Ind. Tech. Support. Task. Force (2018) 14. [19] M. Panteli, P. Mancarella, The grid: stronger, bigger, smarter?: presenting a conceptual framework of power system resilience, IEEE Power Energy Mag. 13 (3) (2015) 5866. [20] P. Maria Luisa, D. Michele, E. All, Resilience of distribution grids, CIRED Working Group, Technical Report, 2018,1107. [21] D. Committee, I. Power, E. Society, Distribution reliability indices, 2012. [22] R.K. Mathew, S. Ashok, S. Kumaravel, Resilience assessment of electric power systems: a scoping study, in: 2016 IEEE Students’ Conference on Electrical and Electronics and Computer Science, SCEECS 2016, Bhopal, India, 2016. [23] M. Panteli, et al., Power systems resilience assessment: hardening and smart operational enhancement strategies, vol. 105, no. 7, 2017. [24] M. Panteli, P. Mancarella, S. Member, D.N. Trakas, S. Member, E. Kyriakides, et al., Metr. Quantification Operational Infrastruct. Resil. Power Syst. 32 (6) (2017) 47324742.

Assessing renewables uncertainties in the short-term (day-ahead) scheduling of DER

[25] G. Strbac, N. Hatziargyriou, C. Moreiara, et al., Microgrids enhancing the resilience of the european megagrid, IEEE Power Energy Mag 53 (9) (2015). [26] Entso-e (The European network for transmission system operators’ electricity), Ten Year Development Plan (TYNDP), public report, 2018. [27] R. Ciavarella, G. Gradit, M. Valenti, T.I. Strasser, Innovative frequency controls for intelligent power systems, in: 2018 International Symposium on Power Electronics, Electrical Drives, Automation and Motion (SPEEDAM), Amalfi, 2018, pp. 656660. [28] ELECTRA IRP Project, Simulations based evaluation of the ELECTRA WoC solutions for voltage and balance controlStand-alone use case simulation results, WP 6, Control schemes for the use of flexibility. ´ [29] M. Patsalides, C.N. Papadimitriou, V. Efthymiou, R. Ciavarella, M. Di Somma, A. Wakszynska, et al., Frequency stability evaluation in low inertia systems utilizing smart hierarchical controllers, Energies 13 (2020) 3506.

387

CHAPTER 12

Load forecasting in the short-term scheduling of DERs Jiajia Yang1, Fengji Luo2, Weicong Kong1,3 and Zhao Yang Dong1 1

School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney, NSW, Australia School of Civil Engineering, University of Sydney, Sydney, NSW, Australia ALDI Stores, Sydney, NSW, Australia

2 3

Abbreviations DER AR MA ARIMA AI ANN SVM AMI SGSC ELM MAPE RNN LSTM QR NILM TE GWAC O&M PV MCP APM PTDF P2P ADMM DSM ESS EMS VPP BESS

distributed energy resource auto regressive moving average auto regressive integrated moving average artificial intelligence artificial neural networks support vector machine Advanced Meter Infrastructure Smart-Grid Smart-City Extreme Learning Machine mean absolute percentage error recurrent neural network long short-term memory Quantile Regression NonIntrusive Load Monitoring trans-active energy Grid Wise Architecture Council operation and maintenance photovoltaic market clearing price Average Pricing Market power transfer distribution factor peer-to-peer Alternating Direction Method of Multipliers demand side management energy storage system energy management system virtual power plant battery energy storage system

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00003-0

r 2021 Elsevier Inc. All rights reserved.

389

390

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

EV TCA NTCA CVPP TVPP

electric vehicle thermostatically controlled appliance nonthermostatically controlled appliance commercial virtual power plant technical virtual power plant

Nomenclature Indices and sets

index of time, indicating the tth time step index of consumers in, i A N index of producers, j A M index of distribution network branches, l AL set of consumers / produces in the market

t i j l N/M

Parameters L rbi / rsj pbi / psj Pmaxl r dl,j xt yt ; y^ t τ

number of branches in the distribution network bidding / offering prices of the ith consumer / jth producer bidding load / output of the ith consumer / jth producer power transfer limit of branch l market clearing price in the distribution market power transfer distribution factor corresponding to branch l and node j it represents a k-dimensional vector of real values at the tth time step, xt Aℛk . the actual load value and the predicted load value at the tth time step the desired quantile τ

Variables

pcbi pcsj st ht gt it ft ot } σ ϕ lτ,t(•)

dispatched load demand of consumer i dispatched generation output of producer j memory cell state at the tth time step of the LSTM model copy of ot of the LSTM model vector input node at the tth time step of the LSTM model vector of the input gate at the tth time step of the LSTM model vector of the forget gate at the tth time step of the LSTM model output vector derived from the output gate at the tth time step of the LSTM model an element-wise multiplication the sigmoid activation function the tanh function pinball loss at the tth time step

12.1 Introduction Human society is experiencing a changing environment, where energy shortage and climate crisis have been identified as two major challenges to the sustainable development of human civilization. On one hand, the ever-increasing energy demand stresses the energy systems that heavily rely on fossil fuels in tradition. On the other hand, the excessive use of primary energy resulted from human activities leads to large green

Load forecasting in the short-term scheduling of DERs

house gas emissions, which deteriorate the climate change. As an evidence, statistical data shows that the global average atmospheric carbon dioxide in 2018 was 407.4 parts per million; the CO2 levels today are higher than at any point in at least the past 800,000 years [1]. Renewable energy technology is identified as a promising countermeasure to the above global challenges. Currently a large number of renewable energy resources are deployed in distribution network side (e.g., rooftop solar panels and wind turbines) to provide local energy production support to end energy consumers. For example, the Australian Electricity Network Transformation Roadmap estimates that by 2050, distributed renewable energy resources may contribute up to 45% of the nation’s electricity generation capacity [2]. Increasing prevalence of Distributed Energy Resources (DERs) make traditional energy consumers become energy ‘prosumers (producers-and-consumers)’, referring to the fact that they are capable of both generating and consuming energy. Deploying DERs at the edge of the grid are beneficial to both end energy customer and the utility grid in terms of enhancing local energy supply support, alleviating network congestion deferring network upgrade, and reducing the grid’s reserve requirement and operation cost. In spite of these benefits, the stochastic and intermittent natures of renewable energy also pose additional complexities to the grid’s operation and control. A direct impact of DER integration is that it introduces inevitable uncertainties into the net-load profile of end energy customers; this subsequentially imposes significant challenges to accurate load forecasting. As a result, performing upper-level energy management tasks becomes a nontrivial issue. Furthermore, traditional load forecasting techniques usually focus on regional-scale forecasting. That is, they predict the time series of regional load at substation- or bus- level. Integration of DERs implies that new load forecasting techniques at individual customer level are necessary, which would lay foundation to precisely understand individual customers’ energy production and consumption patterns for the development of local energy management applications. In summary, developing effective load forecasting and energy management techniques to manage DERs would be meaningful for enhancing distribution network’s efficiency and reliability, improving the value of distributed energy assets, and promoting the sustainable development of modern cities. This chapter provides a review for some recent advances in load forecasting and the associated DER scheduling techniques in different operational environments. We firstly introduce emerging forecasting techniques for individual energy customers. Distinguished from traditional regional load forecasting that is usually performed on bus-level, these techniques forecast the energy demand of an individual customer, making them provide fine-grained decisionmaking support for upper-level demand side management (DSM). Followed by this, this chapter introduces the state-of-the-art of two representative application domains upon load forecasting: trans-active energy (TE) systems and DSM. The former focuses

391

392

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

on integrating load forecasting into energy trading mechanism of prosumers, while the latter focuses on designing energy management strategies for demand side energy entities—buildings, microgrids, and virtual power plants (VPPs). This chapter is expected to provide a reference to researchers, engineers, educators, and higher degree students in relevant fields.

12.2 New trends in load forecasting To build a more sustainable energy future, increasing the uptake of renewable energy sources is an inevitable trend in the transformation of power systems. Countries with rich renewable sources such as the United Kingdom, Germany, and Australia all announced their plans to significantly increase the proportion of renewable energy in the decades to come. Australia, for example, is looking forward to powering the country with 20% renewable generation by 2020 and gradually increasing to 100% by 2050 [3]. On the other hand, the increasing trend of energy demand has never stopped, with an annual growth rate of 2.1% in 2017, doubled from the rate in the previous year [4]. Together, these two trends have posed a great challenge to existing energy systems. That is to make our energy consumption more sustainable while meeting the everincreasing demand. Due to intermittency of renewable generation, technologies such as energy storage [5] and demand response [6] are proposed. In the scenarios of future energy systems, traditional energy users will gradually be transformed to prosumers, who are defined as energy consumers with local power generation. Ideally, prosumers’ power generation will be from renewable energy sources, such as rooftop photovoltaic (PV) panels. Solutions of energy storage and demand response can promote selfconsumption, which lead to less energy loss in transmission and distribution and thus a higher utilization rate of renewable energy. However, all these supportive technologies will rely on highly uncertain information such as the amount of generation from renewables and the flexibility of demand in the near future. Therefore, the focus of load forecasting has gradually shifted from the system level to the individual customer level, which can be further beneficial to future power grid operations.

12.2.1 Introduction of load forecasting for individual energy customers Short-term load forecasting has been a heavily discussed topic in power systems for many years. At the early stage, traditional statistical models such as regression and state space models were applied [7]. Electric load readings are primarily modeled as discrete time series, so other statistical time series models such as Auto Regressive (AR), Moving Average (MA), Auto Regressive Integrated Moving Average (ARIMA) can be applied to predict the load [8]. Many Artificial Intelligence (AI) load forecasting

Load forecasting in the short-term scheduling of DERs

approaches have been proposed as well, including the most commonly adopted methods such as Artificial Neural Networks (ANN) and Support Vector Machine (SVM) [9]. AI approaches are combined to form hybrid models that show better performance in load forecasting, including fuzzy logic system with ANN [10] and ANN and ARIMA models [11]. However, most of the above mentioned methods may be adequate for load forecasting at aggregation level, such as the electricity load on a system or a feeder of a substation. In those cases, due to the diversity of numerous heterogeneous electricity consumption behaviors, the resultant daily load curves are relatively smooth, while abrupt fluctuations and high load variability are frequently observed in the individual load profiles. Thanks to the ongoing worldwide expansion of Advanced Meter Infrastructure (AMI) deployment, it is possible to put differences between aggregated and individual load profiles under scrutiny. For this matter, the Smart-Grid Smart-City (SGSC) Customer Trial project provides an ideal dataset for individual load forecasting research [12]. On aggregate level of all available SGSC customers, the distribution of daily load profiles is shown as Fig. 12.1, while on individual customer level, daily profiles for four randomly picked customers are shown in Fig. 12.2. It is clear to see that aggregate/system level and individual level load profiles exhibit completely different patterns. Electric load on the aggregate level tends to be influenced by many contextual factors such as temperature, weather condition, the time of a day, the day of a week and other occasional events such as holidays, sport

Figure 12.1 Daily profiles of 8000 1 SGSC customers over the summer Dec 2013
Feb 2014 in New South Wales, Australia.

393

394

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 12.2 Daily profile for four randomly selected customers in SGSC dataset.

events, etc. The profiles shown in Fig. 12.1 are drawn from the historical data of the summer in Australia from 2013 to 2014 in New South Wales (NSW). The highest temperature of a summer day in NSW can vary a lot. Correspondingly, the range of daily peak loads is also very wide at aggregate level as seen in Fig. 12.1. Unlike aggregate profiles where consistent and traceable patterns can be clearly seen, load profiles from individual customers often vary from one another and show less correlation to contextual variables such as temperature. The underlying reason for such observation may highly relate to the fact that each household may likely exhibit their unique lifestyles and their consistency in daily routines. For example, Customer 10346328 from SGSC dataset shows high load around midday in the weekends compared to weekdays. This may indicate that this customer can be a working class and usually stays at home during the daytime of weekends. Similarly, Customer 8328038 is likely to demonstrate the load profiles of a holiday house, with most of its daily electricity profiles being flat and occasional with large consumption on weekends. On the other hand, the other two randomly picked customers show rather consistent daily routines in both weekdays and weekends. However, the daily peak demand is likely to happen at different times of every day. Due to the much higher level of volatility present in individual loads, traditional load forecasting methodology designed to work with external contextual features is no longer applicable in individual load forecasting. Specifically, an ensemble model of Extreme Learning Machine (ELM) significantly improved the forecasting performance measured by the mean absolute percentage error (MAPE) to 1.82% for the

Load forecasting in the short-term scheduling of DERs

load forecasting at Australian national level [13], but failed completely in forecasting individual loads with an average MAPE about 122.90% in SGSC [14].

12.2.2 Dynamic probabilistic household load forecasting Given the problem nature of load forecasting for individual energy customers, methods that are more capable of learning intrinsic patterns and correlation from the load profiles themselves shall be more desirable. This is because at residential level, the load consumption can be better explained by the collective effect of the activities done by the residents. Based on the characteristic of their underlying motives, there are four main categories of all practices according to the Practice Theory, namely ‘Practical Understanding’, ‘Rules’, ‘Teleo-affective’ and ‘General Understanding’ [15]. If an actor knows when and what to do for a particular activity, this activity falls into the ‘Practical Understanding’ category. ‘Rules’ type of activities generally refers to those appliance operations that are constrained by the technical limits of the system. For example, the dishwasher has preprogrammed washing procedure once it is started. ‘Teleo-affective’ practices correspond to those goal fulfilling activities, such as making a cup of coffee and watching TV. ‘General Understanding’ type may refer to a broader range of activities with certain persistence and regularity over time, such as family bonding or religion related activities. All these activities practiced by every resident in a household will collectively constitute the interaction and engagement of various electric appliances that result in certain pattern in electricity consumption. Since the electric load of an individual household is generally determined by the activities of residents, the key of accurate forecasting will be to capture the temporal correlations between underlying activities from the observed history load profiles. For this purpose, a recurrent neural network (RNN) is an ideal sequence-based model for this task because of its ability to establish the temporal relationship between previous information and the current states. However, the gradient from long-term components of standard RNNs decrease exponentially to zero, limiting RNN’s ability to learn long-term correlations [16,17]. Therefore, the long short-term memory (LSTM) structure is designed to explicitly store the past information in a memory cell and control the activeness of the memory cell using forget gate. With such refinement, the LSTM model is able to learn correlations across longer terms [18]. Let fx1 ; x2 ; . . .; xT g denote a typical input sequence, where xt Aℛk represents a k-dimensional vector of real values at the tth time step. Each element in the sequence is processed by a single LSTM cell which is illustrated by Fig. 12.3. The temporal connection is maintained and updated by the interactions by the internal memory cell and the intermediate output. Specifically, the memory cell state st is updated by combining the previous memory st21 and the previous output from the previous LSTM cell ht21 through input node gt , input gate it and forget gate f t , while the

395

396

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Copy

Con..

Vector definion

Element-wise operaon

Copy

Unweighted vector transfer

Weighted vector input

Copy

Con.

Vector copy

Vector concatenaon

Neural network acvaon funcon

Figure 12.3 The structure of an LSTM block for a single time step.

LSTM cell output ht is derived by incorporating the updated memory st and the input through the output gate ot . The computations of all these gate functions are given below:
 f t 5 σ Wfx xt 1 Wfh ht21 1 bf ð12:1Þ it 5 σðWix xt 1 Wih ht21 1 bi Þ
 gt 5 φ Wgx xt 1 Wgh ht21 1 bg

ð12:2Þ ð12:3Þ

ot 5 σðWox xt 1 Woh ht21 1 bo Þ

ð12:4Þ

st 5 gt }it 1 st21 }f t

ð12:5Þ

ht 5 φðst Þ}ot

ð12:6Þ

Load forecasting in the short-term scheduling of DERs

where Wgx , Wgh , Wix , Wih , Wfx , Wfh , Wox and Woh are weight matrices for the corresponding inputs of the network activation functions; } stands for an element-wise multiplication; σ represents the sigmoid activation function, while φ represents the tanh function. The loss function of the standard LSTM forecasting model is usually set to be the mean square error (MSE), which is written as lmse 5

T 1X ðyt 2^yt Þ2 T t51

ð12:7Þ

where T is the total number of time intervals in the load sequence, yt is the actual value of load at time t, and y^ t is the forecasted load at time t. With this, the LSTM will learn to map the nonlinear relationship from the past load trajectory to a scalar value for the future load, which is also known as the point forecasting. However, due to the volatility of individual loads, it would be more informative to forecast not only the future load values but also the confidence levels of the forecasted values. This is also known as the probabilistic load forecasting. One way of transforming the LSTM point load forecaster into a probabilistic one is to incorporate with the concept of Quantile Regression (QR). Compared to general point regression model, QR aims to forecast the upper and lower bounds with predefined quantiles. By borrowing the idea of linear QR as proposed in [19], the MSE loss function can be substituted by the quantile function, also known as the pinball function which is formulated as below: T 1X lτ 5 lτ;t ðyt ; y^ τt Þ T t51

ð12:8Þ

where lτ denotes the overall quantile loss, y^ τt is the forecasted load for the τ th quantile. The lτ;t ðUÞ denote the pinball loss at t th time step, which is detailed by: lτ;t ðyt ; y^ τt Þ 5




 ð1 2
τ Þ y^ τt 2yt τ yt 2 y^ τt

ð12:9Þ

The substitution of the loss function allows to train model ℳτ to be trained for specific forecasts given as the predefined target quantile τ. Subsequently, the confidence intervals of the forecasts at each time step can be predicted by using each quantile specific model. One example of the probabilistic load forecasting is demonstrated by Fig. 12.4. From the figure, apart from the point predicted values, the 90%, 95% and 99% confidence intervals are also given. It can be seen that these intervals are robust to encompass the actual testing load values.

397

398

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Figure 12.4 Probabilistic load forecasting for an individual customer over a week.

12.2.3 Consumption behavior-driven household load forecasting As pointed out in the previous sections, the key to better forecast individual load is to learn and perceive the customers’ behavior. The lifestyle patterns of residential customers are generally hidden in the whole-house energy consumption data. However, the most commonly available smart meter data resolution is 1 m reading every 30 minutes. Such resolution is generally too coarse to generate sufficient insight on residents’ behavior according to [20,21]. Generally speaking, the lifestyle of a household can usually be reflected by the way the residents are interacting with their major appliances. That is the time of the day, the frequency, the duration that the users are using, for example, washing machines, air conditioning, dishwashers and so on. The effectiveness of incorporating indirect information regarding the customer behavior into the individual load forecasting has been demonstrated by [22]. Major home appliances such as clothes dryer, clothes washer, dishwasher, heat pump, television, and wall oven tend to involve relatively high level of user interaction, so they may better indicate the lifestyle of a household. With resolution as low as 30-minute-interval data sequences for these appliances, the deep LSTM model has seen noticeable performance improvement for individual load forecasting. In recent years, many countries have published their technological roadmap for massive deployment of smart meter technologies with enhanced functionality. For example, in Australia, a minimum smart meter functionality was proposed to enable higher data refreshing rate as high as 5-second sampling interval for in-home display [23]. At this data resolution, [21] has shown that major appliances can be identified to a high level of accuracy based on deep learning approaches. In the future trend of the individual forecasting research, it would be worth exploring to incorporate the NonIntrusive Load Monitoring

Load forecasting in the short-term scheduling of DERs

(NILM) technology into learning the customers’ behavior from smart meter with higher resolution, which can further improve the load forecasting performance.

12.3 Trans-active energy systems with DERs In future power industry, TE systems are considered to be a promising approach for accommodating a high penetration of DERs while ensuring the operation security and efficiency of power systems [24]. Among various definitions of TE in existing literature, the one proposed by the Grid Wise Architecture Council is widely accepted, namely TE is the economic and control methodology for managing consumption, generation and energy trading within a power distribution network based on market mechanisms [25]. Load forecasting is crucial for the optimal scheduling of DER outputs and the setting of bid parameters for consumers in TE systems. Different from conventional power systems where generation resources are fully dispatchable to meet the real-time load demand, output of DERs are featured by characteristics of intermittency and uncertainty. Considering that currently it is challenging to store energy at a large scale, DERs has to be adjusted to demand in real-time. Therefore, an accurate electric load forecasting can significantly contribute to the localized accommodation of DER outputs in TE systems. Otherwise, curtailment of DER output will occur frequently without an optimal scheduling of generations, such as due to the severe voltage rise issues [26]. Besides, when setting bid parameters in TE systems, consumers also need an accurate forecasting of their own electricity consumptions to optimize their transaction benefits. Penalties will apply to deviations between the winning bids in TE systems and the real-time load demand of consumers [27]. The establishment of TE systems is a prerequisite for the application of load forecasting in future digital grid systems. Key problems of designing TE systems for the integration of DERs into power systems are elaborated in this subsection. In particular, the zero marginal cost of DERs and decentralization of electricity transactions in TE systems are introduced in detail.

12.3.1 Distribution market mechanism for DERs with zero marginal costs During the past decades, the global penetration level, capital investment, and installed capacity of renewable energy generations have been increasing steadily. In particular, DERs are widely deployed in the demand side. For instance, nearly one out of every four households in Australia has already installed rooftop solar panels [28]. In order to enable end-users benefit from DERs and manage behaviors of prosumers in distribution systems efficiently, the design of TE systems has gained widespread interests [29]. Electricity transactions in a distribution system are different from those in a wholesale electricity market and have their own characteristics.

399

400

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

First, the majority of participants in distribution markets are prosumers with smallscale renewable DERs and individual end-users. The bidding or offering quantity of electricity from each participant is usually small. • Second, the traded energy in a distribution market may be 100% from renewable generations. Compared with conventional generation technologies, such as the fossilfueled power generation, the renewable generation is featured by a fuel cost of zero. • Third, participants in a distribution market are often incompetent of developing sophisticated bidding strategies and therefore prefer the set-and-forget method of setting bid parameters when participating in transactions. All these above trading characteristics should be considered in the design of distribution market mechanisms. Existing electricity market mechanisms can be broadly categorized into two types [30]: • Power pool based market model (centralized market). • Bilateral contract based market model (decentralized market). The power pool model is utilized to produce price signals for electricity spot market participants and achieve the supply-demand balance in the concerned power system. It also helps provide a price reference for transactions in the bilateral contract market which is mainly utilized by participants to manage trading risk. In a power pool market, both offers of generation output and bids of load demand are submitted to the market operator. Through the market clearing of these offers and bids, the market operator determines the winning generation output and load demand, as well as the corresponding prices at which every participant is paid (for generation) or charged (by consumption) [31]. The power pool market can be further classified into single-sided and double-sided auction markets. In a single-auction, only producers participate in the market bidding and the objective function is usually to minimize the total system cost of purchasing electricity. In a double-sided auction, both producersand-consumers participate in the bidding, where the objective becomes the maximization of net social welfare. Fig. 12.5A and B show the market clearing mechanism for the single- and double- sided auction electricity markets, respectively. Renewable DERs are featured by a capital intensive investment but a zero fuel cost, which is distinct from conventional generation technologies [32]. In general, generation cost is determined by factors from several aspects including upfront investment, operating expenses, and capacity factors. For a specific generation technology, its cost can be broadly categorized into two types: fixed and variable costs. The fixed cost remains unchanging despite the quantity of electricity generation, such as the upfront capital investment and cost of land for power plant construction. On the contrary, the variable cost usually changes with generation output, such as the fuel cost, labor cost, material cost, start-up/shut down cost, emission cost, as well as operation and maintenance (O&M) cost. In particular, if compulsory maintenance is scheduled for a renewable generator, such as a wind generator or solar PV panel, then its O&M cost can be regarded as fixed cost. Therefore, once a DER is put into

Load forecasting in the short-term scheduling of DERs

Merit order curve of supply

Price

Price

Merit order curve of load demand

Merit order curve of supply

Load demand Equilibrium point that determines the market clearing price

Market clearing price

0 Power (A) Single-sided auction electricity market Dispatched output

Failed generation bids

0 Power (B) Electricity market with double-sided auction

Dispatched output Dispatched load demand

Failed bids

Figure 12.5 Market clearing mechanism for single- (A) and double-sided (B) auction electricity markets.

operation, no additional cost for generating each megawatt-hour of electricity will occur, and thus the marginal cost of DER generations can be recognized as zero. Meanwhile, the market clearing price (MCP) in current marginal cost based electricity markets is merely determined by bidding price of the marginal unit, so when competing with generators that have nonzero marginal costs, DERs with zero marginal costs are not motivated to bid at nonzero prices. This can be verified by the fact that the zero bidding price or even negative bidding price has already been observed in some practically operating electricity markets. However, these electricity markets are still working well because of the following reasons: • The penetration level of renewable energy generation is moderate in current power systems and various subsidies are available from governments for renewable energy generation. This can help renewable generators to survive even when zero and negative price occurs. • Fossil-fueled generation units have not only huge fixed costs of construction but also high marginal operating costs for electricity generation. This results in their nonzero bidding prices in electricity markets. Then, when competing with renewable generators, fossil-fueled generators usually act as marginal generation units and their bidding prices determine the final MCP. In other words, bidding a zero price can bring renewable generators some priority in system dispatch but they are still very likely to being paid by a nonzero MCP. Problem will arise in a 100% renewable energy scenario. Renewable generators that bid at a nonzero price can only decrease its own priority of being dispatched but help rise the MCP for other units if the other units choose to bid at zero. More importantly, marginal cost based market is designed to efficiently price the short-term operation cost of power systems [33], but renewable generators make decision mainly based on their long-term costs, such as their capital and maintenance costs. Therefore, marginal cost based market mechanism would fail to reveal the real market value and generation cost of renewables.

401

402

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

One of the options is to adopt an average pricing mechanism to overcome the above problems, as is shown in Fig. 12.6. In the APM, the weighted average r of participants’ bidding prices is adopted as the MCP, where the weighting factors are their bid quantities. Since it is the average price that acts as the MCP, this market clearing mechanism is called the Average Pricing Market (APM). The mathematical formulation of APM is presented as follows.   X X cs cb b max f pcb ; p p r 2 pcsj rjs ð12:10Þ 5 i j i i iAN

P

P

b b s s iAN ri pi 1 jAM rj pj P P b s iAN pi 1 jAM pj

s:t: r 5

X

ð12:11Þ

b ’iAN; pcb i 5 0 if ri , r

ð12:12Þ

’jAM ; pcsj 5 0 if rjs . r X X pcb pcsj 5 0 i 2

ð12:13Þ

iAN

2pmax # l

jAM

pcb i dl;i 1

iAN

ð12:14Þ

jAM

X

pcsj dl;j # pmax ; ðl 5 1; 2; . . .; LÞ l

ð12:15Þ

jAM b cs s 0 # pcb i # pi ; 0 # pj # pj

ð12:16Þ

Incumbent electricity retail price

Price

Equilibrium of transaction

Merit order curve of generation bids

Market clearing price Merit order curve of load demand bids Feed-in tariff

0

Power rate of demand/output Winning producers

Winning consumers

Figure 12.6 Schematic diagram of the proposed market mechanism.

Failed bids

Load forecasting in the short-term scheduling of DERs cs where N/M is the set of consumers/producers. pcb i /p j denotes the final dispatched th th load demand/generation output of the i consumer/j producer. Pmax indicates the l th transmission limit of the l branch. L represnets the set of branches in a distribution network. dl,j is the power transfer distribution factor which is used to indicate the relative change of the active power that occurs on a particular branch l due to actual power change at node j. In the market clearing model Eqs. 12.10
12.16, Eq. 12.10 is the maximization of social welfare in the electricity market. Eq. 12.11 calculates the MCP. Eqs. 12.12 and 12.13 define the rules of the participants being dispatched. Eq. 12.14 ensures the market equilibrium of supply and demand. Eq. 12.15 shows the network transmission constraint. Eq. 12.16 represents the constraints on decision variables. Compared with the marginal cost based electricity markets, the APM mechanism has the following advantages [34]: • Participants are motivated to set bidding parameter based on their own estimations of the generation costs (for producers) or electricity utilities (for consumers), which is defined as honesty. Honesty is proved to be a dominant strategy for participants in such a market, which enables the proposed mechanism to develop into a setand-forget bidding market. • Using the proposed market mechanism, the problem of always bidding at a zero price by renewable generators in existing markets can be avoided. Even in a scenario where only renewable generation units with zero marginal costs participate in the bidding, the proposed mechanism can still produce a reasonable price signal. • The APM mechanism is compatible with the nodal pricing system. Merits of the nodal pricing can still be retained when adopting the proposed mechanism in the distribution market.

12.3.2 Decentralized market mechanism for DER transactions Different from wholesale electricity market, the distributed nature of DERs and energy consumption in distribution systems requires decentralized market mechanism for DER transactions. In practice, some trials and projects of peer-to-peer (P2P) electricity transaction mechanisms in electricity distribution systems have already been implemented in several countries. These P2P projects aim to stimulate the participation of electricity end-users in energy transactions. As a necessary platform that supports DER transitions in a distribution network, the design of decentralized market mechanisms is of significant importance. In a centralized market, producers-and-consumers need to submit their bids to market operator who takes in charge of the market clearing process and determines the winning bids. Differently, in a decentralized market, transactions are organized directly among participants without a middle man/entity or agent. Fig. 12.7 compares

403

404

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Participant 1

Participant j

Participant 1 Participant 2

Participant N

Market operator

Participant 3

Participant 4

Participant i+1 Participant i

(A)

Participant 2

Participant N

Participant j

Participant 3

Participant 4

Participant i+1 Participant i

(B)

Figure 12.7 Message transmission among participants under different market structures. (A) Structure of communication network for centralized market. (B) Structure of communication network for decentralized market.

the message transmission among participants under centralized and decentralized market structures. However, when implementing a trading mechanism in practice, various types of network topologies can be considered depending on the actual requirements, such as the ring topology, the fully connected topology, and the bus topology. Decentralized market mechanisms for DER transactions can be derived from the centralized nodal pricing model that is widely adopted in existing electricity markets, so merits of existing markets can still be retained, such as the high efficiency in pricing network congestions, and the guarantee of network operation security. In the decentralized formulation of electricity market, the market clearing process is carried out though the iterative computation, where each participant solves its own subproblem and passes selected parameters without releasing its privacy to other participants. The calculation ends when the iteration converges to its optimal results. Besides, it can also be verified that the decentralized formulation can obtain the same market clearing outcomes as the original centralized market mechanism. In mathematical optimization community, endeavors have been made to seek efficient methods to decompose an intractable problem into several subproblems. As a method that combines the advantages of dual decomposition and augmented Lagrangian methods for constrained optimization problems, the Alternating Direction Method of Multipliers (ADMM) is a simple but powerful algorithm for distributed convex optimization problems [35]. Since the ADMM is originally introduced for the special case where there are only two blocks of variables in the optimization problem, in [36] the Gauss-Seidel and Jacobian ADMMs are proposed for cases with three or more blocks of variables. Especially, the Jacobian ADMM is featured by the advantage of enabling a parallelized updating of all variables. In [37
39], the ADMM algorithm has been studied to develop distributed computational methods for optimal power

Load forecasting in the short-term scheduling of DERs

flow (OPF) problems. In [40], a fully decentralized distribution market mechanism is derived using the ADMM method. Furthermore, bidding strategies of renewable generators due to their zero marginal costs are considered in [41]. Since a decentralized distributed market mechanism is computed in an iterative way, its convergence is hence an essential issue. Notably, when using the ADMM method to develop a decentralized market, all convergence results that hold for the ADMM still hold for the decentralized market model. In [36], it is proved that if certain conditions on parameter selection are satisfied, the Proximal Jacobian ADMM can achieve global convergence at an o(1/k) convergence rate. Besides, the added proximal term in the objective function also enables the subproblem to become strictly or strongly convex if it is not originally. Another essential problem in the decentralized market mechanism for DERs is the preservation of privacy. As the bid/offer data from participants are private, the information of each participant should be prevented from leaking to the others. The information leakage may lead to the speculative behaviors of some participants, and even result in the failure of the market mechanism. Compared with the centralized electricity market, the decentralized one does not require a market operator who is in charge of the market operation and manage private information of participants. Therefore, protecting the privacy of participants becomes a more severe problem in a decentralized market mechanism. In [41], the aggregated value of the bidding/ offering data from other participants is used to solve each subproblem, namely the aggregated values of bids/offers are transmitted in the communication network. Each participant solves its own optimization problem after receiving the transmitted messages from his/her neighbors. Through the aggregating of data, the bid/offer information of participants is being encrypted. These data will not be decrypted during the whole process, thus the distributed market mechanism can be operated in a way without exposing the personal information of each participant to the others. The personal information of each participant will only be used for solving his/her own optimization problem. The decentralized distribution market mechanism for DER transactions is characterized by the huge number of participants considering the tens of thousands of small-scale DERs and end-users in distribution systems. Therefore, when designing a decentralized market mechanism, it should have a good scalability for integrating more users. Besides, bidding outputs of DERs are determined by generation forecasting results in a distribution market. The intermittence and uncertainty of renewable energy requires a distribution market to be organized in a flexible way, which can be operated as half-hourly, or hourly ahead market instead of a day-ahead market. Under this circumstance, the forecasting results for renewables can be quite accurate. Consequently, the convergence speed of decentralized market mechanisms also needs to be guaranteed so that transactions can be carried out timely.

405

406

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

12.4 Short-term scheduling of DERs in demand side The high penetration of DERs assigns demand side entities the capabilities of producing energy and actively re-shaping their energy consumption profiles. This adds more flexibilities to design DSM [42] strategies. Load forecasting is playing an increasingly important role in the demand side environment: (1) first, load forecasting is a critical factor for the local energy supply-demand balance of the customer side, which is directly associated to the energy cost of customers. In demand side, computations and decision-making processes are often performed in facilities with limited computing capabilities (e.g., personal computers, smart phones, and smart meters); this makes sophisticated stochastic analysis for load and renewable power output, which are widely applied in grid-level analysis, can hardly be performed in demand side energy management. Therefore, the cost benefit brought by DERs to the customer is largely affected by the accuracy of point load forecasting. (2) Second, load forecasting affects DER scheduling, which consequentially affects customers’ net-load profiles and the distribution of reverse power flows generated by the DERs. Undesired patterns of these factors would influence the external grid’s reliable and secure operation. Based on the above discussions, the load forecasting techniques introduced in Section 12.2 can be considered as a fundamental component for these DSM strategies, which establish higher-level control logics to optimize the energy consumption and production of DERs subjected to different operational objectives and considerations. In this subsection, we provide an introduction on energy management techniques for DERs in different DSM contexts: buildings, microgrids, and VPPs. These can be considered as typical application domains of load forecasting in demand side.

12.4.1 Short-term scheduling of DERs in buildings Buildings are critical infrastructure and large energy consumer in human society. Statistical data show that buildings account for 30%
40% of the final energy usage worldwide [43]. Technological advances in recent years have been significantly transforming modern buildings from a static civil infrastructure to complex cyber-physical systems. These advances include: • Integration of DERs and energy storage systems (ESSs). Increasing capacity of distributed renewable energy sources has been integrated into building side (e.g., solar panels installed in building’s rooftop and facade, on-site wind turbines, geothermal systems). To accommodate renewable energy, ESSs have also been increasingly integrated in buildings. • Electrification. Building electrification refers to replacing fossil fuel-powered appliances such as space heating, water heating, cooking and laundry with electricity, and other fossil fuel-free, zero-carbon alternatives. Building electrification can provide more flexibility for the integration of variable energy resources and assign

Load forecasting in the short-term scheduling of DERs

larger capabilities for buildings to provide demand response support and ancillary services to the grid. • Digitization and automation. Advances of Internet-of-Things and ubiquitous computing technologies in past few years enable buildings to acquire fine-grained data from various devices and real-time monitor their operational environments. Automation technologies make buildings automatically adjust their operations based on different conditions. In this sense, digitization and automation can significantly improve buildings’ efficiency and drive down their operation costs. With the penetration of various energy resources and communication and control facilities, modern buildings can act as very small, self-contained energy systems, referring to as “Nanogrids” in some literature [44]. Managing DERs in building environments is usually associated with the development of building/home energy management systems (BEMSs/HEMSs). A BEMS is a centralized expert system that automatically controls building-side energy resources subjected to different objectives (Fig. 12.8). These energy resources include renewable energy sources and other controllable energy resources, which can be generally categorized as follows: • Distributed renewable energy sources, consisting of on-site energy sources that serve residential loads. The most common renewable energy sources in buildings include PV panels and wind turbines. • Battery energy storage systems (BESSs) that can store energy and provide back-up electricity for buildings.

Figure 12.8 Illustration of a building/home management system.

407

408

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

Plug-in electric vehicles (EVs). In their plug-in period, the building load can be served by batteries of EVs. • Thermostatically controlled appliances (TCAs) include heating & cooling appliances, such as electric water heaters, refrigerators, and air conditioners. • Deferrable, nonthermostatically controlled appliances (NTCAs)—deferrable NTCAs are appliances that have the flexibility to shift their operational time. Examples of these applications include washing machines, clothes dryers, etc. • Distributed, fossil-fueled generators. Distributed generators (such as fuel-cell generators and combined heat and power generators) can also be used as power backup devices in residential buildings. Buildings are essentially occupant-centric environments. Scheduling of DERs in buildings thus need to take into account the occupant’s requirements. One or multiple following factors are usually considered in DER scheduling in the building environments: 1. Building’s energy cost. When the building is operated under time-varying electricity tariffs (e.g., time-of-use and real-time pricing) and variable renewable power output, different operational schedules of DERs would lead to different energy costs. A large number of works (e.g., [45
47])have been devoted to optimally schedule the energy consumption of load and operation time of DERs to minimize the building/home’s energy cost. The occupant’s lifestyle-related requirements on the DERs, such as allowable operation time range for the appliances, desired indoor temperature range, plug-in time period of the EV, are incorporated as constraints of the DER scheduling model. 2. Occupant’s utility. The term “utility” is defined as represent the benefit, happiness, or satisfaction obtained from use of building energy resources. Some works (e.g., [48,49]) establish building energy management models to optimally schedule building-side DERs to maximize the occupant’s utility. Different utility models can be used to model the occupant’s satisfaction. For example, [48] constructs the energy demand and benefit curve from a historical residential dataset, as depicted in Fig. 12.9. The crossover point between the demand and benefit curves indicates the point where the maximum total benefit is achieved. Based on the curve, [49] develops a DER scheduling model that maximizes the difference between the benefit value and the sum of electricity cost and the occupant’s comfort loss. 3. Grid operation support. This refers to using DERs’ operation flexibility to provide demand response support to assist the utility to improve the grid’s operation while satisfying the occupant’s requirements on comfort and energy cost. Various grid operation indices can be considered, such as peak power, bus voltage fluctuation, etc. For example, [50] designs a scheduling model for EVs and appliances, with the aim of controlling the total power consumption of the building below a prespecified limit assigned by the grid. Ref. [36] schedules a residential BESS to minimize the reverse power flow from the home to the grid, so as to avoid voltage

Load forecasting in the short-term scheduling of DERs

Figure 12.9 Illustration of demand and benefit curves from consumer behavior.

fluctuation issues. With the proposed BESS scheduling model, they evaluate the benefit improvement for a set of Australian users under different financial incentive policies, that is, gross metering- and net metering- based incentive policies. The inevitable uncertainties of building load forecasting and renewable output forecasting would affect building’s operation. Some researches use stochastic modeling techniques (e.g., [51
53]) to model such uncertainties in BEMSs/HEMSs. Although the techniques can well capture the stochastic nature in the building environment, to what extent DER’s uncertainty should be considered in building energy management is still a controversial issue. Some researchers [38] argue that deterministic building energy management model is more practical and valuable than the stochastic ones. Their opinion is based on the fact that, compared with the high computational complexity brought by the stochastic methods, the economic value increment to the end user is limited (scale of a few cents). In addition, building energy management programs usually run on building-side processors (e.g., embedded chips in smart meters and personal computers), which are with limited computing capacity and may not be suitable for performing stochastic analysis that is usually compute-intensive.

12.4.2 Short-term scheduling of DERs in microgrids The concept of ‘microgrid’ [54] naturally raises along with the wide deployment of DERs. As its name implies, a microgrid can be considered as a small grid, referring to a group of interconnected loads and DERs with clearly defined electrical boundaries that acts as a single controllable entity with respect to the grid [55]. These DERs usually include: ESSs (BESSs, EVs, and supercapacitor- and flywheel-based ESSs), wind/ solar power sources, distributed fossil-fueled generation units, and flexible loads (i.e., the loads whose power consumptions can be adjusted or temporarily shifted).

409

410

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

•

•

A microgrid can operate in the following modes: Grid-connected mode. In this mode, a microgrid uses the energy generated from its managed DERs to serve its loads, while it also connects to the external macro grid. The microgrid can import power from the grid to serve the loads when it does not have sufficient energy from its own DERs; it can also export surplus power from the on-site DERs to the grid. Islanded mode. In this mode, a microgrid is decoupled from the macro grid and acts as an independent energy system. It uses its managed DERs to serve its own loads in a self-contained manner.

12.4.2.1 Centralized and distributed DER scheduling in microgrids Both centralized and distributed approaches have been developed for scheduling DERs in microgrid environments. In the centralized DER scheduling approaches, a centralized optimization model is usually established to allocate power outputs and power consumptions among the microgrid’s managed DERs, aiming at maximizing/ minimizing a certain operational objective of the microgrid. The primary objective of DER scheduling in a microgrid is minimizing the microgrid’s operation cost [56,57], which is usually equivalent with minimizing the imported energy from the external grid. Other considerations are also incorporated into the DER scheduling objective, such as the environmental impact of the DERs’ operations in terms of CO2 emission [58] and the profile of selling energy to the grid [59]. A microgrid is essentially an environment where energy resources are located and interconnected in a distributed manner. Many distributed scheduling methods are developed to schedule the DERs in a microgrid. Some of these techniques usually do not involve a centralized energy management system (EMS), but rely on the communications and cooperation of the DERs (called the “agents’) to achieve the scheduling objective. For example, [60] develops a distributed scheduling technique that enables interconnected energy storage devices in a microgrid to store the surplus energy generated from renewable sources and serve the microgrid’s internal load in high-price hours. This is achieved through facilitating the energy storage systems to iteratively exchange information with their neighbors and update their power charging/discharging decisions; with this P2E interaction, the ESSs will mutually converge to an optimal charging/discharging scheme. 12.4.2.2 Resilient DER scheduling in microgrids Resiliency-oriented energy management of microgrids refers to enhance the selfenergy supply capability of microgrids subjected to significant fault/disturbances or the islanded operation mode. In these scenarios, economic cost is not the primary objective. Instead, minimizing the curtailed load would be considered as the most important objective. Based on this, the power consumption and generation of DERs in a

Load forecasting in the short-term scheduling of DERs

microgrid can be optimally scheduled to serve as much load as possible over a certain period or implement quick islanding of the microgrid in the events of main grid supply interruption and/or significant system disturbances. The uncertainty is an important consideration in resilient DER scheduling in microgrids. Such uncertainties would not only include forecasting errors of load and renewable power output, but also include other factors such as the main grid interruption time and duration. These uncertainties have direct impact on the amount of curtailed load. Handling load/power output forecasting uncertainties is thus necessary in resilient DER scheduling. Different DER scheduling methods have been developed to support uncertainty-aware microgrid resilient operation. These methods utilize different stochastic modeling/programming techniques, such as robust optimization [61] and stochastic programming [62].

12.4.3 Short-term scheduling of DERs in VPPs Another aggregation form of geographically dispersed DERs is the VPP [63]. As its name implies, a VPP refers to an entity that aggregates a number of DERs and possesses the visibility, controllability, and functionality exhibited by conventional power plants. The concept of VPP is depicted in Fig. 12.10. The dispersed energy resources in different geographical areas (wind/solar farms, distributed renewable energy sources, ESSs, EVs, flexible loads, etc.) are aggregately and remotely managed by a centralized VPP-EMS. The VPP-EMS performs as a control center of the VPP; it monitors the information of the energy resources and makes different control decisions, such as generation scheduling, device fault diagnosis, bidding in energy markets, etc. Area 2

Area 1

Area 3 Generaon scheduling

VPP EMS

Power flow

Grid

Figure 12.10 Schematic of a virtual power plant.

Bidding & trading

Power market

411

412

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

Just like microgrids, a VPP is normally connected to the main grid and bilaterally exchanges power flows with the grid. A main difference between VPP and microgrid is that a microgrid usually covers a specific area, while the energy resources in a VPP can belong to different geographical areas that are not directly connected. The energy sources communicate with each other and with the VPP-EMS via communication networks. Another difference is that a microgrid is usually not eligible to participate in grid-level energy markets due to its capacity barrier, while a VPP can be eligible due to the large capacity of energy resources it aggregates. In this way, a VPP is able to make profits through selling energy to the market. Depending on their VPPs, it can be categorized into two types: commercial VPPs (CVPPs) and technical VPPs (TVPPs) [63]. Correspondingly, the short-term scheduling of DERs in VPP context is subjected to different objectives: 1. For CVPPs, scheduling of DERs usually aims at maximizing the economic profit of the VPP through facilitating it to put bids and sell energy in grid-level energy markets. For example, [64] develop bidding strategies to determine optimal bids for a VPP to put in the wholesale power market and the spinning reserve market, so as to maximize the total profit of the VPP in both markets. Prevalence of distributed energy resources in urban areas has been fostering distributed energy trading activities in recent years, as introduced in Section 12.3. In this context, VPP systems can also be established to directly sell energy to end energy consumers in distributed energy markets. 2. For TVPPs, scheduling of DERs is performed to provide regulation and auxiliary services to maintain the safe, reliable, and efficient operation of the external power system. Power consumption and production of the dispersed DERs can be scheduled and controlled to implement different objectives, such as network peak load reduction, bus voltage regulation, and frequency stabilization.

12.5 Conclusions and future thoughts This Chapter introduces new trends in load forecasting techniques under the smart-grid environment. Due to the rapid growth of renewable generation and the ever-increasing flexibility of load demand in current energy systems, the focus of load forecasting has gradually shifted from aggregated system level forecasting to the estimation of short-term individual load consumptions. In particular, as there is a much higher level of volatility hidden in individual loads, traditional load forecasting methodologies is no longer applicable in individual load forecasting. Considering that the load profile of an individual household is generally determined by activities of residents, the key to carry out accurate load forecasting is to capture the temporal correlations between the underlying activities from the observed load profiles of the past. Then, methods that are more capable of learning intrinsic patterns and correlation from history load profiles

Load forecasting in the short-term scheduling of DERs

are desired. For this purpose, both RNN and LSTM are good options, which are elaborated in this chapter. Besides, the integration of behavior analysis in the individual load forecasting is proved to be an effective way to improve the forecasting accuracy. To incorporate behavior analysis using smart meter data for electricity consumers into the NILM technology will be a critical issue in the future. As a promising approach for accommodating the high penetration of DERs in power systems, the concept of TE and its application in electricity distribution markets are presented in this chapter. In TE systems, load forecasting techniques are crucial for both the optimal scheduling of DERs and the setting of bid parameters for consumers. Accurate load forecasting enables a better match between volatile renewable generations and load demand, thus this can significantly improve the localized accommodation of DERs. Meanwhile, due to the penalties that are applied to deviations between the winning bids and the real-time load demand, an accurate forecasting of electricity consumptions is beneficial to the maximization of transaction benefits for TE system participants. As an essential prerequisite for the application of load forecasting in future digital grid systems, the establishment of TE systems in electricity distribution networks is thus of great importance. Electricity transactions in a distribution system are featured by the participation of prosumers with small-scale DERs and individual end-users, the high penetration of renewable generations, as well as the preference of a set-and-forget bidding mechanism. Consequently, existing marginal cost based market mechanism will face failure under the above circumstance and it calls for the design of innovative distribution market mechanisms. An APM mechanism for transactions among DERs with zero marginal costs is introduced in this chapter which can overcome the challenges in distribution markets. Under the APM mechanism, honesty is proved to be a dominant strategy for participants and this enables the proposed mechanism to develop into a set-and-forget bidding market. Meanwhile, the problem of always bidding at a zero price by renewable generators in existing markets can be avoided. The APM mechanism is also compatible with the nodal pricing system. Merits of the nodal pricing can still be retained when adopting APM. Furthermore, the decentralized market mechanism for DER transactions is also discussed. The ADMM based approaches that are widely adopted for the decentralization of electricity distribution markets are elaborated. Key issues in the design of decentralized market mechanisms are also analyzed, including the convergence of iterative computation, the preservation of privacy, and the scalability for integrating more users. In addition, as a critical factor for maintaining the local balance of energy demand and supply, load forecasting is one of the core components in DSM. The accuracy point load forecasting not only affects the energy cost of customers because of their limited computation and decision-making capability, but also influences customers’ net-load profiles and the distribution of reverse power flows generated by the DERs. Thus, the short-term scheduling of DERs in demand side can be regarded a typical

413

414

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

application of load forecasting in DSM. In this chapter, an introduction on energy management techniques for DERs in different DSM contexts is also provided, namely the short-term scheduling of DERs in buildings, microgrids, and VPPs. As for buildings, technological advances have transformed modern buildings from a static civil infrastructure to complex cyber-physical systems, where these advances include the integration of DERs and ESSs in building side, the building electrification, and the digitization and automation of buildings. BEMSs/HEMSs that are used for managing DERs in building environments are introduced in detail. In terms of the short-term scheduling of DERs in microgrids, state-of-the-art of both centralized and distributed approaches that are proposed in literatures for scheduling DERs in microgrids are reviewed. Notably, there is also resiliency-oriented energy management of DERS in microgrids, which aims at enhancing the self-energy supply capability of microgrids subjected to significant fault/disturbances or the islanded operation mode. Another introduced management strategy of DERs is the VPP. Although the concept of VPP is similar to microgrid to some extent, VPP can overcome the geographical limitation in microgrids with the help of communication networks. Meanwhile, VPP also has the advantage of a larger capacity if aggregates sufficient DERs are compared with a microgrid and this enables VPP to make profits through selling energy to the electricity market.

References [1] Climate Change Atmospheric Carbon Dioxide, 2020. Available from: ,https://www.climate.gov/ news-features/understanding-climate/climate-change-atmospheric-carbon-dioxide. (accessed 9.11.20). [2] Electricity Network Transformation Roadmap, 2017. Available from: ,https://www.energynetworks.com.au/projects/electricity-network-transformation-roadmap/. (accessed 9.11.20). [3] Clean Energy Regulator. About the renewable energy target, 2018. Available from: ,http://www. cleanenergyregulator.gov.au/RET. (accessed 9.11.20). [4] T. Bayar, 2018. Growth in world energy demand. Available from: ,https://www.powerengineeringint.com/articles/2018/03/growth-in-world-energy-demand-doubled-in-2017-says-iea.html.. [5] Y. Zheng, Z.Y. Dong, F.J. Luo, K. Meng, J. Qiu, K.P. Wong, Optimal allocation of energy storage system for risk mitigation of discos with high renewable penetrations, IEEE Trans. Power Syst. 29 (1) (2014) 212
220. [6] A. Mnatsakanyan, S.W. Kennedy, A novel demand response model with an application for a virtual power plant, IEEE Trans. Smart Grid 6 (1) (2015) 230
237. [7] N.F. Hubele, C. Cheng, Identification of seasonal short-term load forecasting models using statistical decision functions, IEEE Trans. Power Syst. 5 (1) (1990) 40
45. [8] P.J. Brockwell, R.A. Davis, Introduction to time series and forecasting, springer, 2016. [9] A.S. Ahmad, et al., A review on applications of ANN and SVM for building electrical energy consumption forecasting, Renew. Sustain. Energy Rev. 33 (2014) 102
109. [10] M. Tamimi, R. Egbert, Short term electric load forecasting via fuzzy neural collaboration, Electr. Power Syst. Res. 56 (3) (2000) 243
248. [11] M. Khashei, M. Bijari, A novel hybridization of artificial neural networks and ARIMA models for time series forecasting, Appl. Soft Comput. 11 (2) (2011) 2664
2675. [12] Smart-Grid Smart-City Customer Trial Data. Available from: ,https://data.gov.au/dataset/ds-dga4e21dea3-9b87-4610-94c7-15a8a77907ef/details?q 5 . (accessed 9.11.20).

Load forecasting in the short-term scheduling of DERs

[13] R. Zhang, Z.Y. Dong, Y. Xu, K. Meng, K.P. Wong, Short-term load forecasting of Australian National Electricity Market by an ensemble model of extreme learning machine, IET Gener, Transm. & Distrib. 7 (4) (2013) 391
397. [14] W. Kong, Z.Y. Dong, Y. Jia, D.J. Hill, Y. Xu, Y. Zhang, Short-term residential load forecasting based on LSTM recurrent neural network, IEEE Trans. Smart Grid 10 (1) (2019) 841
851. [15] B. Stephen, X. Tang, P.R. Harvey, S. Galloway, K.I. Jennett, Incorporating practice theory in subprofile models for short term aggregated residential load forecasting, IEEE Trans. Smart Grid 99 (2016) 1
8. [16] Y. Bengio, P. Simard, P. Frasconi, Learning long-term dependencies with gradient descent is difficult, IEEE Trans. Neural Netw. 5 (2) (1994) 157
166. [17] S. Hochreiter, Y. Bengio, P. Frasconi, J. Schmidhuber, Gradient flow in recurrent nets: the difficulty of learning long-term dependencies, A Field Guide to Dynamical Recurrent Neural Networks, IEEE Press, 2001. [18] F. Gers, J. Schmidhuber, F. Cummins, Learning to forget: continual prediction with LSTM, Neural Comput. 12 (10) (2000) 2451
2471. [19] R. Koenker, K.F. Hallock, Quantile regression, J. Econ Perspect. 15 (4) (2001) 143
156. [20] W. Kong, Z.Y. Dong, J. Ma, D.J. Hill, J. Zhao, F. Luo, An extensible approach for non-intrusive load disaggregation with smart meter data, IEEE Trans. Smart Grid 9 (4) (2018) 3362
3372. [21] W. Kong, Z.Y. Dong, B. Wang, J. Zhao, J. Huang, A practical solution for non-intrusive type II load monitoring based on deep learning and post-processing, IEEE Trans. Smart Grid (2019). 1-1. [22] W. Kong, Z.Y. Dong, D.J. Hill, F. Luo, Y. Xu, Short-term residential load forecasting based on resident behaviour learning, IEEE Trans. Power Syst. 33 (1) (2018) 1087
1088. [23] NSMP Business Requirements Work Group, Smart metering infrastructure minimum functionality specification, 2011, Available from: ,https://link.aemo.com.au/sites/wcl/smartmetering/Pages/ BRWG.aspx.. [24] O. Abrishambaf, F. Lezama, P. Faria, Z. Vale, Towards transactive energy systems: an analysis on current trends, Energy Strategy Rev. 26 (2019) 1
17. [25] T.G.A. Council, Grid Wise Architecture Council-Grid Wise Transactive Energy Framework. Version 1.0, 2015. [26] M. Hasheminamin, V.G. Agelidis, V. Salehi, R. Teodorescu, B. Hredzak, Index-based assessment of voltage rise and reverse power flow phenomena in a distribution feeder under high PV penetration, IEEE J. Photovolt. 5 (4) (2015) 1158
1168. [27] J. Yang, et al., A penalty scheme for mitigating uninstructed deviation of generation outputs from variable renewables in a distribution market, IEEE Trans. Smart Grid (2020). 1-1. [28] R. Morgan. Solar electric panels hot items in Queensland and South Australia. Available from: ,http://www.roymorgan.com/findings/7262-solar-energy-electric-panels-march-2017-201707 061419.. [29] S. Bahramirad, A. Khodaei, R. Masiello, Distribution markets, IEEE Power Eng. Mag. 14 (2) (2016) 102
106. [30] P. Cramton, Electricity market design, Oxf. Rev. Econ Policy 33 (4) (2017) 589
612. [31] A.J. Conejo, M. Carrión, J.M. Morales, Decision Making Under Uncertainty in Electricity Markets, Springer, NY, 2010. [32] B. Elliston, J. Riesz, I. MacGill, What cost for more renewables? The incremental cost of renewable generation
an Australian National Electricity Market case study, Renew. Energy 95 (2016) 127
139. [33] P. Pikk, M. Viiding, The dangers of marginal cost based electricity pricing, Balt. J. Econ. 13 (1) (2014) 49
62. [34] J. Yang, J. Zhao, J. Qiu, F. Wen, A distribution market clearing mechanism for renewable generation units with zero marginal costs, IEEE Trans. Ind. Inform. 15 (8) (2019) 4775
4787. [35] S. Boyd, Distributed optimization and statistical learning via the alternating direction method of multipliers, Found. Trendss Mach. Learn. 3 (1) (2010) 1
122. [36] E.L. Ratnam, S.R. Weller, C.M. Kellett, An optimization-based approach to scheduling residential battery storage with solar PV: assessing customer benefit, Renew. Energy 75 (2015) 123
134.

415

416

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

[37] Q. Peng, S.H. Low, Distributed optimal power flow algorithm for radial networks, I: Balanced single phase case, IEEE Trans. Smart Grid 9 (1) (2018) 111
121. [38] S. Chakrabarti, M. Kraning, E. Chu, R. Baldick S. Boyd, Security constrained optimal power flow via proximal message passing, in: Proceedings of the Clemson University Power Systems Conference, 11-14 March, Clemson, SC, USA, 2014, pp. 1
8. [39] M. Kraning, E. Chu, J. Lavaei, S. Boyd, Dynamic network energy management via proximal message passing, Found. Trends Syst. Control. 1 (2) (2014) 70
122. [40] J. Yang, Z. Dong, G. Che, F. Wen, C. Li, A fully distributed distribution market mechanism using ADMM, in: Proceedings of 2019 IEEE PES General Meeting, 4
8 August, Atlanta, GA, 2019, pp. 1
5. [41] J. Yang, Z.Y. Dong, F. Wen, G. Chen, Y. Qiao, A decentralized distribution market mechanism considering renewable generation units with zero marginal costs, IEEE Trans. Smart Grid 11 (2) (2020) 1724
1736. [42] P. Palensky, D. Dietrich, Demand side management: demand response, intelligent energy systems, and smart loads, IEEE Trans. Ind. Inform. 7 (3) (2011) 381
388. [43] S. Attia, Net Zero Energy Buildings (NZEB) Concepts, Frameworks and Roadmap for Project Analysis and Implementation, Butterworth Heinemann, 2017. [44] B. Nordman, K. Christensen, A. Meier, Think globally, distribute power locally: the promise of nanogrids, Computer 45 (9) (2012) 89
91. [45] Z. Zhuang, L. Won Cheol, S. Yoan, S. Kyung-Bin, An optimal power scheduling method for demand response in home energy management system, IEEE Trans. Smart Grid 4 (3) (2013) 1391
1400. [46] F. Luo, W. Kong, G. Ranzi, Z.Y. Dong, Optimal home energy management system with demand charge tariff and appliance operational dependencies, IEEE Trans. Smart Grid 11 (1) (2020) 4
14. [47] S. Wang, F. Luo, Z.Y. Dong, Z. Xu, Coordinated residential energy resource scheduling with human thermal comfort modelling and renewable uncertainties, IET Generation, Transm. & Distrib. 13 (10) (2019) 1768
1776. [48] V. Pradhan, V.S.K. Murthy Balijepalli, S.A. Khaparde, An effective model for demand response management systems of residential electricity consumers, IEEE Syst. J. 10 (2) (2016) 434
445. [49] J. Ma, H. Chen, L. Song, Y. Li, Residential load scheduling in smart grid: a cost efficiency perspective, IEEE Trans. Smart Grid (1-1)(2015). [50] S. Shao, M. Pipattanasomporn, S. Rahman, Grid integration of electric vehicles and demand response with customer choice, IEEE Trans. Smart Grid 3 (1) (2012) 543
550. [51] C. Zhi, W. Lei, Residential appliance DR energy management with electric privacy protection by online stochastic optimization, IEEE Trans. Smart Grid 4 (4) (2013) 1861
1869. [52] P. Scott, S. Thiebaux, M. Briel, P.V. Hentenryck, Residential demand response under uncertainty, in: Proceedings of International Conference on Principles and Practice of Constraint Programming, 16
20 September, Uppsala, Sweden, 2013, pp. 1
9. [53] S. Wang, F. Luo, Z.Y. Dong, G. Ranzi, Y. Qiao, Stochastic residential energy management system with vehicle-to-home and photovoltaic penetration,” in: Proceedings of 2018 IEEE PES General Meeting, 5
9 August, Portland, OR, 2018, pp. 1
5. [54] R.H. Lasseter, P. Paigi, Microgrid a conceptual solution, in: Proceedings of IEEE 35th Annual Power Electronics Specialists Conference, 20
25 June, Aachen, Germany, 2004, pp. 4285
4290. [55] About Microgrids. Available from: ,https://building-microgrid.lbl.gov/about-microgrids.. [56] F. Luo, Z. Xu, K. Meng, Z. Yang Dong, Optimal operation scheduling for microgrid with high penetrations of solar power and thermostatically controlled loads, Sci. Technol. Built Environ. 22 (6) (2016) 666
673. [57] A.K. Basu, A. Bhattacharya, S. Chowdhury, S.P. Chowdhury, Planned scheduling for economic power sharing in a CHP-based micro-grid, IEEE Trans. Power Syst. 27 (1) (2012) 30
38. [58] M.N. Mojdehi, P.K. Ghosh, Distributed generations scheduling in micro-grid considering CO2 emission cost, in: Proceedings of 2012 IEEE Power and Energy Society General Meeting, 22
26 July, San Diego, CA, 2012, pp. 1
5.

Load forecasting in the short-term scheduling of DERs

[59] N.M.M. Razali, A.H. Hashim, Profit-based optimal generation scheduling of a microgrid,” in: Proceedings of the 4th International Power Engineering and Optimization Conference, 23
24 June, Shah Alam, Malaysia, 2010, pp. 232
237. [60] N. Rahbari-Asr, Y. Zhang, M.Y. Chow, Cooperative distributed scheduling for storage devices in microgrids using dynamic KKT multipliers and consensus networks, in: Proceedings of 2015 IEEE Power and Energy Society General Meeting, 26
30 July, Denver, CO, 2015, pp. 1
5. [61] A. Khodaei, Resiliency-oriented microgrid optimal scheduling, IEEE Trans. Smart Grid 5 (4) (2014) 1584
1591. [62] A. Gholami, T. Shekari, F. Aminifar, M. Shahidehpour, Microgrid scheduling with uncertainty: the quest for resilience, IEEE Trans. Smart Grid 7 (6) (2016) 2849
2858. [63] H. Saboori, M. Mohammadi, R. Taghe, Virtual power plant (VPP), definition, concept, components and types,” in: Proceedings of 2011 IEEE Asia-Pacific Power and Energy Engineering Conference, 25
28 March, Wuhan, China, 2011, pp. 1
4. [64] E. Mashhour, S.M. Moghaddas-Tafreshi, Bidding strategy of virtual power plant for participating in energy and spinning reserve markets—Part I: problem formulation, IEEE Trans. Power Syst. 26 (2) (2011) 949
956.

417

CHAPTER 13

Conclusions and key findings of optimal operation and planning of distributed energy resources in the context of local integrated energy systems Giorgio Graditi and Marialaura Di Somma

Department of Energy Technologies and Renewable Sources of ENEA, Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Rome, Italy

In recent years, depletion of fossil energy resources and global warming problems have prompted worldwide awareness about sustainability of energy supply. In such a context, distributed energy resources (DER) consisting of distributed generation, demand response (DR) and distributed storage have been recognized as a promising option for decarbonization of energy supply and are expected to be widely spread in the next years thanks to the supportive policies and financial incentives. In fact, they present numerous economic and environmental benefits, related to the possibility to integrate and combine several energy resources, including renewables, to recover waste heat from power generation for thermal purposes in buildings, as well as to activate DR programs. From the point of view of the grid, DER can provide ancillary services, such as voltage support and regulation services, whereas, from the point of view of the end-users, DER can improve local reliability, and reduce costs of energy supply and greenhouse emissions. DER organized in local integrated energy systems may offer economically interesting conditions to boost local generation of electricity and heat, and promote local DR, by providing both opportunities and challenges to end-users, utilities, and other participants in distribution systems. These local energy systems are well-placed to meet local energy needs, reduce the need for transmission infrastructure, and bring people together to achieve common goals for well-being. DER within local integrated energy systems can also be employed to support the grid by providing ancillary services and flexibility through different products such as DR, congestion relief, local energy markets, etc. Moreover, these local dimensions also foster consumer engagement while also providing the services needed to enable it across a wider region. Concerns for Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning DOI: https://doi.org/10.1016/B978-0-12-823899-8.00013-3

r 2021 Elsevier Inc. All rights reserved.

419

420

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

resilience and public acceptance of renewable energy sources (RES) and networks have motivated a growing interest in decentralized approaches in recent years. As compared to traditional centralized energy systems, decentralized local energy systems promote an enhanced focus on local security of energy supply, as well as an effective integration of energy systems and carriers through a variety of local generation of electricity and heat, flexible demand, energy storages and electric mobility. However, in order to achieve the expected potentials of DER and enable their local integration, the coordinated operation of generating units together with controllable loads, and storages in all their forms is crucial. The optimal operation of DER is a complex task, by presenting specific issues such as the unbalance between supply and demand sides, due to the typical instantaneous variation of user energy demand, the limited operation flexibility of certain technologies within the system to deal with the fluctuation in uncertain loads, and the presence of renewables with intrinsic stochastic behavior. In the context of local integrated energy systems, the complexity is also linked to their optimal planning due to the presence of different energy carriers (e.g., electricity, gas, heat, cooling, etc.), which need to be optimally coordinated and combined to achieve decarbonization and energy savings and providing flexibility services to the neighbor systems, while also satisfying the user multienergy demand. In addition, the integration of DER in the electricity markets and local energy markets is also a challenging task, which needs to be addressed with a certain urgency especially with the deployment of emerging energy communities in Europe and worldwide. Another important issue is that in operation and planning of multiple DER, the economic optimization alone is not sufficient to guarantee the long-run sustainability of local integrated energy systems, and environmental/sustainability priorities should be also taken into account in the optimization models. In fact, the optimal planning of DER inherently involves multiple and conflicting objectives. For instance, the interest of planners/operators in achieving system configurations with operation strategies with the lowest costs might conflict with the interest of energy legislations, such as the European Union (EU) ones in increasing sustainability of energy supply, which can be attained through a rational use of energy resources, by reducing waste of fossil energy and environmental impacts in terms of CO2 emissions. Therefore, a multiobjective approach should be used in the optimal operation and planning of DER to help in identifying balancing solutions to promote stakeholders’ participation in the decision-making process and facilitate collective decisions. Moreover, in the future energy landscape, with the advent of energy communities, the energy trading between neighbors will become more and more important, and it is needed to understand their benefits and impacts on the operation and planning of distribution networks.

Conclusions and key findings of optimal operation and planning

In order to deal with all these issues, the following topics are comprehensively discussed throughout the book: • Integrating different types of DER including electrical and thermal distributed generation, DR, electric vehicles (EVs), storage devices and various RES in local integrated energy systems, while also investigating the potential benefits and impacts for the larger neighbor system. • Proposing operation optimization models for short-term performance and scheduling of DER in the context of local integrated systems also through a multiobjective approach for their economic and environmental sustainability. • Fostering the integration of DER in the electricity markets through the concepts of DER aggregation. • Addressing the challenges of emerging paradigms as energy communities and energy blockchain applications in the current and future energy landscape. • Assessing and modeling the uncertainties of renewable resources and intermittent load in short-term decision-making of local integrated energy systems. In detail, the book investigates in a comprehensive way the most recent research and policy developments on the issue of optimal operation and planning of DER in the context of local integrated energy systems in the presence of multiple energy carriers and vectors. This assessment is carried out by analyzing impacts and benefits not only at the local level, but also on the distribution network and larger systems, through DER aggregation concepts, functional architectures, and emerging paradigms as energy communities. The interactions among various energy carriers in the local energy systems are investigated in the optimization models. Moreover, in order to foster an effective implementation and deployment of local integrated energy systems while also investigating the benefits and impacts related to the DER integration in the energy supply, multiobjective optimization models are proposed by considering both economic and sustainability-related objectives. The book supports readers finding innovative solutions and detailed insights for the operation and planning of DER while fostering research advances to the state-of-theart on this topic, by presenting planning approaches, methodologies, critical assessments, as well as proper optimization models and algorithms for DER in the context of local integrated energy systems. The proposed optimization frameworks are scalable and flexible for adaptation to a number of real contexts thanks to the wide variety of generation, conversion and storage technologies considered, and the exploitation of demand side flexibility and emerging technologies, as well as through the general mathematical formulations established. These frameworks thus represent valid tools to provide support in decision-making process for DER operation and planning. Uncertainties of RES generation and loads in optimal DER operation and planning are also addressed, and the role of prosumers is investigated through the concepts of neighborhood energy trading and blockchain technologies.

421

422

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

A critical overview of DER technologies and the related benefits and challenges in the context of local integrated energy systems is proposed in Chapter 1, with the analysis of the relevant literature concerning the deployment of local solutions to generate, store and manage energy at small scale and micro-scale. Grid-related aspects and solutions for optimal grid operation through DER are also discussed along with the various emergent aspects referring to local energy systems, markets and energy communities, which are gaining relevance in the present and future context of energy transition. It seems evident that decentralization is a viable and promising alternative for the future development of the power system. It is however important to keep in mind that decentralization is not an ultimate goal itself, but rather a way to deal with growing complexity of the power system. This means in principle that transition to the decentralized system will follow the overall path of changes in the power system corresponding to Europe’s climate ambitions. To address this key topic, the main architectures and concepts for smart decentralized energy systems are discussed in Chapter 2, through the critical analysis of recent documents such as Pan-European roadmaps (ETIP-SNET) and scenarios (TYNDP2020), results of R&D projects and regulatory documents (“Clean Energy Package for all Europeans”). It is clear that in the context of local integrated energy systems, due to the presence of various energy vectors and components, the dependencies are an important factor as they can have impacts on various aspects of a multienergy system, from operation to planning. This issue is not limited to local system operation but also affects the sector coupling and long-term planning. The interdependencies occurring among vectors and systems are addressed and modeled in Chapter 3 to create flexibility services to the larger system. A holistic view of the interdependencies among energy carriers in the local system with DER is discussed, and various DER, as well as the modeling approaches of carrier dependencies that are caused by their presence in the system, are presented. Several benefits can be obtained through the synergies of interdependent sectors. In the analyzed case study, it is shown that the interdependency on various layers of the multienergy system not only brings extra flexibility for the system operator to use in its operational decision-making process but also opens new business opportunities for new roles and players in multienergy system modeling who can address the interdependencies in a more beneficial way for the whole energy sector. In order to foster both the short- and long-run sustainability of local integrated energy systems, a multiobjective optimization framework is presented in Chapter 4 to attain the optimized operation strategies of DER within local integrated energy systems to obtain rational use of energy resources, by considering both economic and environmental/sustainability priorities. With the general mathematical models established, the optimization frameworks developed are flexible and scalable for potential adaptation to real contexts, also considering the wide variety of generation, conversion and storage technologies modeled as DER. The effectiveness of the optimization

Conclusions and key findings of optimal operation and planning

frameworks is tested through different case studies, and it is demonstrated that they can represent powerful tools for both economic and sustainability/environmental objectives, by providing support to decision- and policy-makers in the quantification of the benefits derived by local integrated energy systems and the optimized management of local energy resources to foster efficient and rational use of available energy. The attention is naturally focused on the impacts of DER on larger system, by discussing in Chapter 5 the impact of DER on operation and planning of distribution networks through emerging paradigms as neighborhood energy trading and renewable energy communities. In fact, the recent regulatory framework in Europe is opening the possibility for end-users to directly transact with their neighbors, in the framework of energy communities, which, equipped with both distributed generation units and storage, can favor the local balance between production and consumption during the day, with the advantages for the network of improved efficiency on the one hand and reduced use on the other. Their impact on the operation of distribution networks is mainly associated with the optimal scheduling of the generation and storage units available inside the local energy communities. Their impact is also important for the future planning of networks reinforcement since the increased adoption of neighborhood energy trading mechanisms allows to defer grid investments. The expected increase in the installation of DER at the distribution levels along with the replacement of larger conventional units at transmission are forcing the electricity markets to evolve. The current status of EU regulations together with a country-specific analysis to know what type of DER can provide a certain flexibility service in which country has been assessed in Chapter 6. The massive deployment of DER is also going to change the needs of reserves shortly (2030 horizon) and a methodology to estimate these reserves has been drafted. Additionally, different coordination schemes for the provision of flexibility services have been analyzed from an economic perspective. Local energy markets are also discussed as an opportunity to trade flexibility among the different participants in an economically efficient way, giving a key role to the consumers, who become active consumers. In the policy framework supporting the emerging roles of active consumers in the future energy landscape, the blockchain technology plays a key role by fostering DR and the aggregation of loads and generators, helping the creation of energy communities and the integration of EVs in the power grid, and providing a more transparent and traceable relation between service providers and customers. The role of blockchain technology in the energy sector has been examined in Chapter 7 with a particular interest in DR and vehicle-to-grid (V2G) applications. Energy blockchain applications are deserving a growing attention in latest years, since blockchain architectures, on one hand, provide transparency and solve the information asymmetry problem, and on the other, provide disintermediation. In this way, the dream of an energy market closer to end-users becomes a reality, although, the regulatory framework is not clear yet, especially for

423

424

Distributed Energy Resources in Local Integrated Energy Systems: Optimal Operation and Planning

what concerns the tokenization of assets related to investments in energy projects and energy services. The future challenges of the energy blockchain comprises several aspects to be still discussed and problems to be solved, specifically with regards to the integration with the existing V2G technology and DR scheme. Many research projects and pilot installations are currently testing the potentiality of the blockchain as a disruptive technology for power system for fostering the development of RES and the transition to cleaner and greener configurations. After analyzing the DER issues from the system perspective, the attention is focused on the specific component/technology’s optimal scheduling. In detail, the optimal management issue of battery energy storage systems is addressed in Chapter 8 in the context of nanogrids for a virtual nonsumer community, which is a community that can become essentially self-sufficient in terms of net kWhs purchased from the grid, or in some cases, become a net generator of energy. Achieving growing autonomy, sustainability and efficiency of local energy systems with respect to energy needs is indeed a strategic goal now consolidated and the storage systems are called to become more and more advanced in terms of technology thanks to the “smart grid ready” functions, ready to “work” in harmony with an interconnected, digitalized electrical system, open to the contributions of an increasing number of active consumers. The attention is focused on a nonsumer energy community consisting of several DC nanogrids with the goal to become essentially self-sufficient in terms of net kWhs purchased from the grid using several storage systems integrated in the nanogrids and opportunely managed and controlled implementing a proper optimization management model and the DC Bus Signaling (DBC) control logic. The numerical results obtained from a case study show the effectiveness to satisfy the power profile valuated as output of the proposed optimization management model using the illustrated DBS control logic. The DR role is investigated with the aim to enhance flexibility of local integrated energy systems in Chapter 9, and an energy management framework is proposed to tackle the DER’s uncertainties and enhancing the flexibility of the entire network by adopting the effects of DR programs, as well as the effects of electrical energy storage devices. In detail, a centralized framework is developed for determining the operating points of a multicarrier energy system and improving the flexibility of the local energy systems, considering the price-based DR programs. The mentioned centralized framework can provide the desired solution for the energy vector system and energy communities, considering the flexibility from the consumer engagement in the DR programs. The integration of EVs in smart distribution grids is discussed in Chapter 10. The wide diffusion of EVs represents a great opportunity, not only to make sustainable the mobility sector, but also to work in synergy with distribution grids, since they can work as distributed storage to improve the quality of the power and increase the

Conclusions and key findings of optimal operation and planning

percentage of RES that they can embed. This opportunity is introduced by smart and bidirectional charging infrastructures allowing V2G functionalities for using EVs as DER and provide ancillary services to smart distribution grids or towards the transmission systems for frequency regulation. Considering that the uncertain and nonpredictable behavior of RES and intermittent loads are crucial aspects to be considered in the short-term scheduling of DER, the last two chapters of the book are dedicated to these topics. The stochastic nature of renewables imposes great challenges to the emerging energy grid. This inherent characteristic of RES creates many technical problems with respect to settings of protection systems, voltage drops, congestion management and other quality attributes of power systems. As distributed renewable resources increase, their intermittent and variable nature, if not compensated, adds more uncertainty to the system, which adversely affects the reliability of the interconnected systems. Therefore, it is clear that RES uncertainties influence the impact of how DER as emerging technologies, will be scheduled in the short term, to provide the available flexibilities as an important contribution to the balancing of the system at all times. Chapter 11 addresses the issue of RES uncertainties and assesses how DER operating strategies can soften their effect on the grid. An analytical approach is proposed to assess the uncertainties of active distribution networks and the effect on the short-term scheduling of operators. The effectiveness of the presented assessment is validated using extensive numerical studies on a real grid of Cyprus with high RES penetration. Finally, Chapter 12 discusses new trends in load forecasting techniques under the smart grid environment. Due to the rapid growth of RES generation and the ever-increasing flexibility of load demand in current energy systems, the focus of load forecasting has gradually shifted from aggregated system level forecasting to the estimation of short-term individual load consumptions. In particular, as there is a much higher level of volatility hidden in individual loads, traditional load forecasting methodologies are no longer applicable in individual load forecasting. Considering that the load profile of an individual household is generally determined by activities of residents, the key to carry out accurate load forecasting is to capture the temporal correlations between the underlying activities from the observed load profiles of the past. Then, methods that are more capable of learning intrinsic patterns and correlation from history load profiles are desired, and for this purpose, both recurrent neural network and long short-term memory are good options. Besides, the integration of behavior analysis in the individual load forecasting is proved to be an effective way to improve the forecasting accuracy.

425

Index Note: Page numbers followed by “f” and “t” refer to figures and tables, respectively.

A

B

Absorption chiller model, 297 Absorption chillers (ACs), 283 285 AC microgrids (ACMGs), 244 Active consumers, 17 18, 183 Active Distribution System Management (ADSM), 40 Adaptive frequency containment control, 48 49 Advanced Meter Infrastructure (AMI), 393 “AGCControl” functions, 376 Agency for Cooperation of Energy Regulators (ACER), 57, 185 Agents, 32 Agent-based method, 80 82 Aggregation, 183 184 Aggregator, 13 14, 200 AI. See Artificial intelligence (AI) model AirBnb of energy, 219 220 Alternating current (AC), 316 charging infrastructures, 320 321 Alternating direction method of multipliers (ADMM), 129, 135, 404 405 convergence, 136 Ancillary services market, 263 programs, 290 Artificial intelligence (AI) model, 392 393 Artificial neural networks (ANN) model, 392 393 Auto regressive (AR) model, 392 393 Auto Regressive Integrated Moving Average (ARIMA) model, 392 393 Automatic frequency restoration control (aFRC), 41 42 Automatic frequency restoration reserves (aFRR), 177, 355 356 Automatic generation control (AGC), 37 38 Automatic voltage regulator (AVR), 37 Automation, 407 Auxiliary boilers, 68 Auxiliary gas-fired boiler (AB), 110 Average pricing market (APM), 402

Balance responsible parties (BRPs), 52, 198 201 Balance restoration control (BRC), 47 Balance steering control (BSC), 48 Balancing or frequency control, 177 Battery electric vehicle (BEV), 316 Battery energy storage (BES), 128 129 Battery energy storage systems (BESSs), 54, 407 Battery Management System (BMS), 236 Bidding type, 202 Bids formats, 201 Biomass boiler, 97 Bitcoin, 209 212 Black start, 179, 181, 267 Blackout. See Power—outage Blockchain, 208 210 applications for P2P, DR and V2G, 215 217 consensus algorithms, 210 213 energy blockchain, 218 222 laboratory setup for energy blockchain testing, 223 228 SC, 213 215 Blockchain for Renewable Energies research project (BLORIN research project), 208, 221 smart contracts in, 226 228 Boiler model, 295 296 Box-and-whisker plot, 379 Boxplot, 379 Brooklyn Microgrid project, 128 129, 219 Budgetary interdependency, 66 67 Building/home energy management systems (BEMSs/HEMSs), 407 408 Byzantine Fault Tolerance (BTF), 210 211

C C/sells, 40 Capacity constraint, 97 credit, 353 market programs, 290

427

428

Index

Carnot factor, 94 95, 100 Carrier dependency, 64 65 “Casper” protocol, 212 Cell(s), 32 controller, 50 Central Slave, 256 Centralized DER scheduling in microgrids, 410 Chaincodes, 214 215 Charging infrastructures planning, 332 333 Citizen Energy Community (CEC), 20, 58, 208 “Clean Energy for all Europeans” framework, 128 129 Clean Energy Package, 20 for all Europeans, 182 184 Clearing rule, 202 Collective self-consumption (CSC), 18, 261 Combined Cooling Heat and Power (CCHP), 68 69, 93 94 Combined heat and power systems (CHP systems), 36, 67 68, 93 94, 180, 283, 295 Commercial VPPs (CVPPs), 412 Community Energy Provider (CEP), 261 Complex system method, 82 Compressed-air energy storage system (CAES system), 235, 283 285 Congestion management, 39, 177 178, 181 Consensus algorithms, 210 213 Consumers, 282 Consumption behavior-driven household load forecasting, 398 399 Continental Europe (CE), 188 189 Conventional generators, 180 Convertors, 67 Coordinated charging, 341 Coordination scheme (CS), 193 Cost-benefit analysis (CBA), 194 of market participation of DERs, 194 197 Costing of NNSs, 157 Coupling model of components and services, 78 of local energy systems, 78 80 energy hub method, 78 80 energy network method, 80 large-scale coupling, 80 82 Critical peak pricing (CPP), 287, 289 Cumulative distribution functions (CDF), 152 153 Curtailable IDR program, 291, 302 303

D Day-ahead scheduling, 130 131 DC bus signaling (DBS), 242, 253, 424 DC microgrids (DCMGs), 244 Decentralization, 32 decentralized market mechanism for DER transactions, 403 405 in European future scenarios, 34 in European R&D projects, 34 35 and markets, 54 56 operation and control of decentralized system, 39 pros and cons, 35 roles and responsibilities, 57 58 Decentralized architecture, 36 41 control decentralized system, 40 41 level of decentralization, 36 40 Degree of centralization, 22 23 DELTA H2020 project, 216 217 Demand bidding/buyback programs, 290 Demand response (DR), 1 2, 5 6, 148 149, 155, 183, 218, 281 282, 350 351, 419 challenges of using blockchain technology for DR applications, 221 222 comprehensive assessment of DR programs, 287 291 programs for local energy systems, 286 291 models, 299 Demand side management (DSM), 2, 391 392 Dependency categories, 65 66 Digitization, 407 Direct current (DC), 316 charging infrastructures, 322 nanogrid, 244 Direct load control (DLC), 289 291 Dispersed generating systems (DGs), 283 285 Distributed approach for day-ahead scheduling of LEC, 129 137 distributed optimization model formulation, 132 137 Distributed DER scheduling in microgrids, 410 Distributed energy resources (DER), 1 7, 36, 92, 149, 176, 281 282, 349, 391, 419 barriers to market access of, 187 benefits of DER on power system grids, 360 363 combined production of different energy carriers, 4 5

Index

current status of DERs as flexibility providers, 186 demand response, 5 6 distributed generation based on different energy sources, 3 4 distributed storage, 6 7 emergent paradigms and solutions, 16 26 flexibility needs in power systems, 188 192 grid side aspects, 7 16 local energy markets, 198 202 in local integrated energy system, 97 99 market value of flexibility in distribution system, 193 197 modeling of DER in local energy community, 109 111 as providers of flexibility services, 176 181 characterization, 179 181 products and services for voltage and frequency control, 176 179 regulatory framework, 181 187 renewables uncertainties in short-term scheduling of DER, 370 384 grid system under investigation, 371 374 methodology, 370 371 operational strategies, 374 377 scenario under study, 377 simulation case results, 377 384 trans-active energy systems with, 399 405 Distributed generation (DG), 1 2, 148 149, 183, 286 based on different energy sources, 3 4 Distributed Ledger Technology (DLT), 208 Distributed method, 332 Distributed multi-energy systems (DMES), 4 Distributed storage (DS), 1 2, 6 7 Distribution market mechanism for DERs with zero marginal costs, 399 403 Distribution networks distributed approach for day-ahead scheduling of LEC, 129 137 implementation and numerical tests, 138 147 scalability of distributed approach, 141 145 scenario considering uncertainties on energy generation and consumption, 145 147 planning model, 148 160 case studies and analysis of results, 160 169 costing of NNSs, 157 NET modeling, 156 157

nonnetwork solutions modeling, 154 155 planning with neighborhood energy trading, 151 problem formulation, 157 159 risk-managed planning, 150 151 solution strategy, 159 160 uncertainties modeling, 151 153 Distribution system operator (DSO), 7 8, 32, 176, 182 183, 201, 216, 330, 355 Domestic hot water (DHW), 94 95 “dPBorder” function, 377 DSNS, 150 151 index, 160 Dual active bridge (DAB), 248 Dual Half-Bridge (DHB), 248 Dynamic probabilistic household load forecasting, 395 397 Dynamic reactive power, 178

E Eco-environmental optimization of local energy community in United States, 113 119 case study results, 114 119 input data, 113 114 Eco-exergetic operation optimization of local integrated energy system, 103 107 case study results, 104 107 input data, 103 104 Economic interdependency, 66 67 Economic objective, 100 ELECTRA cell, 45 ELECTRA Web-of-Cells control concept, 45 46 Electric heat pump (EHP), 283, 296 297 Electric heaters (EHs), 289 290 Electric Power Research Institute (EPRI), 33 Electric vehicles (EVs), 3, 69 71, 180, 217, 316, 350, 408, 421 characteristics, 316 320 and charging infrastructures, 316 322 EV supply equipment (EVSE), 320 high power DC charging infrastructures, 322 integration of electric vehicles in smart distribution grids, 322 333 impact of charging infrastructures on distribution grids, 325 332 planning of charging infrastructures, 332 333 low power AC charging infrastructures, 320 321

429

430

Index

Electric vehicles (EVs) (Continued) parking lots, 69 71 synergies between RES and, 342 344 vehicle-to-grid, 333 344 Electrical energy storage (EES), 281 282, 297 298 Electrical Energy Trading, 216 Electrical heater model, 296 Electrical storage, 180 Electricity, 233 and gas network, 71 73 and hydrogen, 73 74 Electricity gas heating/cooling systems, 74 75 Electricity gas hydrogen, in transportation sector, 75 76 Electricity billing procedure, 131 132 Electricity market directive II (EMDII), 259 261 Electricity network codes and guidelines, 185 186 Electricity shifting potential (ESP), 14 15 Electrification, 406 407 Electrochemical capacitors. See Supercapacitors Emergency demand response programs, 290 Energy carriers, 64 65 communities, 95 96, 208, 245 blockchain for, 227 228 paradigm, 19 24 management framework for DER integrated distribution networks, 292 304 modeling of energy balances, 99 100, 111 112 network integration among, 11 12 method, 80 storage, 68 69 trading projects, 219 220 transaction between hubs, 299 transition, 3 Energy Balancing Guideline (EBGL), 177 Energy blockchain, 218 222. See also Blockchain challenges of using blockchain technology for DR and V2G applications, 221 222 future applications, 227 228 laboratory setup for energy blockchain testing, 223 228 peer-to-peer energy exchanges among prosumers, 218 221 Energy hubs (EH), 4, 78 80

Energy management system (EMS), 129 Energy not supplied (ENS), 293 Energy storage systems (ESSs), 148 149, 216, 265, 334, 350, 406. See also Integrated energy systems (IES) as distributed flexibility, 234 239 flexibility in distribution grid, 234 235 flexibility services, 238 239 technologies, 235 238 for grid ancillary service, 263 267 ancillary services market, 263 potential benefits of using energy storage to provide ancillary services, 263 267 in nanogrid, 239 258 Enterprise service buses (ESBs), 351 352 Environmental objective, 102 Environment-based energy, 3 Ether, 214 Ethereum Virtual Machine (EVM), 214 European Commission (EC), 182 European green deal, 184 185 European legislator, 208 European Network of Transmission System Operators for Electricity (ENTSO-E), 33 European power industry, 33 European regulatory context, 182 186 clean energy package for all Europeans, 182 184 electricity network codes and guidelines, 185 186 European green deal, 184 185 European Technology & Innovation Platforms (ETIPs), 17 18 European Technology & Innovation Platforms for Smart Networks for Energy Transition (ETIP-SNET), 34 European Union (EU), 16, 182, 259 261 “Event Location” function, 377 Event tree analysis (ETA), 152 153 Exergetic objective, 100 102 Exergy, 94 95 efficiency, 102 input rate of biomass, 101 of natural gas, 101 Extended range electric vehicle, 317 External dependencies, 65 information/communication, 77 78

Index

multienergy demand, 76 77 in smart local system, 76 78 Extreme Learning Machine (ELM), 394 395

F Fault tree analysis (FTA), 152 153 Flexi User, 198 199 Flexibility, 64 65, 281 282, 366 368 activations, 41 assessment of local energy systems in presence of ESS and DR programs, 292 in distribution grid, 234 235 estimation of future needs of reserves in power systems with high shares of DERs, 190 192 market beneficiaries, 193 needs in power systems, 188 192 in power and energy systems, 15 practices in estimation of flexibility requirements, 188 190 frequency control reserves, 188 189 voltage control reserves, 190 providers characterization of DERs as, 179 181 status of DERs as, 186 services, 176 Flexible Heat and Power project (FHP project), 54 Forgers, 211 212 Forward backward approach, 150 FP7 ELECTRA IRP project, 44 45 Fractal Grid, 40 Frequency nadir, 357 358 regulation, 263 265, 341 restoration control effectivity, 356 zenith, 357 Frequency containment control (FCC), 41 42 Frequency containment reserves (FCR), 177, 188 189, 355 Frequency control reserves, 188 192 frequency containment reserves, 188 189 frequency restoration reserves, 189 replacement reserves, 189 services, 181 Frequency Restoration Reserve (FRR), 177, 189, 355 356 Fuel-cell electric vehicle, 317

Functional interdependency, 66 67 Future system grid projection, 350 352

G Gas energy systems, 71 76 Gaussian mixture model (GMM), 152 Genetic algorithm (GA), 285 Geographic information systems (GIS), 330 Grid. See also Nanogrids grid-link, 39 grid-secure activations for ancillary services, 41 45 power, 106 side aspects, 7 16 analysis and optimization of grid operation with local energy systems, 12 14 evolution of grid connection issues and standards, 7 9 integration among energy networks, 11 12 microgrids and local energy networks, 9 11 provision of grid services, 14 16 stabilization, 220 system under investigation, 371 374 Grid-to-Vehicle (G2V) mode, 217, 334 Grid4EU (EU FP7 project), 34 35 Gridable EVs (GEVs), 333 334

H Harmonics, 359 Hash functions, 209 Heating/cooling, 74 75 High power DC charging infrastructures, 322 High Slave, 256 High voltage (HV), 359 360 High-impact low-probability (HILP), 351 352 Hybrid electric vehicle (HEV), 317 Hybrid energy systems, 64 Hybridization degrees (HD), 317 Hydrogen, 73 76 Hyperledger Fabric, 214 215

I IDR programs for multicarrier energy systems, 291 IEEE 13-bus radial feeder, 163 168 IEEE 33-bus radial feeder, 168 169 Incentive-based demand response programs, 289 291

431

432

Index

Independent system operator (ISO), 287 Inertia, 178 control, 49 emulation, 266 Inertial response, 178 179 Information and Communication Technologies (ICT), 2, 64, 195, 208, 233, 351 352 Information/communication, 77 78 Insulated Gate Bipolar Transistor (IGBT), 359 360 Integrated energy systems (IES), 4, 64, 92. See also Distributed energy resources (DER) operation optimization of multiple integrated energy systems in local energy community, 108 119 eco-environmental optimization of local energy community, 113 119 under study and mathematical formulation, 108 112 Interdependency modeling, 78 82 coupling model of components and services, 78 coupling model of local energy systems, 78 80 Interdependent MES model, case study on, 82 84 Intergovernmental Panel on Climate Change (IPCC), 65 Internal Combustion Engine vehicles (ICE vehicles), 316 Internal dependencies, 65 Internal Electricity Market (IEM), 55 Internal energy market (IEM), 185 Internal multicarrier dependency in smart local system, 66 76 components of local energy systems, 67 71 electricity, 73 76 gas energy systems, 71 73 heating/cooling, 74 75 hydrogen, 73 74 transportation, 75 76 International Grid Control Cooperation (IGCC), 41 42 Interruptible/curtailable services, 290

J Jointly acting renewable self-consumers, 18

K Key Performance Indicators (KPIs), 348, 371

L Large-scale coupling, 80 82 agent-based method, 80 82 complex system method, 82 Lead acid (LA), 235 Li-Ion battery model, 246 248 Li-Ion DC/DC converter, 248 Linear programming (LP), 285 286 LINK project, 39 Lithium batteries, 236 Lithium-ion cells, 236 Lithium/ion batteries (Li-Ion battery), 236 237 Load curtailment devices, 180 Load duration curves (LDCs), 151 152 Load forecasting, 391 392 consumption behavior-driven household load forecasting, 398 399 dynamic probabilistic household load forecasting, 395 397 for individual energy customers, 392 395 new trends in, 392 399 short-term scheduling of DERs in demand side, 406 412 trans-active energy systems with DERs, 399 405 Load shifting devices, 180 Load tap changing (LTC), 358 Local energy community (LEC), 128, 131f distributed approach for day-ahead scheduling, 129 137 Local energy markets (LEMs), 18 19, 198 202 components of functional LEMs, 201 202 Local energy networks, 9 11 Local energy systems, 281 282. See also Integrated energy systems (IES); Multi-energy systems (MES) analysis and optimization of grid operation with, 12 14 components of, 67 71 coupling model of, 78 80 demand response programs for, 286 291 simulation results, 304 310 Local integrated energy systems, 419 420 Long short-term memory (LSTM), 395 Low frequency oscillation, 327 Low power AC charging infrastructures, 320 321 Low Slave, 256

Index

Low voltage (LV), 130, 358 grids, 34 35 Low-impact high-probability events (LIHP), 363

M Major event days (MED), 364 Manual Frequency Restoration Control (mFRC), 41 42 Manually frequency restoration reserves (mFRR), 177, 355 356 Marginal pricing (MP), 202 Market design, 202 interdependency, 66 67 models, 39 value of flexibility in distribution system, 193 197 cost-benefit analysis of market participation of DERs, 194 197 flexibility market beneficiaries, 193 Market clearing price (MCP), 400 401 Master, 255 256 Master Absorbs, 255 256 Master Injection, 256 Matching, 202 Maximum Power Point Trekking mode (MPPT mode), 245 Maximum profit electricity reduction (MPER), 14 15 Mean absolute percentage error (MAPE), 394 395 Mean square error (MSE), 397 Medium voltage (MV), 358 359 Micro combined heat and power (mCHP), 283 285 Microgrids, 9 11, 218, 350 351 MicroSource (MS), 244 Miners, 211 Mixed-integer linear programming (MILP), 93 94, 283 Mixed-integer nonlinear programming model (MINLP model), 283 285 Mixed-integer programming (MIP), 333 Modified PSO algorithm (MPSO algorithm), 160 Modularity of DG plant, 3 Monte Carlo analyses, 12 Moving Average (MA) model, 392 393 MS DC/DC converter model, 246

Multi-energy systems (MES), 4, 64 case study on interdependent MES model, 82 84 dependency categories, 65 66 external dependencies in smart local system, 76 78 infrastructure and carrier dependency, 64 65 interdependency modeling, 78 82 internal multicarrier dependency in smart local system, 66 76 objectives, 66 Multicarrier energy systems, 281 282 Multienergy aggregator, 83 84 Multienergy demand, 76 77 Multiobjective operation optimization eco-exergetic operation optimization of local integrated energy system, 103 107 energy supply with sources, 95f of local integrated energy system, 96 103 solution methodologies, 102 103 under study and mathematical formulation, 96 102 operation optimization of multiple integrated energy systems in local energy community, 108 119 for short and long-run sustainability of local integrated energy systems, 92 96 Multistage distribution expansion planning model (MSDEP model), 148 149

N Nanogrid management system (nMS), 244 Nanogrids, 9 case study, 268 275 problem formulation, 268 269 simulation results, 269 275 simulation setup, 269 configuration schemes with integrated energy storage systems, 244 245 as enabling technology, 242 244 energy storage system in, 239 258 modeling and control, 245 258 optimal energy management for virtual nonsumers nanogrid community, 259 263 Natural gas (NG), 282 Neighborhood energy trading (NET), 148 149 modeling, 156 157 planning with, 151

433

434

Index

Net load, 353 Net present value (NPV), 149 150 Network microgrids, 351 Network solutions (NSs), 148 149 New South Wales (NSW), 393 394 Non-fossil value, 220 NonIntrusive load monitoring (NILM), 398 399 Nonnetwork solutions (NNSs), 148 149 costing, 157 modeling, 154 155 Nonreserved, energy-based product, 178 Nonsumer community, 262, 268, 424 Nonthermostatically controlled appliances (NTCAs), 408 Northern Europe (NE), 188 189

O Objective function, 112, 132, 293 On-load tap changers (OLTC), 46 Operating cost of energy hubs’ assets, 293 295 Operation and maintenance cost (O&M cost), 400 401 Optimal energy management for virtual nonsumers nanogrid community, 259 263 mathematical model, 261 262 solution algorithms, 262 263 virtual nonsumers community review, 259 261 Optimal operation and planning, 420 Optimization of grid losses, 181 methods, 13 Outage, 327

P Parallel PHEV, 318 Pareto analysis, 13 Pareto frontier, 93, 104 105 Pareto set, 93 Particle swarm optimization (PSO), 149 Passive grid. See Traditional grid Pay-as-bid pricing (PABP), 202 Peak shaving, 340 Peer-to-peer (P2P) electricity trade, 55 electricity transaction, 215 217, 403 energy exchanges among prosumers, 218 221 BLORIN project, 221 Brooklyn Microgrid, 219 energy trading projects, 219 220

grid stabilization and vehicle to grid applications, 220 PPA management, 221 market, 202 PEI DC/AC converter model, 246 Photovoltaics (PV), 33 35, 93 94 panel, 392 system, 129, 283 285 Physical interdependency, 66 67 ΦΛEΠ Resilience quantitative framework, 366 Pinball function, 397 Plug-in hybrid electric vehicle (PHEV), 316 317 Point forecasting, 397 Point of common coupling (PCC), 243 Pool Hub, 198 199 Post-primary voltage control (PPVC), 46 Power balance constraints, 303 flow constraints, 303 Ledger, 220 node model, 4 outage, 327 Power blackout. See Power—outage Power cut. See Power—outage Power electronic interface (PEI), 242 Power failure. See Power—outage Power out. See Power—outage Power Purchase Agreements (PPA), 218 management, 221 Power to hydrogen model, 251 252 Power-electronic-DER (PE-DER), 351 352 Power-to-gas (P2G) storage system, 283 285 technologies, 73 74, 240 241 Power-to-hydrogen DC/DC converter model (P-to-H DC/DC converter model), 252 Practical-Bizantine-Fault-Tolerance algorithm (pBFT algorithm), 212 213 Price-based demand response programs, 287 289 Price formation, 202 Primary reserve power. See Frequency containment reserves (FCR) Probabilistic cost of NSs, 158 159 Probabilistic load forecasting, 397 Probability distribution functions (PDFs), 149 Probability of exceedance (POE), 150, 153 Producer-Link, 39 Proof of Authority (PoA), 212 213 Proof of Stake consensus (PoS consensus), 211 212

Index

Proof of Work (PoW), 211 Prosumers, 199, 208 Proton exchange membrane (PEM), 237 based power-to-hydrogen, 237 238 Pumped hydro energy storage (PHES), 236

Q Quantile regression (QR), 397

R Rate of change of frequency (RoCoF), 356, 363 Real-time ancillary services, 41 42 Real-time pricing (RTP), 5 6, 283 285, 289 Realistic 747-bus radial feeder, 168 Recurrent neural network (RNN), 395 Regulatory barriers, 25 26 Regulatory framework, 181 187 Renewable DERs, 400 401 Renewable energies model, 283 285, 298 299 Renewable Energy Community (REC), 20, 227 Renewable energy directive II (REDII), 259 261 Renewable energy sources (RES), 1 2, 6, 12, 33, 176 178, 215, 233, 292 293, 323, 348, 419 420 integration, 183 impact of RES on power system grids, 352 360 impact on overall inertia, 355 357 impact on voltage regulation, 358 359 impacts of RES on system, 359 360 impact of variability in secure and efficient operation of power system, 352 355 support, 341 uncertainties description and assessment, 352 363 Renewable energy technology, 391 Renewable generation, 33 Renewable self-consumers, 17 18 Replacement reserves (RR), 177, 189 Research, Development & Innovation (RD&I), 17 18 Reserved, capacity-based product, 178 Resilience, 363 increasing resilience of high RES system with flexible resources, 368 matrix, 366 368 operational measurements, 368 370 trapezoid, 365 366 Resilient AC distribution systems (RACDS), 351

Resilient DER scheduling in microgrids, 410 411 Risk-managed planning, 150 151 Robust optimization methods, 12

S SC DC/DC converter, 249 251 Scalability of distributed approach, 141 145 Scheduling function, 130 “SecControl” functions, 376 Sector coupling, 64 Self-consumption, 17 18 Self-sufficiency, 17 18 Series PHEV, 318 Series Resonant Converter (SRC), 249 251 Series-parallel PHEV, 319 320 Shiftable demand response program, 300 Shiftable IDR program, 291, 301 Short circuit levels of network, 360 Short-term load forecasts (STLFs), 331 Short-term scheduling of DERs in buildings, 406 409 in demand side, 406 412 in microgrids, 409 411 in VPPs, 411 412 Silicon Controlled-Rectifier (SCR), 359 360 Slave, 256 258 Small medium enterprises (SMEs), 18 Smart contracts (SC), 213 215, 225t in BLORIN project for DR and V2G implementation, 226 228 Smart decentralised energy systems, 33 35 adaptive frequency containment control, 48 49 balance steering control, 48 BRC, 47 decentralization in European future scenarios, 34 decentralization in European R&D projects, 34 35 decentralized architecture, 36 41 decentralizing the DA/ID energy market clearing and grid prequalification of ancillary services, 49 57 ELECTRA Web-of-Cells control concept, 45 46 grid-secure activations for ancillary services, 41 45 inertia control, 49 post-primary voltage control, 46 pros and cons of decentralization, 35

435

436

Index

Smart distribution grids, integration of electric vehicles in, 322 333 Smart Energy Service Provider (SESP), 198 199 Smart Grid Architecture Model (SGAM), 39 Smart grids, 323 framework, 2 Smart microgrids, 233, 240 Smart Networks for Energy Transition (SNET), 17 18 Smart prosumers, simulation and emulation of, 223 225 Smart-Grid Smart-City (SGSC), 393 SmartNet (H2020 project), 35 SNOCU blockchain, 224 Social barriers, 25 26 Solar exergy input rate, 101 102 Solar thermal plant, 98 Solidity, 214 Space cooling (SC), 96 97 Space heating (SH), 96 97 Spatial-temporal model, 331 Spinning reserve, 340 reduction, 265 266 Spot pricing, 5 6 Stake, 211 212 Standard deviation (SD), 151 152 State of Charge (SoC), 180, 332 State of energy (SoE), 134 135 Steady-state reactive power, 178 Stochastic programming models, 12 Storage, 2 Storage-Link, 39 Suboptimal solution, 93 Subsidiarity principle, 34 Substation automation unit (SAU), 39 Supercapacitors, 237, 249 Superconductive magnetic energy storage (SMES), 235 Supplier, 200 Support Vector Machine (SVM) model, 392 393 Sustainability, 92 Synchronous generators (SGs), 178 System Average Frequency Index (SAIFI), 364 System average interruption duration index (SAIDI), 159, 364 System average interruption frequency index (SAIFI), 159 System flexibility, 292 services, 176

System operator (SO), 178 System overloading, 327 System-based energy, 3

T Tariff rates, 5 6 Technical barriers, 25 26 Technical failure, 359 Technical VPPs (TVPPs), 412 Ten-Year Network Development Plan 2020 (TYNDP), 34 Thermal energy storage (TES), 7, 283, 297 298 Thermostatically controlled appliances (TCAs), 408 Thermostatically controlled loads (TCL), 180 Time-of-use (TOU), 288 289 demand response program, 300 301 pricing, 288 289 rates, 5 6 Time-series data, 371 Tokens, 215 Traditional grid, 322, 326 Trans-active energy system (TE system), 19, 391 392 with DERs, 399 405 Transferrable IDR program, 291, 301 302 Transmission and Distribution networks (T&D networks), 35 Transmission system operator (TSO), 32, 182 183, 201, 330 Transportation sector, carrier dependencies in, 75 76 Truncated Gaussian distribution, 155

U Uncertainties assessment, 151 153, 348 ΦΛEΠ Resilience quantitative framework, 366 flexibility and resilience matrix, 366 368 future system grid projection, 350 352 increasing resilience of high RES system with flexible resources, 368 metrics for assessing distribution system resilience, 363 365 signs of vulnerability, 364 total restoration cost, 365 present and future energy landscape, 349 350 renewables uncertainties in short-term scheduling of DER, 370 384

Index

RES uncertainties description and assessment, 352 363 resilience trapezoid, 365 366 uncertainties affecting system resilience, 363 370 Universal Smart Energy Framework (USEF), 52, 198 199 Use cases (UC), 54 Utility, 408

V Validators, 212 213 Value of customer reliability (VCR), 159 Variable renewable generation (vRES), 40, 179 Vehicle-to-grid (V2G), 217, 330, 333 344, 423 424 applications, 6 7, 220 challenges of using blockchain technology for DR and, 221 222 functions for frequency regulation, 339 341 use of EVs for grid support, 334 339 Vehicle-to-home (V2H), 334 336 Vehicle-to-vehicle (V2V), 217, 334, 336 337

Virtual node (VN), 216 217 Virtual nonsumers community review, 259 261 Virtual power plants (VPP), 9 10, 391 392 Vision 2050, 34 Voltage adaptive method. See Distributed method Voltage control, 178, 181 reserves, 190, 192 Voltage droop control (PVC), 46 Voltage regulation, 266 267 Voltage regulator (VR), 167 Voltage Source electronic Converter (VSC), 267 Voltage source inverter (VSI), 351 352 Voltage support, 266

W Web-of-Cells concept (WoC concept), 9, 38 39, 44 45 architectures, 350 351

Z Zero marginal costs, DERs with, 399 403

437