Energy Systems Evaluation (Volume 2): Multi-Criteria Decision Analysis (Green Energy and Technology) 3030673758, 9783030673758

Citation preview

Green Energy and Technology

Jingzheng Ren   Editor

Energy Systems Evaluation (Volume 2) Multi-Criteria Decision Analysis

Green Energy and Technology

Climate change, environmental impact and the limited natural resources urge scientific research and novel technical solutions. The monograph series Green Energy and Technology serves as a publishing platform for scientific and technological approaches to “green”—i.e. environmentally friendly and sustainable—technologies. While a focus lies on energy and power supply, it also covers “green” solutions in industrial engineering and engineering design. Green Energy and Technology addresses researchers, advanced students, technical consultants as well as decision makers in industries and politics. Hence, the level of presentation spans from instructional to highly technical. **Indexed in Scopus**.

More information about this series at http://www.springer.com/series/8059

Jingzheng Ren Editor

Energy Systems Evaluation (Volume 2) Multi-Criteria Decision Analysis

Editor Jingzheng Ren Department of Industrial and Systems The Hong Kong Polytechnic University Hong Kong, China

ISSN 1865-3529 ISSN 1865-3537 (electronic) Green Energy and Technology ISBN 978-3-030-67375-8 ISBN 978-3-030-67376-5 (eBook) https://doi.org/10.1007/978-3-030-67376-5 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Contents

Overview of Multi-criteria Decision Analysis and Its Applications on Energy Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ruojue Lin and Jingzheng Ren

1

Development of Decision Support Tools: Methodological Framework of System Dynamics Energy Models . . . . . . . . . . . . . . . . . . . . . Athar Kamal and Sami G. Al-Ghamdi

27

Multi-Criteria Decision Analysis Methods for Sustainability Assessment of Renewable Energy Systems and Its Potential Application to Sustainable STEM Education . . . . . . . . . . . . . . . . . . . . . . . . . Jin Su Jeong and David González-Gómez National Energy Sustainability and Ranking of Countries . . . . . . . . . . . . . Vassilis S. Kouikoglou, Evangelos Grigoroudis, and Yannis A. Phillis

39 63

Multi-criteria Assessment of Sustainability for Energy Systems Under Uncertainty: Grey-Based Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Yakup Çelikbilek, Nurdan Tüysüz, and Fatih Tüysüz Coupling LCSA and Multi-criteria Decision Analysis for Energy System Prioritization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Weichen Li, Lichun Dong, Jingzheng Ren, and Ruojue Lin Multi-criteria Decision Analysis Methods for Sustainability Assessment and Improvement of Energy Systems Under Uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 Xusheng Ren, Lichun Dong, and Jingzheng Ren An Integrated Hesitant Fuzzy Decision Model for Sustainable Wind Farm Site Selection: The Case Study in the Central Anatolian Region of Turkey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 Beyzanur Cayir Ervural

v

vi

Contents

Rough Set-Based Multi-Criteria Decision Analysis Methods in Sustainability Assessment of Photovoltaic Projects . . . . . . . . . . . . . . . . . 219 Jing Li and Wenyan Song Multicriteria-Oriented Optimization of Building Energy Performances: The Annex 72 IEA-EBC Experience . . . . . . . . . . . . . . . . . . . 239 Francesco Montana, Sonia Longo, Harpa Birgisdottir, Maurizio Cellura, Rolf Frischknecht, Francesco Guarino, Benedek Kiss, Bruno Peuportier, Thomas Recht, Eleonora Riva Sanseverino, and Zsuzsa Szalay Multi-objective Genetic Algorithm Optimization of HVAC Operation: Integrating Energy Consumption, Thermal Comfort, and Productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 Sokratis Papadopoulos and Elie Azar

Overview of Multi-criteria Decision Analysis and Its Applications on Energy Systems Ruojue Lin and Jingzheng Ren

Abstract Energy systems are popularly studied to maintain and improve the energy supply, which is an essential and irreplaceable element for modern society. With the development of technology, stakeholders need to face more prioritization and optimization problems in the establishment, operation, and maintenance process of the energy system. To make various and complicated decisions with multiple criteria to consider in the life cycle of an energy system, multi-criteria decisionmaking (MCDM) methods have been widely applied. Since the MCDM method is used in the research and development process for an electricity utility company in 1991, MCDM methods have been extended, improved, and adopted in an increasing number of studies regarding the energy system. In order to provide a clear reference for researchers in the related field, this study overviewed 1995 literature solving problems related to the energy system by applying MCDM methods. The definition and classification of energy systems have been clearly explained with illustrations and examples. The general framework of MCDM is introduced, and the MCDM methods used in the studies concerning energy systems are summarized and explained. To comprehensively analyze the relative works of literature, all articles using MCDM methods to solve energy system-related problems are collected and analyzed by conducting a bibliometric analysis. In the bibliometric analysis, the co-citation networks of journals, publications, and authors and the co-occurrence network of keywords are examined and evaluated. As the result, the MCDM has been proven a feasible method for energy system; multi-criteria decision analysis for energy system has been shown as a heated topic, and the most highly cited papers, authors, research countries/regions, and sources are identified. Keywords Multi-criteria decision analysis · Energy system · Sustainability assessment · Multi-criteria decision-making · Energy

R. Lin · J. Ren (B) Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong Special Administrative Region, China e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 J. Ren (ed.), Energy Systems Evaluation (Volume 2), Green Energy and Technology, https://doi.org/10.1007/978-3-030-67376-5_1

1

2

R. Lin and J. Ren

1 Introduction With the development of technology and society, the supply of energy became increasingly essential to maintain the operation of society [1, 2]. On the individual level, the normal life of mankind will be significantly influenced without energy supply, as the supply of food, clothing, housing, and transportation will not continue without energy. Besides, at the national level, energy supply is highly related to national security issues and is always an important factor for country safety [3, 4]. At the world level, the development of science and technology will be greatly restricted without energy supply. For instance, the development of science and technology now requires a large amount of energy input, and it also requires energy as a foundation. Therefore, long-term and stable supply of energy is very important. The energy system refers to all systems related to the production, conversion, transportation, storage, use, and recovery of energy [5]. The concept of the energy system can provide a more orderly energy supply chain and a clear framework for overall optimization, making it easier to optimize system efficiency. Therefore, the energy system is an important research object, and many studies have discussed the development and optimization of the energy system. Now that energy systems are diversified, which energy system to use has also become an important decision-making issue. Energy systems are complex and have multiple judgment standards. In the analysis of energy systems, it is difficult to show that a certain option has overwhelming advantages in all aspects. Taking biorefinery technologies as an example [6], Fischer–Tropsch diesel is efficient but high in cost, while grain ethanol is economic but not environmentally friendly. Therefore, because of each energy system with its advantages and disadvantages, MCDM has become an effective decision-making tool. MCDM is a tool that can analyze, sort, or filter out the best solutions among multiple solutions based on multiple dimensions, and it has effectively solved some comparative problems of energy systems, for example, articles [7–9]. Because MCDM plays an important role in the optimization and analysis of energy systems, it is necessary to conduct a literature review on the topic of energy systems using MCDM methods. In this article, we will • • • •

define energy system; review the energy system research topics that apply the MCDM method; classify the MCDM methods used in related topics; and analyze the development trend of analysis and research.

In addition to this section, this article will be divided into four parts: The concept of the energy system will be defined in Sect. 2; Sect. 3 illustrates the details of MCDM methods; a bibliometric analysis is conducted to comprehensively analyze the pieces of literature related to energy system using MCDM methods in Sect. 4; this chapter is concluded thereafter in Sect. 5.

Overview of Multi-criteria Decision Analysis …

3

2 Energy System Energy systems literally mean the system related to energy. However, seldom of researches use this concept as the study scope, and the definition of this concept is still unclear. In this section, the energy and energy system will be clearly defined and introduced to classify the scope of this study.

2.1 Definition and Classification of Energy There are different ways to classify energy. According to the basic form of energy, energy can be divided into primary energy and secondary energy (see Fig. 1) [10]. Primary energy refers to the energy resources that exist in the natural world without processing and conversion. Secondary energy refers to the resources obtained after processing and conversion of primary energy. Primary energy can be divided into renewable energy and non-renewable energy. Any energy that can be continuously replenished or regenerated in a relatively short period is called renewable energy [11]. The energy that has been formed after hundreds of millions of years and cannot be recovered in the short term is called non-renewable energy. Secondary energy can be divided into procedural energy and energy-contained substance. Energy-contained substance refers to substances containing energy, which can be directly stored and transmitted [12]. Procedural energy refers to the energy produced during the movement of materials with relatively concentrated energy [12]. Electricity is the most widely used procedural energy, while gasoline and diesel are currently the most widely used energy sources [13–15].

Fig. 1 Classification of energy (based on [10])

4

R. Lin and J. Ren

As for electricity, the generation and transmission of it are the most commonly discussed topics, for example, the selection of several electricity generation pathways [16] and large-scale power grids management [17]. The transportation of heat energy mostly exists in winter heating supply and industrial heat supply [18, 19]. The chemical energy is partly used for transportation [20] and other aspects that require special forms of energy or temporarily reserved fuel forms for the convenience of energy transmission, such as hydrogen [21, 22] and natural gas [23, 24].

2.2 Definition and Classification of the Energy System Taking a structural viewpoint, the energy system is defined as “all components related to the production, conversion, delivery, and use of energy” [25]. In addition, current energy storage has also become an important part of the energy supply. There are many forms of energy, mainly including mechanical, electric, magnetic, gravitational, chemical, ionization, nuclear, elastic, radiant, and thermal energy [10]. According to the situation of modern society, the commonly used forms of energy supply are electricity, thermal energy, and chemical energy. Therefore, in this article, an energy system is defined as any system related to the production, conversion, transportation, storage, and use of electrical, thermal, and chemical energy. However, due to the diverse ways of using energy, for example, buses need fuel, household appliances need electricity, and food processing plants need heat. If all energy use scenarios are considered, the processing and production, supply chain, shipping, transportation, and architectural design of the factory will be designed (see Fig. 2). The fields involved are too extensive, and the use of energy will not be the focus of this article. Production refers to the collection of primary energy, such as mining and natural gas collection. Conversion refers to the production process of secondary energy, for instance, crude oil refining, thermal power plant power generation, and hydrogen production. Energy storage refers to the preservation of various forms of energy. The energy storage discussed in this article is mainly electrical and thermal energy, such as compressed air energy storage, pumped hydroelectric storage, and flywheel energy storage. Delivery refers to the transmission process of various forms of energy, including natural gas pipelines, power transmission cable systems, and oil tank fleets.

Fig. 2 Components in the energy system

Overview of Multi-criteria Decision Analysis …

5

2.3 Related Research Topics The MCDM methods have been widely applied in various research topics regarding energy system establishment, management, and optimization, including location selection [26, 27], technology prioritization [28], resource selection [29], region analysis [30], strategy planning [31], policy-making [32], and investment analysis [33]. Among the reviewed literature, the most popular topic is location selection analysis and technology prioritization.

3 Multi-criteria Decision-Making Method Multi-criteria decision-making (MCDM) is an effective tool for alternative selection and prioritization when multiple criteria are considering. It is widely acknowledged that MCDM could provide a logical dynamic analysis system for information collection, organization, calculation, and analysis, while it can rapidly reflect the preference of decision-makers in the quantitative terms. Since the rapid development of technologies led to the variety of alternatives provided in a different field, each option has its advantages and disadvantages which cause difficulty in selection. The MCDM shows its advance in problem with multiple options and multiple criteria, and therefore, it has been adopted and proven efficient by many researchers in the multi-criteria ranking or selection problem. In this section, the general framework of MCDM is introduced to give readers an overall idea of this method. The two important steps, criteria weighting and ranking are especially illustrated, and the methods commonly used in these two steps are explained in detail then. Thereafter, some special theories, that have been adopted to combined with MCDM in the literature, are reviewed and summarized in Sect. 3.4.

3.1 General Framework of MCDM In most cases, the multi-criteria decision-making problems regarding the energy system were solved based on a general MCDM framework. The general framework includes six steps (as shown in Fig. 3), including criteria system determination, criteria weighting, alternative determination, information collection, decision-making matrix determination, and ranking [34, 35]. In the beginning, the research problem is identified, and the alternatives can be selected based on reality and the research field. Then, the criteria system should be determined. In the criteria system, the criteria should be selected to descript the performance of alternatives in the target aspect or aspects thoroughly and without overlap. For example, if a suitable hydrogen generation pathway is selected based on economic sustainability, then the criteria to describe benefit and cost should be

6

R. Lin and J. Ren

Fig. 3 General framework of MCDM methods

included, respectively. For criteria regarding benefit, the net present value (NPV), return on investment (ROI), productivity, and net profit can be the parts of the criteria system. The capital cost, operation cost, raw material cost, and maintenance cost can be adopted as part of the criteria system with regard to cost. But there should not be any overlap in the criteria system, for instance, the labor cost cannot be considered as a criterion in the previous criteria system because labor cost has already been included in the criterion and operation cost. Based on the previous tips, the determination of alternatives and criteria can be completed as the first step. The criteria data regarding alternatives should be collected, and this information together with criteria weights forms the basis of decision-making. The data could be collected from different sources including literature reviews, field visits, simulations, and judgment by experts. The data sources can be selected and decided depending on the scope of study, availability, and feasibility of data. As for criteria weights, they can be determined by different criteria weighting methods. Since there are various criteria methods used in the MCDM method and the criteria weighting methods are usually complex with multiple steps, the details of criteria weighting methods are introduced in Sect. 3.2. Assume that m alternatives need to be prioritized based on n criteria in a study. With criteria data and criteria weights, the decision-making matrix can be built presented as Eq. (1). This matrix is commonly adopted in MCDM as the prepared format for further ranking and analysis.

Overview of Multi-criteria Decision Analysis …

A1 A2 .. . Am

C C · · · Cn ⎡1 2 x11 x12 · · · x1n ⎢ x21 x22 · · · x2n ⎢ ⎢ . . . ⎣ .. .. . .

7

⎤ ⎥ ⎥ ⎥ ⎦

(1)

xm1 xm2 xmn w1 w2 · · · wn

 where xi j indicates the data of the j-th criterion C j with regard  to the i-th alternative (Ai ); w j represents the criteria weights of the j-th criterion C j . After the establishment of the decision-making matrix, ranking and scoring can be conducted accordingly. Many scoring and ranking methods have been developed and improved, so the details of the scoring and ranking methods are clearly explained in Sect. 3.3. The selection or ranking results will be generated as the reference for decision-makers after the last step of scoring and ranking.

3.2 Criteria Weighting Methods The criteria weighting is one of the most important steps used in MCDM, as it determines the different weights for each criterion when a decision is made. The accuracy of the criteria weighting results, generated by quantifying decision-makers’ preferences into numerical weights, significantly influences ranking or selection result afterward. The criteria weighting method can be classified into a subjective method and an objective method. In a subjective method, the opinion of decision-makers is the basis for criteria weighting determination and usually presented in linguistic terms. On the contrary, the objective method adopts criteria data as the sources and generates criteria weight by statistical calculation. Some classical subjective and objective criteria weighting methods applied in the energy system are introduced below.

3.2.1

Subjective Methods

The analytic hierarchy process (AHP) is the most famous and classical subjective criteria weighting method proposed by Saaty [36]. It illustrates a precise method for quantifying the criteria weights based on the opinion of experts on the estimation of relative magnitudes of factors through pairwise comparison. This method is still popular currently and has been applied in the MCDM study of energy systems [37–39]. However, since the criteria considered in AHP are assumed mutual independent, the case study with interdependent criteria cannot be correctly analyzed by AHP. Therefore, the analytic network process (ANP) [40] was developed by extending AHP to a method that can proceed MCDM problem with complex and interrelated

8

R. Lin and J. Ren

relationships between decision elements. The ANP has been used in the case study with interdependent criteria. For instance, Cayir Ervural et al. [41] used ANP in energy planning for Turkey; Sakthivel et al. [42] selected the best biodiesel blend based on the ANP-based MCDM method. Decision-making trial and evaluation laboratory (DEMATEL) technique is a criteria weighting method for interrelated criteria as well. Though DEMATEL, the complicated causal relationships among criteria can be visualized through diagraphs, and the cause and effect relationships can be analyzed. The critical factors of the system can be thereafter identified. The DEMATEL can be used as an identification method of influence factors [43, 44] and a criteria weighting method in energy system analysis combining with ANP. Integration of DEMATEL and ANP (DANP) is a perfect method to proceed weighting problem with interdependence among criteria and has been used in energy system problem solving, for example, Hsu et al. [45] adopted the DANP method in selection of an outsourcing provider; DANP has been used as the criteria weighting method for low-carbon energy planning [46]. The best-worst multi-criteria decision-making method (BWM) proposed by Rezaei [47] in 2015 is a criteria weighting method adopting the idea of pairwise comparison as well. Comparing with AHP, the BWM simplifies the steps of pairwise comparison and only conducts the best-to-others comparison and the others-to-worst comparison. Due to the advantageous characteristics of BWM, it was adopted in several decision-making problems concerning the energy system in recent years. For instance, the external forces affecting supply chain sustainability in the oil and gas industry were evaluated by using BWM [48]; the selection biomass thermochemical conversion technology in the Netherlands was a study based on BWM [49]; Wang et al. [2] have studied the energy performance contracting industry in China using BWM. The method of measuring attractiveness through a categorical-based evaluation technique (MACBETH) requires only qualitative judgments about the difference of attractiveness between two elements at a time, and it can generate quantitative criteria weights. A software called “M-MACBETH” was developed by this research team and assist researchers to solve MCDM based on MACBETH at a faster speed. This method is also adopted in studies regarding the energy system. For example, Ertay et al. [50] evaluated the renewable energy options by MACBETH and fuzzy AHP; the on-board hydrogen storage technologies are using the MACBETH method [51]; MACBETH method was also used in assessment for power plants technologies [52].

3.2.2

Objective Methods

The entropy weighting method is the most widely used objective weighting method. The idea of entropy by Shannon was originally developed for information theory. The entropy is defined as the disorder degree and its utility in system information. The smaller the entropy value is, the smaller the disorder degree of the system is. Similarly, the entropy value in the MCDM method represents the disorder degree of information provided in the decision-making matrix. Therefore, the criteria weights

Overview of Multi-criteria Decision Analysis …

9

can be determined by the entropy weight method based on the amount of information. Regarding the MCDM study about the energy system, the entropy weight method contributes to criteria weighting in several related studies. For example, Wei et al. used the entropy weighting method in energy saving and emission reduction effect evaluation in thermal power plants [53]; the energy regulation in China was studied based on the entropy weighting method as criteria weighting method [54].

3.2.3

Extended Methods

In order to address the hesitations and ambiguity existing in human judgments, many studies have extended criteria weighting methods to uncertain environments by combining the methods with interval numbers, fuzzy numbers, and rough numbers. For example, the interval DEMATEL can determine the lower and upper boundaries of criteria weights, which is used to assess the distributed energy system [55]; the fuzzy AHP [56, 57] which is a combination of AHP and fuzzy set theory is an excellent weighting tool since the fuzzy numbers are more suitable to express linguistic preferences than crisp numbers; the BWM has been extended to rough BWM to evaluate the suitable location for wind farms [58].

3.3 Scoring and Ranking Methods As the last step of the MCDM method, the scoring and ranking step helps to integrate criteria weights and all information regarding alternatives to further generate a feasible ranking or an optimal choice. Based on the results generated by each model, the traditional scoring and ranking methods could be classified into two categories: scoring methods and outranking methods. As for a scoring method, a score is generated for an alternative, and the prioritization results can be determined by ranking those scores in either descending or ascending sequences. In the outranking method, a ranking sequence is generated as a result. The ranking can be obtained in an outranking method, but the integrated differences among alternatives are not quantitative. The details of each type of scoring and ranking methods are illustrated below.

3.3.1

Scoring Methods

The VIKOR method is a traditional MCDM method developed by Opricovic and Tzeng [59]. It was developed based on the idea that the best option is the closest to the ideal, and a score for each alternative is generated according to the closeness of each alternative to the ideal option. This classical MCDM method was popularly used in solving MCDM problems in the energy system. For instance, the VIKOR method was used to select the best materials for thermal energy storage technology

10

R. Lin and J. Ren

[60]; the renewable energy system schemes in tourist resorts are evaluated based on VIKOR [61]; VIKOR was adopted to optimize a system regarding biomass-based cell [62]. The technique for order of preference by similarity to ideal solution (TOPSIS) is also a classic MCDM method proposed by Hwang and Yoon [63]. Different from the VIKOR, the TOPSIS method not only considers the best alternative to have the highest closeness between the alternative and the positive ideal solution (PIS), but also the longest distances between this alternative and the negative ideal solution (NIS). The score is given to each alternative based on the integrated performance of closeness to PIS and NIS. This classical MCDM has been widely used in energy system problem solving, such as building energy performance benchmarking [64], environmental analysis of energy in European countries [65], and energy conservation evaluation of coal company [66]. Gray relational analysis (GRA) developed by Deng [67] is one of the important MCDM methods. In the gray system theory, the uncertainty can be expressed by the degrading of gray, as the gray is the color between white and black, and the closeness level of the gray color to black color or white color means the degree of the alternative to be positive or negative. Based on this idea, the gray relational score generated from GRA can reflect the ranks of alternatives. This method is also widely adopted in the energy-related studies, for example, flat plat collector process optimization in solar energy system [68], evaluation for batteries [69], and energy-saving evaluations for residential buildings [70]. Data envelopment analysis (DEA) is a nonparametric method in operations research and economics for the estimation of production frontiers. It is used to empirically measure the productive efficiency of decision-making units (DMUs). Thereafter, the DEA method has been modified as a ranking method based on efficiency measurement for the energy system. The larger the value of efficiency, the best the option is. By using DEA, the efficiency of several energy sources was evaluated and ranked [71, 72].

3.3.2

Outranking Methods

The preference ranking organization method for enrichment of evaluations (PROMETHEE) is a series of MCDM that provide ranking sequence as results based on the pairwise comparison. The PROMETHEE I and PROMETHEE II differ merely in the last step. PROMETHEE I provides the ranking with possibly more than one alternative in the same place, while PROMETHEE II can generate strong preference so that all alternatives will be ranked in order. These two methods have been used in energy system problem solving such as site selection for waste-to-energy plant site [73], evaluation for energy performance of hotels [74], and bioenergy pathways analysis [75]. The method of ELimination Et Choice Translating Reality (ELECTRE) is a family of outranking MCDM methods published by Roy [76]. The relationships between any two alternatives are finally figured out as indifference, strict preference, weak

Overview of Multi-criteria Decision Analysis …

11

preference, and incomparability in the original version of ELECTRE. Based on the pair relationship, the alternatives could be prioritized. The ELECTRE has been improved and extended to be applied in different situations as the ELECTRE family. In the energy system, the ELECTRE method and its family have been widely applied as well to solve some problems such as photovoltaic production pathways selection [77], energy recovery technology evaluation [78], and risk evaluation for energyrelated investment [79].

3.3.3

Extended Methods

To modify MCDM to be suitable in the decision-making environment with uncertainty, researchers have extended commonly used MCDM methods by additionally considering the case of uncertainty, such as using interval numbers and triangular fuzzy numbers instead of a crisp number to better describe criteria data collected from energy system in reality. For example, a fuzzy VIKOR is used to optimize renewable energy planning [38], the interval TOPSIS, the interval GRA, fuzzy ELECTRE, and rough PROMETHEE.

3.4 Combination with Other Concepts To solve problems existing in the energy systems, some concepts were innovatively combined with MCDM. Obtained from the reviewed literature, these concepts commonly used in multi-criteria problems include life cycle sustainability assessment (LCSA), geographic information system (GIS), and quality function deployment (QFD).

3.4.1

Life Cycle Sustainability Assessment (LCSA)

With worsening conflicts between limited resources and increasing demands, the concept of sustainability has arisen more and more attention. To evaluate the sustainable performances of target objects, the method of sustainability assessment is necessary and essential. In this case, LCSA has been developed as a useful tool to analyze the sustainability performance of energy systems from cradle to grave. LCSA is an integrated assessment extended from life cycle assessment (LCA) and refers to the evaluations of impacts in environmental, economic, and social aspects. LCSA is consisting of LCA for environmental assessment, life cycle costing (LCC) for economic assessment, and social life cycle assessment (S-LCA) for social assessment. Every single assessment in LCSA follows the same steps: goal and scope definition, inventory analysis, impact assessment, and interpretation (see Fig. 4). By conducting LCA, LCC, and S-LCA, the impacts in each aspect are obtained and

12

R. Lin and J. Ren

Fig. 4 Framework for LCA, LCC, and S-LCA [80]

transformed into quantitative attributes and are displayed in unified values. Therefore, the application of LCSA can provide researchers numerical results of each attribute in environmental, economic, and social aspects, respectively. Since the ranking and scoring methods of MCDM problems are data-based, the LCSA is a good source for MCDM models. Each criterion in the MCDM model can refer to an attribute measured by LCSA, and MCDM can provide each alternative an integrated score for its sustainable performance. Therefore, the combination of LCSA and MCDM is an efficient and useful framework. This integrated framework has been widely applied in sustainability-oriented MCDM problems for energy systems. For example, Ren and Toniolo [81] established the life cycle sustainability decision-support framework to evaluate hydrogen production pathways based on LCSA and MCDM method; this framework was also used to rank electricity generation technology [1].

3.4.2

Geographic Information System (GIS)

Literately, GIS is a data management system integrating geographic information. It is a system where researchers can input, save, inquire, analyze, and display geographic data. Different from other information systems, GIS provides data with both numbers and corresponding spatial information, which can assist decision-makers to analyze any situations of any location with more accuracy information support. Therefore, GIS technology has been applied to scientific investigation, resource management, property management, development planning, mapping, and route planning. According to related kinds of literature, the GIS was frequently chosen to solve location selection problems by combining with MCDM methods. As mentioned above, MCDM is a data-based ranking methodology which requires data for multiple criteria with regard to alternatives. As for a location selection problem, GIS is one of

Overview of Multi-criteria Decision Analysis …

13

the best sources to provide sufficient and accurate information. Also, some value of criteria can be obtained by inferencing from GIS, such as distance and transportation costs. In this case, the method combining GIS and MCDM has been adopted in a lot of studies selecting a suitable site for energy systems. For instance, GIS was combined with AHP and TOPSIS to determine the best location for wind farm installation in Grace [37]; GIS was also used to evaluate the locations for solar farms [82, 83] and landfill site [84].

3.4.3

Quality Function Deployment (QFD)

The QFD method is a classical quality management tool to improve product quality and production processes in the product design stage. To fulfill the demand of customers, the QFD is conducted to transform the opinion of customers into the numerical expression of measurable criteria and examined the performances of the new products and other products in competition. A QFD matrix is created to assist the conduction of the QFD process (see Fig. 5). Taking music player design as an example, customers replied that they want to purchase a portable music player. Then this idea will be expressed as lighter weight, and a smaller size of music player should be designed. With the reflection of customers, all products or product designs will be scored and ranked.

Fig. 5 QFD matrix (modified by [85])

14

R. Lin and J. Ren

Fig. 6 Framework of QFD-based MCDM (modified by [86])

This framework of QFD has been revised and combined into the MCDM method to select suppliers because the framework of QFD shares the same elements of MCDM, such as linguistic opinions from decision-makers, criteria to evaluate, and an integrated score for each criterion. The framework of QFD-based MCDM can be illustrated in Fig. 6. Therefore, the QFD-based MCDM framework is adopted by many researchers in the study of suppliers’ selection [86, 87], fuel selection [88], and risk evaluation for energy policy [89].

4 Bibliometric Analysis Bibliometrics is the use of statistical methods to analyze publications on certain topics. A general introduction to the basic information of the energy system-related publications is given in this section. To comprehensively analyze the multi-criteria decision-making researches related to energy systems, the bibliometric analysis in terms of co-citation networks of journals, publications, and authors and the co-occurrence network of keywords are conducted to review all relative articles.

4.1 Data Sources and Statistics Summary The literature related to energy systems can be collected by searching related keywords in the academic information system Web of Science (WoS). To comprehensively study the literature, the information is collected by searching method keywords combining featured keywords (see Table 1). The logic for keywords combination is that all keywords about the energy system are inserted as Topic Keywords with relation OR and then combine keywords about methods with AND. To be specific, it

Overview of Multi-criteria Decision Analysis … Table 1 Keywords for literature selection

15

Keywords about methods Multi-criteria decision-making MCDM Multi-criteria decision analysis MCDA

Keywords about energy system AND

Energy Mining Energy storage Biorefinery Refinery Battery Electricity Heat Steam Hydropower

Fuel Nuclear Wind power Solar Gas Oil Gasoline Petrol Hydrogen Biomass

should be written as “(energy OR mining OR … OR biomass) AND (multi-criteria decision-making OR MCDM)”. According to the searching results, the first article studying decision-making problem solving by MCDM is published in January 1991 [90]. We selected all articles and journals in the searching results publishsssssed from 1991 to September 29, 2020, and there are 1995 publications adopted in this study. As mentioned above, the increasing significance of energy supply attracts the rapid development and construction of energy systems. The decision-making problems related to the energy system have been popularly studied, and the most cited article is a review of MCDM applied in sustainable energy planning with 896 citations [91]. Based on the searching results in WoS, the annual publications and the annual citations of these energy system studies based on MCDM methods are summarized and presented in Fig. 7. Till the data collection date, the publication information for the last quarter of 2020 has not been updated. Therefore, the data from 1991 to 2019 is analyzed, and it shows upward trends for both annual publication and annual citation, which demonstrate the heated research tendency of this topic and the strong potential to further research. The countries or regions where these 1995 publications are distributed have been investigated as presented in Fig. 8. In the figure, the deeper color of the countries/regions, the more publications related to energy system solving by MCDM

Fig. 7 Summary of publications and citations

16

R. Lin and J. Ren

Fig. 8 Cartogram of publications

have been published. Researchers from China, Iran, India, and Turkey contributed the most in this topic, with 319, 235, 188, and 186 publications in total, respectively. According to the statistics, researchers from 100 countries or regions participated in the relative researches, and the USA, England, and Spain take the fifth to seventh positions in the number of publications. In these seven countries mentioned above, researchers have contributed more than 100 publications each, which indicates the admitted value of this topic worldwide. However, as displayed in Fig. 8, not all countries participated in the research and study of the energy system by using MCDM methods. Therefore, this method can be promoted and applied in a wider range worldwide. Based on the WoS database, the collected literature has been published in 112 different journals. Among these sources, there are 30 journals published more than 10 articles with relative topics as presented in Table 2. As reflected by the statistics, the articles with the related topic were more welcomed and accepted by journals regarding clean energy and sustainability, although journals with regard to computing and mathematics published some of the studied articles.

4.2 Bibliometric Analysis In this section, the VOSviewer software [92] is used to analyze the 1995 publications collected and selected as mentioned above. Based on the 1995 collected literature, we

Overview of Multi-criteria Decision Analysis … Table 2 Source summary of publications

17

Source titles

Record count

Journal of cleaner production

114

Energy

84

Renewable sustainable energy reviews

73

Energies

67

Sustainability

60

Renewable energy

52

Energy policy

45

Applied energy

36

Energy conversion and management

33

Clean technologies and environmental policy

24

Energy and buildings

23

Expert systems with applications

23

Sustainable cities and society

20

European journal of operational research

19

International journal of hydrogen energy

19

Journal of environmental management

17

Sustainable energy technologies and assessments

17

Science of the total environment

16

Building and environment

14

Computers industrial engineering

14

Annals of operations research

13

Applied soft computing

13

Energy sources part B economics planning and policy

13

Journal of civil engineering and management

13

Journal of intelligent fuzzy systems

13

Information sciences

12

Applied sciences basel

11

Resources policy

11

Technological and economic development of economy

11

Arabian journal for science and engineering

10

draw three figures with regard to the co-occurrence network of research keywords, the citation network of authors, countries, and sources. The keywords co-occurrence network is firstly analyzed to figure out the relatedness of keywords based on the number of publications in which they occur together. To accurate the analysis, this network generation requires keywords selection and

18

R. Lin and J. Ren

words to merge. The keywords are extracted from the title and abstract fields with minimum ten occurrences of the terms as the threshold, while structured abstract labels and the copyright statements are ignored. The words with the same meaning but in different forms should be merged to avoid the elimination of impacts. For example, the “analytical hierarchy process” and “analytic hierarchy process” were merged into “app”; “location selection” and “locations selection” were combined as “site selection.” The top 600 keywords of these 1995 publications are presented in Fig. 9. The larger the size of the dot, the higher the correlation of the keyword in the topic is. As shown in Fig. 9, the keywords have been categorized into five clusters in different colors, and each cluster represents keywords of the articles that have high co-occurrence. The keywords in blue color can be categorized as the research related to the energy system applied in the construction industry using MCDM methods. In this category, the “optimization” and “algorithm” are the two most frequently used and cited keywords. The yellow dots in Fig. 9 indicate that one of the major research topics in selected literature is the location selection problem of energy systems or facilities. The studies focusing on mathematical improvement are gathered in green color. The keywords in red are related to the sustainable discussion of energy systems, and the keywords in purple reflect the criteria used in MCDM. The co-occurrence network can provide a general overview of the MCDM studies. The co-citation network of authors is presented in Fig. 10. In this figure, the larger size of the dots means the higher citations of articles written by the author

Fig. 9 Co-occurrence map with regard to keywords

Overview of Multi-criteria Decision Analysis …

19

Fig. 10 Co-citation map with regards to authors

indicated. The color illustrated the situations of the citation of responding author in each year. In this study, if the color of the dot is closer to blue, the citation is relatively old in time. Similarly, if the dot color is closer to yellow, the citation is more updated. Observed from Fig. 10, the major dots with big size indicate an explosion of research interests in this topic during 2015–2018, and studies with high quality have been increased during this period as well. As reflected in Fig. 10, Ren, Zavadskas, and Streimikiene contributed the highest citations among all authors conducting related studies, and their publications published on around 2016–2019 have been the most highly cited. Ren and his team have published 27 related articles and applied MCDM methods to sustainability assessment of energy systems, such as biorefinery production systems [6], distributed energy systems [55], home heating technologies [93], energy storages [94, 95], and electricity generation [96]. The major contributions of Zavadskas, Streimikiene, and their colleagues are to develop a sustainable decision-making framework applied in renewable resources [97, 98] and mining systems [99–101]. The co-citation network regarding the countries/regions of authors shown in Fig. 11 indicates the countries/regions the most highly cited publications contributed in. Similar to the co-citation network of authors shown in Fig. 10, the color of dots in the diagram refers to the year of publications and size indicates the frequency of citations. As demonstrated in the figure, the latest studies conducted by researchers from China have been cited highly. The countries or regions taking the following positions include Iran, India, Turkey, England, and the United States of America. Since more papers have been published, more papers will be cited, and this information is consistent with the statistical summary discussed as shown in Fig. 8.

20

R. Lin and J. Ren

Fig. 11 Map with regard to the country of authors

The relatedness of sources that being cited by other articles is analyzed based on the number of papers they are cited together, and it is presented as the co-citation analysis of sources shown in Fig. 12. Similarly, a dot indicates a journal source, and the size of the dot illustrates the frequency of the source cited by other publications. As shown in Fig. 12, the Journal of Cleaner Production is the most highly cited source, and its latest articles have been continuously cited. Renewable & Sustainable Energy,

Fig. 12 Map regarding sources that being cited

Overview of Multi-criteria Decision Analysis …

21

Energy, Energies, and Sustainability are four journals that have the top frequency of citations excepting the Journal of Cleaner Production. Those journals are influential sources in the field of sustainability and energy; therefore, the studies of an energy system based on MCDM are efficient direction for future development in this field. The result of co-citation analysis is validated, as it is consistent with what is shown in Table 2.

5 Conclusions This chapter overviews the multi-criteria decision-making models and their applications in the energy system. The definition and classification of energy systems have been clearly explained with illustrations and examples. The general framework of MCDM is introduced, and the MCDM methods used in the studies about energy systems are summarized and explained. To comprehensively analyze the relative works of literature, all articles using MCDM methods to solve energy system-related problems are collected and analyzed by conducting a bibliometric analysis. In the bibliometric analysis, the co-citation networks of journals, publications, and authors and the co-occurrence network of keywords are examined and evaluated. As a result, the MCDM has been proven a feasible method for energy system; multi-criteria decision analysis for energy system has been shown as a heated topic, and the most highly cited papers, authors, research countries/regions, and sources are identified. Acknowledgements The work described in this paper was supported by the grant from the Research Committee of The Hong Kong Polytechnic University under student account code RK22 and was also financially supported by The Start-up Grant of The Hong Kong Polytechnic University for New Employees (1-ZE8W) and the Departmental General Research Funds (UAFT) of Department of Industrial and Systems Engineers, The Hong Kong Polytechnic University (G-UAFT).

References 1. Ren J, Ren X, Shen W, Man Y, Lin R, Liu Y et al (2020) Industrial system prioritization using the sustainability-interval-index conceptual framework with life-cycle considerations. AIChE J. https://doi.org/10.1002/aic.16961 2. Wang Z, Xu G, Lin R, Wang H, Ren J (2019) Energy performance contracting, risk factors, and policy implications: identification and analysis of risks based on the best-worst network method. Energy. https://doi.org/10.1016/j.energy.2018.12.140 3. Ren J, Sovacool BK (2014) Quantifying, measuring, and strategizing energy security: determining the most meaningful dimensions and metrics. Energy. https://doi.org/10.1016/j.ene rgy.2014.08.083 4. Zhang L, Yu J, Sovacool BK, Ren J (2017) Measuring energy security performance within China: toward an inter-provincial prospective. Energy. https://doi.org/10.1016/j.energy.2016. 12.030 5. Groscurth HM, Bruckner T, Kümmel R (1995) Modeling of energy-services supply systems. Energy. https://doi.org/10.1016/0360-5442(95)00067-Q

22

R. Lin and J. Ren

6. Lin R, Man Y, Lee CKM, Ji P, Ren J (2020) Sustainability prioritization framework of biorefinery: a novel multi-criteria decision-making model under uncertainty based on an improved interval goal programming method. J Clean Prod. https://doi.org/10.1016/j.jclepro. 2019.119729 7. Lee HC, Ter CC (2018) Comparative analysis of MCDM methods for ranking renewable energy sources in Taiwan. Renew Sustain Energy Rev. https://doi.org/10.1016/j.rser.2018. 05.007 8. Kabak M, Daˇgdeviren M (2014) Prioritization of renewable energy sources for Turkey by using a hybrid MCDM methodology. Energy Convers Manag. https://doi.org/10.1016/j.enc onman.2013.11.036 9. Streimikiene D, Balezentis T, Krisciukaitien I, Balezentis A (2012) Prioritizing sustainable electricity production technologies: MCDM approach. Renew Sustain Energy Rev. https:// doi.org/10.1016/j.rser.2012.02.067 10. Øvergaard S (2008) Definition of primary and secondary energy. Stand Int Energy Classif Int Recomm Energy Stat 1–7 11. Lin R, Ren J (2020) Renewable energy and sustainable development. Renew Energy Sustain Dev 6:3. https://doi.org/10.21622/resd.2020.06.1.003 12. National Bureau of Statistics (2006) Manual of Energy Statistics 13. Lin R, Liu Y, Man Y, Ren J (2019) Towards a sustainable distributed energy system in China: decision-making for strategies and policy implications. Energy Sustain Soc. https://doi.org/ 10.1186/s13705-019-0237-9 14. Luo L, van der Voet E, Huppes G (2009) Life cycle assessment and life cycle costing of bioethanol from sugarcane in Brazil. Renew Sustain Energy Rev 13:1613–1619. https://doi. org/10.1016/j.rser.2008.09.024 15. Sharma A, Strezov V (2017) Life cycle environmental and economic impact assessment of alternative transport fuels and power-train technologies. Energy 133:1132–1141. https://doi. org/10.1016/j.energy.2017.04.160 16. Ahmad S, Tahar RM (2014) Selection of renewable energy sources for sustainable development of electricity generation system using analytic hierarchy process: a case of Malaysia. Renew Energy 63:458–466. https://doi.org/10.1016/j.renene.2013.10.001 17. Bakken D, Askerman A, Srivastava A, Panciatici P, Seewald M, Columbus F et al (2018) Towards enhanced power grid management via more dynamic and flexible edge computations. 2017 IEEE Fog World Congr. FWC 2017. https://doi.org/10.1109/FWC.2017.8368539 18. Rämä M, Wahlroos M (2018) Introduction of new decentralised renewable heat supply in an existing district heating system. Energy. https://doi.org/10.1016/j.energy.2018.03.105 19. Nielsen S, Thellufsen JZ, Sorknæs P, Djørup SR, Sperling K, Østergaard PA et al (2020) Smart energy aalborg: matching end-use heat saving measures and heat supply costs to achieve least-cost heat supply. Int J Sustain Energy Plan Manag. https://doi.org/10.5278/ijsepm.3398 20. Zhou M, Jin H, Wang W (2016) A review of vehicle fuel consumption models to evaluate eco-driving and eco-routing. Transp Res Part D Transp Environ. https://doi.org/10.1016/j.trd. 2016.09.008 21. Abe JO, Popoola API, Ajenifuja E, Popoola OM (2019) Hydrogen energy, economy and storage: review and recommendation. Int J Hydrogen Energy. https://doi.org/10.1016/j.ijh ydene.2019.04.068 22. Hosseini SE, Wahid MA (2016) Hydrogen production from renewable and sustainable energy resources: promising green energy carrier for clean development. Renew Sustain Energy Rev. https://doi.org/10.1016/j.rser.2015.12.112 23. Hao H, Liu Z, Zhao F, Li W (2016) Natural gas as vehicle fuel in China: a review. Renew Sustain Energy Rev. https://doi.org/10.1016/j.rser.2016.05.015 24. Liu W, Hu J, Li X, Sun Z, Sun F, Chu H (2018) Assessment of hydrate blockage risk in long-distance natural gas transmission pipelines. J Nat Gas Sci Eng. https://doi.org/10.1016/ j.jngse.2018.10.022 25. Sabine C (2014) The IPCC fifth assessment report. Carbon Manag. https://doi.org/10.4155/ cmt.13.80

Overview of Multi-criteria Decision Analysis …

23

26. Wang CN, Nguyen VT, Thai HTN, Duong DH (2018) Multi-criteria decision making (MCDM) approaches for solar power plant location selection in Viet Nam. Energies. https://doi.org/10. 3390/en11061504 27. Wu Y, Zhang T, Xu C, Zhang B, Li L, Ke Y et al (2019) Optimal location selection for offshore wind-PV-seawater pumped storage power plant using a hybrid MCDM approach: a two-stage framework. Energy Convers Manag. https://doi.org/10.1016/j.enconman.2019.112066 28. Ali T, Ma H, Nahian AJ (2019) An analysis of the renewable energy technology selection in the southern region of Bangladesh using a hybrid multi-criteria decision making (MCDM) method. Int J Renew Energy Res 29. Chalvatzis KJ, Malekpoor H, Mishra N, Lettice F, Choudhary S (2019) Sustainable resource allocation for power generation: the role of big data in enabling interindustry architectural innovation. Technol Forecast Soc Change. https://doi.org/10.1016/j.techfore.2018.04.031 30. Mastrocinque E, Ramírez FJ, Honrubia-Escribano A, Pham DT (2020) An AHP-based multicriteria model for sustainable supply chain development in the renewable energy sector. Expert Syst Appl. https://doi.org/10.1016/j.eswa.2020.113321 31. Solangi YA, Tan Q, Mirjat NH, Ali S (2019) Evaluating the strategies for sustainable energy planning in Pakistan: an integrated SWOT-AHP and Fuzzy-TOPSIS approach. J Clean Prod. https://doi.org/10.1016/j.jclepro.2019.117655 32. Alizadeh R, Soltanisehat L, Lund PD, Zamanisabzi H (2020) Improving renewable energy policy planning and decision-making through a hybrid MCDM method. Energy Policy. https:// doi.org/10.1016/j.enpol.2019.111174 33. Guo S, Zhang W, Gao X (2020) Business risk evaluation of electricity retail company in China using a hybrid MCDM method. Sustainability 12. https://doi.org/10.3390/su12052040 34. Lin R, Man Y, Ren J (2020) Life cycle decision support framework: method and case study. Life Cycle Sustain Assess Decis. https://doi.org/10.1016/b978-0-12-818355-7.00009-9 35. Rezaei J (2015) A systematic review of multi-criteria decision-making applications in reverse logistics. Transp Res Procedia. https://doi.org/10.1016/j.trpro.2015.09.030 36. Saaty RW (1987) The analytic hierarchy process-what it is and how it is used. Math Model. https://doi.org/10.1016/0270-0255(87)90473-8 37. Konstantinos I, Georgios T, Garyfalos A (2019) A decision support system methodology for selecting wind farm installation locations using AHP and TOPSIS: case study in Eastern Macedonia and Thrace region, Greece. Energy Policy. https://doi.org/10.1016/j.enpol.2019. 05.020 38. Kaya T, Kahraman C (2010) Multicriteria renewable energy planning using an integrated fuzzy VIKOR & AHP methodology: The case of Istanbul. Energy. https://doi.org/10.1016/j. energy.2010.02.051 39. Toossi A, Camci F, Varga L (2013) Developing an AHP based decision model for energy systems policy making. Proc IEEE Int Conf Ind Technol. https://doi.org/10.1109/ICIT.2013. 6505886 40. Saaty TL (2006) The analytic network process. Int Ser Oper Res Manag Sci 95:1–26. https:// doi.org/10.1007/978-1-4419-1153-7_32 41. Cayir Ervural B, Zaim S, Demirel OF, Aydin Z, Delen D (2018) An ANP and fuzzy TOPSISbased SWOT analysis for Turkey’s energy planning. Renew Sustain Energy Rev. https://doi. org/10.1016/j.rser.2017.06.095 42. Sakthivel G, Ilangkumaran M, Gaikwad A (2015) A hybrid multi-criteria decision modeling approach for the best biodiesel blend selection based on ANP-TOPSIS analysis. Ain Shams Eng J. https://doi.org/10.1016/j.asej.2014.08.003 43. Hou J, Zhou D (2011) Study on influence factors of distributed energy system based on DEMATEL and analysis of influence factors of (DES). Int J Nonlinear Sci 44. Bacudio LR, Benjamin MFD, Eusebio RCP, Holaysan SAK, Promentilla MAB, Yu KDS et al (2016) Analyzing barriers to implementing industrial symbiosis networks using DEMATEL. Sustain Prod Consum. https://doi.org/10.1016/j.spc.2016.03.001 45. Hsu CC, Liou JJH, Chuang YC (2013) Integrating DANP and modified grey relation theory for the selection of an outsourcing provider. Expert Syst Appl. https://doi.org/10.1016/j.eswa. 2012.10.040

24

R. Lin and J. Ren

46. Liu R, Sun H, Zhang L, Zhuang Q, Zhang L, Zhang X et al (2018) Low-carbon energy planning: A hybrid MCDM method combining DANP and VIKOR approach. Energies. https://doi.org/ 10.3390/en11123401 47. Rezaei J (2015) Best-worst multi-criteria decision-making method. Omega (United Kingdom) 53:49–57. https://doi.org/10.1016/j.omega.2014.11.009 48. Wan Ahmad WNK, Rezaei J, Sadaghiani S, Tavasszy LA (2017) Evaluation of the external forces affecting the sustainability of oil and gas supply chain using best worst method. J Clean Prod. https://doi.org/10.1016/j.jclepro.2017.03.166 49. van de Kaa G, Kamp L, Rezaei J (2017) Selection of biomass thermochemical conversion technology in the Netherlands: a best worst method approach. J Clean Prod. https://doi.org/ 10.1016/j.jclepro.2017.07.052 50. Ertay T, Kahraman C, Kaya I (2013) Evaluation of renewable energy alternatives using MACBETH and fuzzy AHP multicriteria methods: the case of Turkey. Technol Econ Dev Econ. https://doi.org/10.3846/20294913.2012.762950 51. Montignac F, Noirot I, Chaudourne S (2009) Multi-criteria evaluation of on-board hydrogen storage technologies using the MACBETH approach. Int J Hydrogen Energy. https://doi.org/ 10.1016/j.ijhydene.2008.09.098 52. Andrade GN, Alves LA, Andrade FVS, De Mello JCCBS (2016) Evaluation of power plants technologies using multicriteria methodology Macbeth. IEEE Lat Am Trans. https://doi.org/ 10.1109/TLA.2016.7430079 53. Wei L, Yuan Z, Yan Y, Hou J, Qin T (2016) Evaluation of energy saving and emission reduction effect in thermal power plants based on entropy weight and PROMETHEE method. Proc 28th Chinese Control Decis Conf CCDC 2016. https://doi.org/10.1109/CCDC.2016.7530970 54. Wang ZX, Li DD, Zheng HH (2018) The external performance appraisal of china energy regulation: an empirical study using a TOPSIS method based on entropy weight and Mahalanobis distance. Int J Environ Res Public Health. https://doi.org/10.3390/ijerph15020236 55. Wang Z, Xu G, Wang H, Ren J (2019) Distributed energy system for sustainability transition: a comprehensive assessment under uncertainties based on interval multi-criteria decision making method by coupling interval DEMATEL and interval VIKOR. Energy. https://doi. org/10.1016/j.energy.2018.12.105 56. Heo E, Kim J, Boo KJ (2010) Analysis of the assessment factors for renewable energy dissemination program evaluation using fuzzy AHP. Renew Sustain Energy Rev. https://doi.org/10. 1016/j.rser.2010.01.020 57. Liu SY, Lee RS (2019) Analysis of the dilemmas of solar energy application for Taiwan building with Fuzzy AHP approach. IOP Conf Ser Earth Environ Sci. https://doi.org/10. 1088/1755-1315/237/4/042006 58. Pamucar D, Gigovic L, Bajic Z, Janoševic M (2017) Location selection for wind farms using GIS multi-criteria hybrid model: an approach based on fuzzy and rough numbers. Sustain 9. https://doi.org/10.3390/su9081315 59. Opricovic S, Tzeng GH (2002) Multicriteria planning of post-earthquake sustainable reconstruction. Comput Civ Infrastruct Eng. https://doi.org/10.1111/1467-8667.00269 60. Wang Y, Zhang Y, Yang W, Ji H (2015) Selection of low-temperature phase-change materials for thermal energy storage based on the VIKOR method. Energy Technol. https://doi.org/10. 1002/ente.201402098 61. Zheng G, Wang X (2020) The comprehensive evaluation of renewable energy system schemes in tourist resorts based on VIKOR method. Energy. https://doi.org/10.1016/j.energy.2019. 116676 62. Mojaver P, Khalilarya S, Chitsaz A (2020) Multi-objective optimization and decision analysis of a system based on biomass fueled SOFC using couple method of entropy/VIKOR. Energy Convers Manag. https://doi.org/10.1016/j.enconman.2019.112260 63. Hwang C-L, Yoon K (1981) Multiple attribute decision making: methods and applications a state-of-the—art survey, vol 186. Springer Berlin Heidelberg, Berlin, Heidelberg. https://doi. org/10.1007/978-3-642-48318-9

Overview of Multi-criteria Decision Analysis …

25

64. Wang E, Alp N, Shi J, Wang C, Zhang X, Chen H (2017) Multi-criteria building energy performance benchmarking through variable clustering based compromise TOPSIS with objective entropy weighting. Energy. https://doi.org/10.1016/j.energy.2017.02.131 65. Vavrek R, Chovancová J (2019) Assessment of economic and environmental energy performance of EU countries using CV-TOPSIS technique. Ecol Indic. https://doi.org/10.1016/j. ecolind.2019.105519 66. Gao Y, Zhang L, Zhou D (2011) Performance evaluation of coal enterprises energy conservation and reduction of pollutant emissions base on GRD-TOPSIS. Energy Procedia. https:// doi.org/10.1016/j.egypro.2011.03.093 67. Deng J (1989) Introduction to grey system. J Grey Syst 68. Jeffrey Kuo CF, Su TL, Jhang PR, Huang CY, Chiu CH (2011) Using the Taguchi method and grey relational analysis to optimize the flat-plate collector process with multiple quality characteristics in solar energy collector manufacturing. Energy. https://doi.org/10.1016/j.ene rgy.2011.03.065 69. Li X, Wang Z, Zhang L, Zou C, Dorrell DD (2019) State-of-health estimation for Li-ion batteries by combing the incremental capacity analysis method with grey relational analysis. J Power Sources. https://doi.org/10.1016/j.jpowsour.2018.10.069 70. Xie Y, Mao Z (2012) Evaluation of residential energy-saving buildings based on grey relational analysis method. Procedia Eng. https://doi.org/10.1016/j.proeng.2012.01.456 71. Chen X, Gong Z (2017) DEA efficiency of energy consumption in China’s manufacturing sectors with environmental regulation policy constraints. Sustainability. https://doi.org/10. 3390/su9020210 72. Lee SK, Mogi G, Shin SC, Kim JW (2007) An AHP/DEA hybrid model for measuring the relative efficiency of energy efficiency technologies. IEEM 2007 2007 IEEE Int Conf Ind Eng Eng Manag. https://doi.org/10.1109/IEEM.2007.4419150 73. Wu Y, Wang J, Hu Y, Ke Y, Li L (2018) An extended TODIM-PROMETHEE method for waste-to-energy plant site selection based on sustainability perspective. Energy. https://doi. org/10.1016/j.energy.2018.05.087 74. Vujoševi´c ML, Popovi´c MJ (2016) The comparison of the energy performance of hotel buildings using promethee decision-making method. Therm Sci. https://doi.org/10.2298/TSCI15 0409098V 75. Schröder T, Lauven LP, Beyer B, Lerche N, Geldermann J (2019) Using PROMETHEE to assess bioenergy pathways. Cent Eur J Oper Res. https://doi.org/10.1007/s10100-018-0590-3 76. Roy B (1991) The outranking approach and the foundations of electre methods. Theory Decis 31:49–73. https://doi.org/10.1007/BF00134132 77. Cavallaro F (2010) A comparative assessment of thin-film photovoltaic production processes using the ELECTRE III method. Energy Policy. https://doi.org/10.1016/j.enpol.2009.09.037 78. Karagiannidis A, Perkoulidis G (2009) A multi-criteria ranking of different technologies for the anaerobic digestion for energy recovery of the organic fraction of municipal solid wastes. Bioresour Technol. https://doi.org/10.1016/j.biortech.2008.11.033 79. Peng H, Shen K, He S, Zhang H, Wang J (2019) Investment risk evaluation for new energy resources: an integrated decision support model based on regret theory and ELECTRE III. Energy Convers Manag. https://doi.org/10.1016/j.enconman.2019.01.015 80. ISO 14040 (2006) Environmental management–life cycle assessment—Principles and framework. Int Organ Stand 81. Ren J, Toniolo S (2018) Life cycle sustainability decision-support framework for ranking of hydrogen production pathways under uncertainties: an interval multi-criteria decision making approach. J Clean Prod 175:222–236. https://doi.org/10.1016/j.jclepro.2017.12.070 82. Chen CR, Huang CC, Tsuei HJ (2014) A hybrid MCDM model for improving GIS-based solar farms site selection. Int J Photoenergy. https://doi.org/10.1155/2014/925370 83. Sánchez-Lozano JM, Teruel-Solano J, Soto-Elvira PL, Socorro G-C (2013) Geographical information systems (GIS) and multi-criteria decision making (MCDM) methods for the evaluation of solar farms locations: case study in south-eastern Spain. Renew Sustain Energy Rev. https://doi.org/10.1016/j.rser.2013.03.019

26

R. Lin and J. Ren

84. Alejano LR, Alonso E (2005) Integrating GIS and MCDM to Deal with landfill site selection. Int J Rock Mech Min Sci 85. Chan LK, Wu ML (2002) Quality function deployment: a literature review. Eur J Oper Res. https://doi.org/10.1016/S0377-2217(02)00178-9 86. Yazdani M, Chatterjee P, Zavadskas EK, Hashemkhani ZS (2017) Integrated QFD-MCDM framework for green supplier selection. J Clean Prod. https://doi.org/10.1016/j.jclepro.2016. 10.095 87. Yazdani M, Hashemkhani Zolfani S, Zavadskas EK (2016) New integration of MCDM methods and QFD in the selection of green suppliers. J Bus Econ Manag. https://doi.org/ 10.3846/16111699.2016.1165282 88. Akba¸s H, Bilgen B (2017) An integrated fuzzy QFD and TOPSIS methodology for choosing the ideal gas fuel at WWTPs. Energy. https://doi.org/10.1016/j.energy.2017.02.153 89. Dinçer H, Yüksel S, Martínez L (2019) Balanced scorecard-based analysis about European energy investment policies: a hybrid hesitant fuzzy decision-making approach with quality function deployment. Expert Syst Appl. https://doi.org/10.1016/j.eswa.2018.07.072 90. Stewart TJ (1991) A multi-criteria decision support system for r&d project selection. J Oper Res Soc. https://doi.org/10.1057/jors.1991.3 91. Pohekar SD, Ramachandran M (2004) Application of multi-criteria decision making to sustainable energy planning—a review. Renew Sustain Energy Rev. https://doi.org/10.1016/ j.rser.2003.12.007 92. VOSViewer (2018) VOSviewer : visualizing scientific landscapes. VOSviewer 2018 93. Yang Y, Ren J, Solgaard HS, Xu D, Nguyen TT (2018) Using multi-criteria analysis to prioritize renewable energy home heating technologies. Sustain Energy Technol Assess. https:// doi.org/10.1016/j.seta.2018.06.005 94. Ren J (2018) Sustainability prioritization of energy storage technologies for promoting the development of renewable energy: a novel intuitionistic fuzzy combinative distance-based assessment approach. Renew Energy 121:666–676. https://doi.org/10.1016/j.renene.2018. 01.087 95. Ren J, Ren X (2018) Sustainability ranking of energy storage technologies under uncertainties. J Clean Prod 170:1387–1398. https://doi.org/10.1016/j.jclepro.2017.09.229 96. Ren J (2018) Multi-criteria decision making for the prioritization of energy systems under uncertainties after life cycle sustainability assessment. Sustain Prod Consum. https://doi.org/ 10.1016/j.spc.2018.06.005 97. Yazdani-Chamzini A, Fouladgar MM, Zavadskas EK, Moini SHH (2013) Selecting the optimal renewable energy using multi criteria decision making. J Bus Econ Manag. https:// doi.org/10.3846/16111699.2013.766257 98. Yazdani M, Chatterjee P, Zavadskas EK, Streimikiene D (2018) A novel integrated decisionmaking approach for the evaluation and selection of renewable energy technologies. Clean Technol Environ Policy 20:403–420. https://doi.org/10.1007/s10098-018-1488-4 99. Yazdani-Chamzini A, Yakchali SH, Zavadskas EK (2012) Using A integrated MCDM model for mining method selection in presence of uncertainty. Econ Res Istraživanja. https://doi.org/ 10.1080/1331677x.2012.11517537 100. Azimi R, Yazdani-Chamzini A, Fouladgar MM, Zavadskas EK, Basiri MH (2011) Ranking the strategies of mining sector through ANP and TOPSIS in a SWOT framework. J Bus Econ Manag. https://doi.org/10.3846/16111699.2011.626552 101. Fouladgar MM, Yazdani-Chamzini A, Zavadskas EK (2011) An integrated model for prioritizing strategies of the Iranian mining sector. Technol Econ Dev Econ. https://doi.org/10. 3846/20294913.2011.603173

Development of Decision Support Tools: Methodological Framework of System Dynamics Energy Models Athar Kamal and Sami G. Al-Ghamdi

Abstract Decision support tools are essential instruments in making informed choices and decisions. These tools can be found in all organizations including governments and multi-corporations to even small businesses. The underlying knowledge used to build these can span from the areas of pure engineering and economics, etc., to diverse interdisciplinary fields. Systems thinking is one of these multi-disciplinary approaches which looks at problems and projects through a holistic manner. Energy projects involve broad socio-environ-economic concerns, and system thinking seems to be the right approach to take. One of the essential systems thinking tools for policy and decision-making is System Dynamics and it was used for over 50 years. System dynamics is based on causal variables and inferences that change over time. This book chapter introduces system dynamics as the tool and goes step by step to develop a methodological framework for the techniques required to approach problem solving using this tool. At the end of this chapter, you will be able to think of an energy problem in terms of system thinking and approach it to address issues that are unique and specific depending on the available knowledge and within the various constraints. Keywords Systems thinking · System dynamics · Decision support tool

1 Introduction Energy in its most theoretical definition is the ability or capacity to do work. It is found in the form of potential, kinetic, thermal, electrical, chemical and various other forms [1]. Generally, and in more broader terms, energy is often referred to as the medium that provides fuel for any sort of economic activity. It is the electricity that runs homes, offices and industries as well as the oil that moves cars, airplanes and ships. Primary energy is the raw form of a resource, such as wind, solar as well as the crude oil dug up from the ground. Transforming this into a more usable form gets us the secondary and tertiary energy, like electricity and petrol, etc. The production and A. Kamal · S. G. Al-Ghamdi (B) Division of Sustainable Development, College of Science and Engineering, Hamad Bin Khalifa University, Qatar Foundation, Doha, Qatar e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 J. Ren (ed.), Energy Systems Evaluation (Volume 2), Green Energy and Technology, https://doi.org/10.1007/978-3-030-67376-5_2

27

28

A. Kamal and S. G. Al-Ghamdi

consumption of energy, often called the energy supply and demand, widely varies between different countries as extraction, production, transportation and consumption are largely dependent on the geography, demographics (population, age, income, etc.), raw resources, geopolitics and economic structure. These constraints are what lead to policy and decisions makers strategies, in investing for short- and long-term solutions. To help make right decisions in the complex energy sector, stakeholders often rely on tools that assist and support them through rigorous scientific arguments. The modeling tools help in addressing energy planning and management concepts such as but not limited to decentralized planning, energy conservation, waste recycling, renewable energy and energy forecasting [2]. These modeling tools are based on advanced engineering, economic, environmental and social knowledge, often used in conjunction of each other. However, as the scale and scope of the project grow, the complexity of these tools increases, and the time and effort needed expand rapidly. Thus, depending on the restrictions and obstacles, usually in terms of time and money as well as the available technology, there are several assumptions that are made, and multiple tools are overlooked. Although specialized methodologies have been created to minimize any unforeseen outcomes, these are rarely able to give satisfactory outcomes in all conditions, given their development and implementation in developed countries as compared to developing ones [3].

1.1 Background of Decision Support Tools in Energy Modeling Decisions regarding the supply and demand of energy require thorough assessment of the economic, environmental and social aspects [4, 5], with strategic energy plans at urban and national scale having time spans of around 20 to 50 years [6]. Decision support tools are key instruments in making informed choices and decisions. The underlying knowledge used to build these can span from the areas of pure engineering and economics, etc., to diverse interdisciplinary fields. Decision support tools in energy planning are used as a part of the decision support process [7] and can be divided into top-down, bottom-up and hybrid energy models. Bottomup models are technical models developed and used by engineers and scientists, whereas top-down models are often used by economists and policy makers. Hybrid models as the name applies are a combination of top-down and bottom-up models to address the limitations of the two techniques. Top-down energy models include techniques such as input-output, econometrics, computable general equilibrium and system dynamics. Whereas, bottom-up models comprise modeling techniques such

Development of Decision Support Tools: Methodological Framework …

29

as partial-equilibrium, optimization, simulation and multi-agent models [8]. Systems thinking is one of these multi-disciplinary approaches which looks at problems and projects through a holistic manner.

2 System Dynamics as a Decision Support Tool In the case of expansive and all-encompassing analysis, energy system models are often used. Systems models are based on “Systems Thinking”, which is the skill, capacity and ability to see the world as a complex system [9, 10]. System dynamics is one of the systems thinking tool, which can be applied “to the behavior of human as well as physical and technical systems”, and it draws on the “cognitive and social psychology, economics and other social sciences” [9, 10]. It is the methodology to develop and analyze systems over time based on system thinking approach [11]. The methodology uses feedback loops and time delays to explore the changes in behaviors and trends, because of some policy interventions [12]. It is an efficient technique to look at energy projects from a broad socio-environ-economic spectrum and can be used to assess decisions in the long run (as is the case in energy projects).

2.1 SD Methodology The system dynamics methodology can be broken down into three main modeling phases. The first or initial stage involves identifying the set of problems, setting up the system boundary and determines the influencing factors. These are then used in the second or development phase of the models. With the help of the variables and factors identified in the first step, a system structure is developed, usually in the form of causal loop diagrams (CLD). The CLDs are the mind mapping structure of the system dynamic methodology, where the cause and effect of the different variables are modeled using causal links, with various different loops being intuitively formed as further complexity is added to the model. Once the CLDs are developed, the modeler moves on to the quantifying phase of the analysis, where stock and flow diagrams are developed. The stock and flow diagrams have data requirements, which when input can be then simulated to get the results. There are often times when data is not available, and here is the area where system dynamics methodology shines, as the whole methodology is a cyclical and iterative process [13]. Modelers can use causal links to infer and interpret decisions and outcomes, which would have been otherwise impossible. Once the modeler is satisfied with the initial model, validation can be done with the help of external data, models and/or experts. Any changes to the model can be incorporated in this phase after which the scenario or policy analysis stage starts. In this phase, different variables and measures are changed dependent on finding the outcome of different strategies and policies. The different outcomes

30

A. Kamal and S. G. Al-Ghamdi

are then compared and analyzed, with suggestions in the form of policy notes, briefs and documents developed and argued with the help of the model. It is essential to understand that even though system dynamics uses a holistic approach, there are limitations in the modeling technique. Flaws in mental models, scientific reasoning skills, not identifying discrepancies between outcomes and their expectations as well as not taking enough time to think and reflect on the outcomes [9, 10] can result in models giving erroneous results. If any of the problems are identified in the model, it is expected that the modeler is going to follow the methodology of iterating and improving the model to reduce any uncertainties.

2.2 Previous Use (Literature of SD) The use of system dynamics modeling as a decision support tool has been done extensively to model energy systems. Some of the early works in the field of energy that utilize the system dynamics tool can be seen in the 1970s [14, 15]. The tool gained popularity starting in the late 1980s [16–18] and throughout the 1990s [19–22] as more researchers began model development and analysis for miscellaneous policy plans. Utilization of system dynamics for different energy systems has continued since the 2000s, and newer research continues to emerge as the energy world continues to evolve with the exploration, advancements, growth and progress of newer methods and resources. The methodology is being used throughout the world, with studies evaluating energy projects and policies of various countries including but not limited to China [23], Canada [24], Ecaudor [25], Iran [26], Finland [12], Malaysia [27] and UK [28]. If we go into further detail, we see that system dynamic modeling has been developed to support integrated energy policymaking for efficiency measures, as Dyner et al. [20] show and analyze alternative technology diffusion, growth in electricity and gas consumption and the effects of pricing on energy demand by substituting household appliances for more efficient ones. Use of system dynamics and what-if scenarios in the short and medium term in the energy sector have also been captured by Caponio et al. [29], who analyze the implementation of energy-efficiency policies in residential buildings of a medium-sized Italian city, Bari. Furthemore, Bajracharya and Bhattarai [30] developed a system dynamics model that examines urban residential lighting electricity demand for Nepal, with the author measuring impacts of three different lighting measures (baseline, LED lamp and incandescent lamp removal). System dynamic modeling has also been used to estimate electricity demand, with Akhwanzada and Tahar [27] forecasting the electricity demand for Malaysia between 2011 and 2022, Feng et al. [23] modeling energy consumption of Beijing, China for the period 2005 to 2030 and Kamal et al. [4, 5, 31] estimating the electricity consumption for both the building sector as well as the water sector for Qatar. Additionally, building sectors have been studied by Yücel [32] and Filchakova et al. [33] who studied impact of energy transition in Switzerland and Netherland. Fazeli and Davidsdottir [34] studied the behavioral dynamics affecting the implementation of retrofitting measures for Danish housing.

Development of Decision Support Tools: Methodological Framework …

31

2.3 Tools Comparison The underlying principle of system dynamics methodology is based on the theory of nonlinear dynamics and feedback controls, and the mathematical underpinning is based on differential and integral calculus. This means that the tools (or software) that are used have similar interfaces. However, they differ based on the reach, expanse and specific use and needs of the customers, where Table 1 shows a qualitative Table 1 Popular system dynamics software tools (rating out of 5, 5 highest and 1 lowest) Software tool

User-friendly (Ease of learning for new users)

Analysis of Integration with multiple systems other programming languages

Free version available

Vensim

★★★★★

★

Vensim DLL can be used to control Vensim from Visual Basic, Delphi or any other programming language

✔ Vensim PLE for education

Ventity

★★★

★★★★ Through multiple entity models

As of now, Ventity ✔ cannot be 60-Day Trial dynamically integrated with other software

AnyLogic

★★

★★★★

Models can be integrated with external Java applications

✔ AnyLogic PLE for education

Stella

★★★★

★★ Some agent-based modeling (limited)

In some cases, models can be integrated with other applications through the use of command line

✔ 30-Day Trial

PowerSim

★★★★

★★★★ Through sub-models

While data can be read from excel, dynamic integration with other software is not available

✔ 30-Day Trial (Full Version) 6-Months (Express Version)

Simulink

★

★★★★★ Integration with other MATLAB code

Simulink’s main MATLAB environment means access to other software

✔ 30-Day Trial

32

A. Kamal and S. G. Al-Ghamdi

assessment of different available environments. The system dynamic methodology can be learned easily with the use of these tools, as most of them often have student versions (or trials) that are free and have sufficiently comprehensive feature sets. Vensim [35] from VENTANA systems is one of the most well-known system dynamics modeling environment. It has a free environment for early stage modelers and students and has a very intuitive and easy to learn interface as it is a declarative language (no programming). It can be accessed through other programs, but hierarchy and object orientation are not its strong suite. To compensate for these shortcomings, VENTANA systems have introduced a new program called Ventity [36], which supports object orientation and hierarchy. Because of these new implementations, compared to Vensim, Ventity is a bit difficult to learn. Anylogic [37] from the AnyLogic Company supports discrete event, agent-based and system dynamics modeling that can be used in any combination. Models can use GIS integration to assess systems like supply chains and logistic networks, and models can also be exported as standalone Java applications to facilitate easy use for clients. Because of the complexities mentioned, Anylogic can be difficult for new users who want to learn the complete package. PowerSim [38] is also a system dynamic tool that can be utilized to develop models, with a software development kit (SDK) available that can be used to create custom Web-based or desktop interfaces. Stella from isee systems [39] can also model system dynamics and discrete event models, but only some agent-based model. The Simulink environment in MATLAB can also be used to develop system dynamics model. While the interface is based on drag-and-drop feature, the more generalized nature of the software means that for new users, it can be difficult to learn. However, the seamless integration of MATLAB and Simulink means that commands in MATLAB can be called to access Simulink and Simulink can call variables from the main MATLAB environment.

3 Developing an Energy Decision Support Tool in SD As seen in the brief literature review in Sect. 3.1, as well as because of its methodological ease, system dynamics is an ideal tool to support the decisions in the complex energy sector as well as the economy. The interactions between different sectors of the economy along with the interplays and complexities of the energy sector can be modeled with the help of the causal loops, and the quantitative analysis can be implemented with the help of the stock and flow diagrams. Depending on the extent, scope and limitations of the project, the methodology below gives an overview and step-wise procedure in designing and building a decision support tool to address the energy sector in system dynamics modeling. It is important and vital to remember that system dynamics modeling is an iterative approach. As you learn and understand more of the problem and the resulting knowledge, you can go back and adjust your goals from the start and include new information in the decision support tool. The process described below is similar to the traditional SD modeling technique mentioned in Sect. 3.1 and literature [9, 10, 13] but is tailored specifically

Development of Decision Support Tools: Methodological Framework …

33

toward the complexities and scope of the energy sector. Where necessary, examples in the form of an outcome of system dynamics designing are presented along with each step, taken from [4, 5, 31], to enhance the comprehension of the technique.

3.1 Initial Phase The initial phase is often referred to as the boundary selection stage. • The first step in boundary selection involves the identification of the problem. Here, the modeler asks what is and why is the problem. In case of an energy systems model, this can be related to the supply and demand of energy. Modelers maybe looking to enhance the efficiency of any of the three (transportation, domestic (residential) and business, commerce and industrial) consumption sectors or the productivity of the supply side. It can also be used to find the future energy consumption demands, with which various socioeconomic and environmental decisions are made. Boundaries are necessary in system dynamic modeling as an unspecified boundary means that almost anything and everything can influence the variables defined in the model. – Kamal et al. [4, 5] defined the boundary selection as the study of efficiency in the building sector in Qatar and also addressed the transportation sector in [31], where the impact of introduction of efficient vehicles was addressed. Boundaries in both studies were set with external factor such as weather not included in the dynamic system. • Once the problem is articulated, the key variables and processes are identified. For the energy sector, it can include any processes, actions as well as stakeholders, etc. – Kamal et al. [4, 5] defined key variables of the building sector as the population to estimate the housing requirements, as well as the number and type of buildings, their age, energy demands and the construction and demolition rates. In [31] for the transportation sector, in addition to the population variable, the authors identified the vehicle sales/year, type of vehicle, size of vehicle and the vehicle scrappage age, etc. These variables served as the backbone of the simulation models and interacted amongst each other based on the causal links specified. • The problem identification stage and key variables then support the next step, which is the time horizon being considered for the problem at hand. Similar to boundary conditions, it is critical to pick the right time period as picking a very short or very long interval can result in biased or partial results. – In both cases, Kamal et al. [4, 5, 31] pick a simulation period from 2010 to 2050. This allows them to see the impact of turnover in the building and transport sector, which have relatively long time spans and it requires several decades to

34

A. Kamal and S. G. Al-Ghamdi

see the long-term impact of policy measures in both sectors. The model was not modeled after 2050, because it is uncertain what technologies might be available that impact the efficiencies in both the transportation and building sector. • Dynamic problem definition or reference models are then identified, which can serve to explain historical behavior of key concepts and variables and explain the expected functioning of the new approach. This helps the modeler in addressing some initial problems such as finding the right approach to address the concerned issues.

3.2 Model Development Phase • Before moving on to the development of the model structure, the initial hypothesis is generated, which contains information about the current theories of the problem at hand. Also known as the dynamic hypothesis, it is a statement that is the basis of the model which explains the underlying nature of the dynamic problem. – For the case of the building sector in Qatar, the dynamic hypothesis is: “The dynamic hypothesis for the state of Qatar is that long-term energy consumption in the building (residential and commercial) sector depends on the type, age, renovation period, and construction and demolition of building stocks” [4, 5]. This statement is very important as it gives a theoretical bottom line and a mental understanding of the modeling process. • With the help of all the previous steps (i.e., initial hypothesis, key variables, etc.), the system structure is developed. Alternatively called “Mapping”, the structure is developed mostly in the form of causal loop diagrams. Causal loop diagrams are built by linking one variable to another with the help of causal links. Positive or negative connotations are given at the end of those links, which suggest the positive or negative influence of one variable on another. Positive influence means that one variable is moving in the same direction as the other, e.g., increase in total water consumption in a city as population grows. Negative influence means that one variable is influencing the other in an opposite direction, e.g., as population increases, the total water available per capita decreases. Other tools can also be used in tandem of causal loop diagrams, such as “model boundary diagram”, “subsystem diagram” and “policy structure diagrams”, to better explain the underlying nature of the system. • The system structure, parameters, initial conditions, behavioral relationships and other identified boundaries and requirements are then used to develop the “stock and flow diagram”, which gives the quantitative aspect of the model developed. Any shortcomings and deficiencies typically found in the form of limited or no available data are then incorporated by iterating through the previous steps of

Development of Decision Support Tools: Methodological Framework …

35

the model development procedure. If the shortcomings are weakening and undermining the underlying structure of the model, then the whole modeling procedure should be repeated from the start, by coming up with better and suitable hypothesis. However, if the flaws are minor, new subsystems that can be represented with other causal parameters can be included. Additionally, appropriate assumptions can be made with arguments from the literature, previous models and available causal structures. • The stock and flow model are then run with initial conditions (quantitative data) and tested for consistency. This testing phase involves comparison to reference modes, which answer whether the model reproduces the behavior adequately for the required project. The robustness of the model is also put to test under extreme conditions to see whether the model shows realistic behavior when variables or parameters change because of drastic measures and conditions. Sensitivity analysis is also performed to see the behavior of the model resulting from the uncertainty in initial conditions, parameters, variables and boundary conditions, etc. Other tests such as checking the unit consistency and getting feedback from stakeholders and experts can also increase the validity and effectiveness of the model.

3.3 Policy Analysis Phase • The policy analysis phase is usually the end product or the outcome of the system dynamics decision support tool developed to address and resolve a problem. This phase starts once the modeler is satisfied by the model scope, timeframe and validity. Depending on the research, policy or project outcomes, scenario specifications about different environmental, parametric and variable adjustments are examined. With the help of these adjustments, policy designs are made in the form of decision rules and strategies that are anticipated to influence real-world conditions, for which the model is made. Because of the nature and type of model being developed (i.e., a decision support tool), multiple scenarios are established and simulated. This scenario analysis is also known as the “What-if” analysis, where different effects of the policies are compared and analyzed. Several visual plots are used in this case, to get a better understanding of the impacts of different interventions. – In the case of estimating the impacts of efficiency measures on the buildings sector in Qatar, Kamal et al. [4, 5] analyzed seven policy measures in addition to the business as usual scenario. These measures included the impact of the magnitude of renovation measures and implementation of building codes and combinations of both. For the transport sector [31], the authors estimated the impact of four policy scenarios in addition to the business as usual case. These measures were analyzed to see the impact of fuel efficiency restrictions on new cars as well as the move away from traditional fuel vehicles (petrol) to alternate vehicles such as hybrids and battery electric vehicles.

36

A. Kamal and S. G. Al-Ghamdi

4 Conclusions In this chapter, we have looked at the history and use of system dynamics in the energy sector. The use of system dynamics modeling as a decision support tool has been done extensively, and it provides appropriate results to analyze policy decisions in the short and especially long term. The tool is based on logical real-world problem solving and is easily understandable to anyone who has a slight understanding of what system thinking is. System thinking is the approach of looking at problems from a holistic point of view, with assumptions involved that facilitate quicker yet robust modeling. The technique is ideal for understanding the complex socio-environ-economic outcomes of the energy sector with several tools available. System dynamics is one of these tools that makes it easy to produce complex mechanisms. The system dynamics framework heavily involves both qualitative and quantitative modeling, to ensure that understanding of the system is the first priority. Three main steps can be followed, which start with theoretical considerations and move to mental models in the form of causal loop diagrams. These mental models are then used to develop stock and flow which provide policy and decision makers with the decision support tool that can be used for analyzing different choices based on different constraints.

References 1. Encyclopedia Britannica (2020) Energy| definition, types, & examples. In: Encyclopedia Britannica. https://www.britannica.com/science/energy 2. Jebaraj S, Iniyan S (2006) A review of energy models. Renew Sustain Energy Rev 10(4):281– 311 3. Bhattacharyya SC, Timilsina GR (2010) A review of energy system models. Int J Energy Sect Manage 4. Kamal A, Al-Ghamdi SG, Koc M (2019) Revaluing the costs and benefits of energy efficiency: a systematic review. Energy Res Soc Sci 54:68–84 5. Kamal A, Al-Ghamdi SG, Koç M (2019) Role of energy efficiency policies on energy consumption and CO2 emissions for building stock in Qatar. J Cleaner Prod 6. Moret S, Babonneau F, Bierlaire M, Maréchal F (2020) Decision support for strategic energy planning: a robust optimization framework. Eur J Oper Res 280(2):539–554 7. Sahota PS, Jeffrey P (2005) Decision-support tools: Moving beyond a technical orientation. Proc Institution Civ Eng—Eng Sustain 158(3):127–134. https://doi.org/10.1680/ensu.2005. 158.3.127 8. Herbst A, Toro F, Reitze F, Jochem E (2012) Introduction to energy systems modelling. Swiss J Econ Stat 148(2):111–135 9. Sterman JD (2000a) Business dynamics: sysems thinking and modeling for a complex world. McGraw-Hill 10. Sterman J (2000b) Business dynamics: systems thinking and modeling for a complex world. McGraw-Hill 11. Forrester JW (1989) The beginning of system dynamics. 1–16 12. Aslani A, Helo P, Naaranoja M (2014) Role of renewable energy policies in energy dependency in Finland: system dynamics approach. Appl Energy 113:758–765. https://doi.org/10.1016/j. apenergy.2013.08.015

Development of Decision Support Tools: Methodological Framework …

37

13. Barlas Y (2007) System dynamics: systemic feedback modeling for policy analysis. System 1(59) 14. Naill RF (1977) Managing the energy transition: a system dynamics search for alternatives to oil and gas 15. Sasser DW (1976) System dynamics model of national energy usage. STIN 78:13577 16. Ford A, Bull M (1989) Using system dynamics for conservation policy analysis in the Pacific Northwest. Syst Dyn Rev 5(1):1–16 17. Ford A, Bull M, Naill RF (1987) Bonneville’s conservation policy analysis models. Energy Policy 15(2):109–124 18. Sterman JD (1988) Modeling the formation of expectations: The history of energy demand forecasts. Int J Forecast 4(2):243–259 19. Bunn D, Dyner I (1996) Systems simulation to support integrated energy analysis and liberalised planning. Int Trans Oper Res 3(2):105–115 20. Dyner I, Smith RA, Peña GE (1995) System dynamics modelling for residential energy efficiency analysis and management. J Oper Res Soc 46(10):1163–1173. https://doi.org/10.2307/ 2584613 21. Ford A (1997) System dynamics and the electric power industry. Syst Dyn Rev J Syst Dyn Soc 13(1):57–85 22. Naill RF (1992) A system dynamics model for national energy policy planning. Syst Dyn Rev 8(1):1–19 23. Feng YY, Chen SQ, Zhang LX (2013) System dynamics modeling for urban energy consumption and CO2 emissions: a case study of Beijing, China. Ecol Model 252:44–52. https://doi. org/10.1016/j.ecolmodel.2012.09.008 24. Qudrat-Ullah H (2013) Understanding the dynamics of electricity generation capacity in Canada: A system dynamics approach. Energy 59:285–294 25. Robalino-López A, Mena-Nieto A, García-Ramos JE (2014) System dynamics modeling for renewable energy and CO2 emissions: A case study of Ecuador. Energy Sustain Dev 20:11–20. https://doi.org/10.1016/j.esd.2014.02.001 26. Vafa-Arani H, Jahani S, Dashti H, Heydari J, Moazen S (2014) A system dynamics modeling for urban air pollution: a case study of Tehran, Iran. Trans Res Part D: Trans Environ 31:21–36 27. Akhwanzada SA, Tahar RM (2012) Strategic forecasting of electricity demand using system dynamics approach. Int J Environ Sci Dev 3(4):6 28. Chi KC, Nuttall WJ, Reiner DM (2009) Dynamics of the UK natural gas industry: system dynamics modelling and long-term energy policy analysis. Technol Forecast Soc Chang 76(3):339–357 29. Caponio G, Massaro V, Mossa G, Mummolo G (2015) Strategic energy planning of residential buildings in a smart city: a system dynamics approach. Int J Eng Bus Manag 7:20 30. Bajracharya I, Bhattarai N (2016) System dynamics modeling of lighting electricity demand in the urban residential sector of Nepal. J Dev Adm Stud 23(1–2):33. https://doi.org/10.3126/ jodas.v23i1-2.15447 31. Kamal A, Al-Ghamdi SG, Koç M (2020) Modeling and understanding the impacts of efficiency measures on fleet fuel consumption in vehicle importing countries: a case study of Qatar. J Cleaner Prod 120619 32. Yücel G (2013) Extent of inertia caused by the existing building stock against an energy transition in the Netherlands. Energy Build 56:134–145. https://doi.org/10.1016/j.enbuild.2012. 09.022 33. Filchakova N, Wilke U, Robinson D (2009) Energy modelling of city housing stock and its temporal evolution. Iji 2(4):0–1 34. Fazeli R, Davidsdottir B (2015) Energy modeling of Danish housing stock using system dynamics. In: Proceedings of the 27th international conference of the system dynamics society, Boston 35. Ventana systems, inc. (2010) Vensim PLE 36. Ventana systems, inc. (2018) Ventity—Simulation software for complex systems (2.0) [Computer software]. http://ventity.biz/

38

A. Kamal and S. G. Al-Ghamdi

37. AnyLogic: Simulation Modeling Software Tools & Solutions for Business. (2020). https://www. anylogic.com/ 38. Powersim Software (2020). https://www.powersim.com/ 39. Isee systems (2020). https://www.iseesystems.com/

Multi-Criteria Decision Analysis Methods for Sustainability Assessment of Renewable Energy Systems and Its Potential Application to Sustainable STEM Education Jin Su Jeong and David González-Gómez Abstract Multi-criteria decision analysis (MCDA) method as a significant tool has been used in various and distinct disciplines. In the case of assessment in the sustainability of renewable biomass energy systems, it is important to identify and choose the location, which is an important concern due to geographically and spatially dispersion of biomass feedstocks. Here, the techniques via Geographic Information System (GIS) are controlling the suitable locations of biomass plants in terms of renewable energy systems. Thus, because of the characteristics it has, it is only few occasions to employ and operate its concept into the academic field, especially in Science, Technology, Engineering, and Mathematics (STEM) education. This work presents a collective approach of MCDA-GIS and the Fuzzy-DEcision-MAking Trial and Evaluation Laboratory (F-DEMATEL) to find optimal and favorable locations for biomass plants in the context of long-term sustainability and its possible application to sustainable STEM education. Along with the operational method of MCDAGIS/F-DEMATEL, the most optimal locations of biomass plant can be found through the main criteria, constraints and their weight coefficients assigned in a case study proposed. The case study area has a characteristic that forest and agriculture are the typical land uses. To check the reliability of this work, the weighted linear combination (WLC) method and a sensitivity analysis are applied. The outcomes of MCDA-GIS/F-DEMATEL show that the best locations for biomass plants can be placed close to forests and low transport costs zone. Among criteria and sub-criteria, the vegetation cover, agricultural area, transport cost and potential demand are the most optimal criteria of MCDA-GIS/F-DEMATEL model. This proposed method of renewable energy policy planning can be pertained to decision-making glitches in private sectors, at government levels and academic fields. Here, particularly, it is essential to employ the experience to the possible sustainable STEM education, which always necessitates continuous attainment of knowledge and distribution. Keywords Sustainability · MCDA · Renewable energy · STEM · Fuzzy-logic J. S. Jeong (B) · D. González-Gómez Departamento de Didáctica de las Ciencias Experimentales y Matemáticas, Universidad de Extremadura, Cáceres, Spain e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 J. Ren (ed.), Energy Systems Evaluation (Volume 2), Green Energy and Technology, https://doi.org/10.1007/978-3-030-67376-5_3

39

40

J. S. Jeong and D. González-Gómez

1 Operational Techniques in ESDSS The Multi-Criteria Decision Analysis-Geographic Information System (MCDAGIS) technique is a significant tool, which has been employed in numerous and distinct disciplines ([1]). With optimal criteria and sub-criteria for numerous types of installations on the basis of maps, it can be used to select locations and decrease environmental impacts and natural threats, disperse restricted resources, etc. [2–4]. Particularly, decision-makings on energy management are complicated and combine multiknowledge issues and foundations, such as physical, technological, political, social, environmental and economic ones [5–10]. The above-mentioned techniques can integrate some strategies’ effects to lessening threats in decision-makings and energy system management [11]. These suggested techniques can distinguish numerous evolving risks on the basis of a large number of preferred criteria and be utilized to select weights to generate suitable outcomes and classify possible locations for determining renewable biomass plants [12, 13]. Accordingly, these projected techniques together with the participatory approach can deliberate the decisions of a group and can decrease the subjectivity of incomplete important criteria and attain reliability among decision-makers as well because of evaluated criteria [9, 14, 15]. Consequently, the MCDA-GIS method is one of the most efficient approaches for selecting probable locations of potential renewable biomass plants [16–18]. Here, as a dominant instrument, the analytic hierarchy process (AHP) is used in the renewable bioenergy and biomass area for regulating its potential plant locations. Along with a earlier approach, the Fuzzy-DEcision-MAking Trial and Evaluation Laboratory (F-DEMATEL) technique was pertained to shape a organizational procedure among the criteria and their weights [19–21]. In the Geographic Information System (GIS), the procedure of the two above-mentioned approaches, as MCDA-GIS/FDEMATEL, is established on a systemic criterion that has flexibility characters. This organization rises the quality of problem assessment and the optimization of biomass plants’ location for renewable and sustainable energy managing and longrange sustainability [5–8, 22]. This method generates a holistic and seamless vision of certain decision-making procedures with specific compensations [23, 24]. Consequently, the MCDA-GIS/F-DEMATEL technique is a reliable and accurate valuation approach, which can support the energy spatial decision support system (ESDSS) on land usage matters. Also, it can be applied to manufacture optimal biomass plant management maps, energy planning maps, renewable energy management maps for long-range sustainability in policy/strategy decision-makings.

2 Geographical and Spatial Position with Operational Techniques Toward Sustainability The best location choice of biomass plants has been appraised by abundant researches at local, regional, national and international measures [25–27]. To regulate a biomass plant location, the dissemination of physical, geographic and spatial biomass feedstocks is very significant [28–30]. Aside from the levels and processes pertained,

Multi-Criteria Decision Analysis Methods for Sustainability …

41

such locations’ optimization is generally executed on the basis of a GIS [31, 32]. As a valuable tool, a GIS can assimilate geographic and spatial evidence along with statistical information to attain a graphic production that combines diverse resources and descriptions. A GIS, for instance, allows the presentation of statistical extractions and modest data schemes in maps, which can display more applicable results competed to traditional exposure [33]. A change of issues and methodologies that effect the biomass plants’ location optimization and decision-making processes have been pondered in the practical literature. In accordance with the comments of Viana et al. [34], they explored, mapped, regulated, and classified the locations of optimal thirteen biomass power facilities in Portugal, which can be defined by the GIS data along with feedstock radius. Herrea-Seara et al. [35] employed a different method with the MCDA to choose a best location for biomass power plants using the AHP and the GIS in the Spanish Granada Province. A comparable approach was projected in Valencia Province according to Prepina et al. [26]. In this province, it has also been deliberated together with diverse methodologies, which associate the GIS techniques and mathematical encoding procedures to improve a biomass plant location and size [36]. Also, Freppaz et al. [37] as a similar combination proposed an example in Italy. The mixed integer programming (MIP) was used by Schmidt et al. [38] that this case was applied to optimize bioenergy plant locations along with the evaluation of the combined heat and power (CHP) biomass possibility in Austria. Moreover, the MIP model has been employed in the USA to distinguish the probable locations of a biomass plant [5–8, 39, 40]. In Sweden, Leduc et al. [25] reflected the biomass price and accessibility based on the geographical and spatial location to classify the optimal biomass refineries’ locations. As the proposal of Singh et al. [41], at both the spatial and sequential stages, apposite planning and development are essential to locate biomass plants and implement effective bioenergy and biorefinery programs.

3 Sustainability Assessment of Renewable Energy Systems The glitches of management and planning in energy related with fast economic and social progress worldwide have been of critical issue into the level of local and national authorities and international governments [10, 42]. High energy burdens, environmental deprivation and climate change for many decades have shaped a new market occasion and attention in using organic biomass feedstocks and materials, which are originated from biological schemes for renewable and sustainable energy functions [9, 13, 43]. Using biomass materials to produce energy stipulates significant compensations and environmental remunerations, which are a segment of the European strategy to endorse renewable and sustainable energy sources [44]. In the European Union (EU), renewable energy sources can be outlined and constructed by a policy, especially through 2020, the European Parliament and Council. It was formulated a policy agenda on the basis of Directive 2009/28/EC [45, 46]. Each EU partner state favored by this directive fosters and encourages a national action strategy on

42

J. S. Jeong and D. González-Gómez

renewable and sustainable energy for implementing policies of 2011 to 2020 sustainable energy evolution [47]. The government of Spain has focused on a low-carbon economy and encouraged long-term sustainability. It was performed by not exploiting the previous Feed-In Tariff (FIT) procedure (Diversification and Energy Saving Institute [48, 49]. In a Spanish Province, Extremadura regional government has launched the Extremadura Bioenergy Plan 2015–2020 and engrossed on smart adaptation. It was a vital operating axis in the sector for energy elaboration (Diversification and Energy Saving Institute [48–51]. Bioenergy in Extremadura together with national energy policy is now contemplated the sector of energy with the greatest growth possibility. Particularly, forestland is occupied by 68% of the province is covered, and agricultural land is occupied by 30% of the province. More than 4 million tons of raw materials can be produced by this province, which can be altered into biomass materials [50]. This majority potential (96%) is presently not utilized that regional demand (45%) has affected to be substituted by imports. Subsequently, in many sectors in Extremadura, the regional government is increasing energy generation with biomass and biogas [51, 52]. Hence, renewable and sustainable energy administration and projection as a consistent and precise assessment is obligatory to define the optimal biomass plants’ locations and endorse long-term policy decisions’ sustainability.

4 Sustainability STEM Education Deliberating the present circumstances aforementioned, the goals, standards, movements and values of sustainable Science, Technology, Engineering, and Mathematics (STEM) education on the basis of the Decade of Education for Sustainable Development (DESD) of United Nations Educational, Scientific and Cultural Organization (UNESCO) in the United Nation (UN) endorse public awareness and follow to spread a life-long instruction and increase standing in various instructive areas [53, 54]. For educational background of human possible illumination and economic, biological and communal interdependence, Sterling mentioned that sustainable STEM education is an adjustment, which will start into transformative education [5–8, 55]. In the transformative education context, here Mezirow represented the educators’ obligation is to help learners that can complete their goals in a more reliable and autonomous manner [56–58]. Teaching/learning procedures and determinations in the pedagogical and cultural setting distillate on supporting learners in accordance with skills, values, data and a mind-set that can achieve as a transformation facilitator to sustainability [59]. Thus, the STEM education for sustainability development is linked with knowledge acceptance such as the standards and principles for sustainable education, while a certain area of research can associate with its individual measurements, procedures and competences and scientific and practical abilities [53, 58]. Particularly, as a teaching/learning procedure, flipped and e-learning scheme are pondering an appropriate educational style. It permits adaptable learner-focused instruction due to Information and Communication Technologies (ICTs), being online and virtual teaching/learning podiums, in sustainable STEM development education ([60, 61]).

Multi-Criteria Decision Analysis Methods for Sustainability …

43

Here, students in flipped e-learning programs as Hansen emphasized normally have better experience and knowledge awareness, which precede to optimistic and affirmative transformative education [60, 62–64]. Furthermore, students’ learning accomplishments and achievements by Paechter et al. stated are carefully associated to the flipped e-learning programs’ and systems’ characteristics owing to education structures’ flexibility and comprehension exchange, that is, multi-dimensional communications [65]. In addition to the ICTs and original information, flipped e-learning learning/teaching in sustainable STEM education can be of excellent connection in genuine life-long education. It is for sustainable development together with countless criteria and sub-criteria [66]. Nevertheless, in a higher education, sustainable STEM education is still an initial phase and rare solicitation actuality although they have performed numerous positions and shares in transforming civilizations by enlightening decision-makers [65, 67, 68]. It is for common e-learning occasions and is demanded to accomplish the efficiency and effectiveness of a research on particular flipped e-learning prototypes. Current researches available define the inspection and discussion in detail examination and valuation for STEM education for sustainable development via flipped/e-learning schemes in higher education acclimating the sustainable energy systems integrating operational methods [63, 64, 66, 68].

5 Assessment Approaches to Sustainable and Renewable Energy Systems and Its Potential Application to STEM Education The MCDA-GIS/F-DEMATEL can be showed as a conceptual framework of assessment approach (see Fig. 1). Along with a proposed case study, the assessment approach optimizes a biomass plant location and its potential application to STEM education. Here, renewable and sustainable energy planning and management can be promoted by the approach for long-term sustainability. As a certain outranking technique, the F-DEMATEL is pertained, and, successively, the weighted linear combination (WLC) and a sensitivity analysis are applied and examined to produce an optimization map for a biomass plant. ArcGIS 10.2 software is utilized to generate the transformation process and the optimization assignments and results on the basis of various data source [69, 70].

5.1 A Realized Case Study The selected study area has more than 4 million tons of raw materials that can be transformed into biomass and can be considered to apply the established methodology. In Extremadura Province, 68% of the land is forestland, and over 30% is farmland as shown in Table 1, and currently, 96% of this probable land is not used [50, 71]. Here, in

44

J. S. Jeong and D. González-Gómez

Fig. 1 MCDA-GIS/F-DEMATEL conceptual assessment outline approach

Table 1 Extremadura surface distribution as a case study area [71]

Category Agricultural zone

Territorial surface (km2 )

Territorial percentage (%)

9383.68

22.54

28,316.51

68.02

Urban and reservoir zone

666.46

1.60

Marginal agricultural zone

3267.92

7.84

41,634.57

100.00

Forest zone

Territorial Total

terms of geographical features, the case study area is circumscribed by the Provinces of Castilla and Leon and the foothills of the Bejar and Gredos Mountains. The area covers Agro-Silvo-Pastoral System (ASPS), which is a multifunctional system that is a system in land use. It comprises deciduous forest ruled by chestnut trees along with pre-existing agrarian actions at the identical location. With the current situation that 45% local request that is satisfied with foreign import, multi-criteria decision analysis methods for sustainability assessment of renewable energy systems and its potential application to sustainable STEM education are necessary to have an accurate and reliable evaluation [51, 52].

5.2 Criteria and Constraints Criteria’, sub-criteria’ and constraints’ selection as MCDA-GIS/F-DEMATEL first step can express a direct inspiration for potential location assessment that can show

Multi-Criteria Decision Analysis Methods for Sustainability …

45

a precise action, connection and/or capability in this question. In Table 2, all the criteria, sub-criteria and constraints for this research are described by the authors as decision-makers with a help with a group of experts. Particularly, GIS data as a real data matching their theme were decided on the basis of EU directives in regard with national and regional strategies and literature review. Here, socioeconomic, environ-sustainable and geo-physical criteria are divided into the twelve sub-criteria to find optimal and favorable locations for biomass plants in the context of long-term sustainability and its possible application to sustainable STEM education. Thus, the specific territory can express as the constraints that can limit and contain artificial and natural zones in the research proposed as shown in Table 2. These zones indicated should be conserved due to their distinct features that were protected by the Spanish government and the Extremadura parliament current laws.

5.3 MCDA-GIS/F-DEMATEL Method On the basis of criteria, sub-criteria and constrains selected, MCDA-GIS/FDEMATEL second step firstly uses the F-DEMATEL technique. This is an outranking method that can apply and compute criteria, sub-criteria and constraints coefficients to find optimal and favorable locations for biomass plants in the context of long-term sustainability and its possible application to sustainable STEM education in the AHP method context. Here, a fuzzy-logic is engaged to regulate and normalize the data of the criteria, sub-criteria and constraints. With the fuzzy-logic function and membership, all pixels can satisfy the necessities of fuzzy-logic that can be evaluated [72]. Three criteria, twelve sub-criteria and seven constraints on the basis of common 0 to 1 scale were enumerated to find optimal and favorable locations for biomass plants (0 signifies the least scale for suitability, and 1 signifies the most scale for suitability). Their weight and comparative significance should be determined in regard with a pair-wise comparison matrix (PCM). Here, the matrix of a consistency ratio (CR) can be produced by the PCM, and their values can be dogged to measure each matrix arithmetic consistency as shown in Eq. (1) [4]. CR =

CI RI

(1)

Furthermore, in pair-wise comparison in this equation, the consistency index (CI) can portion the discrepancies, and also, average of the CI assessments can produce the random index (RI) randomly [3, 73]. Less than 10% CR value is adequate for its application; however, if it is higher than 10%, decision-makers consider to change their evaluation [5–8, 74]. CI =

λmax − n n−1

(2)

46

J. S. Jeong and D. González-Gómez

Table 2 Criteria, sub-criteria and constraints measured for a biomass facility location Criteria

Sub-criteria

Criteria context

Socioeconomic

Transport cost

Spatial coverage arrangement of 0.148 biomass group and dissemination cost

Weight

Potential demand

Spatial coverage arrangement of energy ingestion and demand

0.069

Economic area

Spatial coverage arrangement of financial actions and population concentration

0.035

Site access

Spatial coverage arrangement of transportation grids, highways, local highways and railways

0.014

Validation

λmax = 4.247; CI = 0.082; CR = 0.091 T 7 > T 4 > T 6 > T 2 > T 3 > T 5. From the ranking result, it can be concluded that the method with the best sustainability performance is nuclear power (T 1). For the two biomass power systems, the route using miscanthus as raw material has greater economic advantages than that using wood as raw material, and therefore, the sustainability of T 7 is better than T 6 due to the high weight of economic criteria. At the same time, solar power is less sustainable because it is economically

8.887E-02

1.072E-01

1.114E-01

T5 (Solar)

T6 (Biomass w)

T7 (Biomass m)

6.785E-09

9.664E-09

1.746E-08

1.252E-08

3.762E-01

T 4 (Gas)

5.960E-10

4.252E-09

1.118E-02

T2 (Wind)

5.405E-10

C2 (kg CFC-11 eq./kWh)

T 3 (Coal) 1.068E + 00

6.194E-03

T1 (Nuclear)

C1 (kg CO2 eq./kWh)

3.094E-03

2.859E-03

4.323E-04

1.347E-04

1.777E-03

8.291E-05

4.391E-05

C3 (kg SO2 eq./kWh)

4.666E-01

2.822E-01

4.476E-03

4.108E-04

2.728E-02

3.741E-04

5.433E-04

C4 (m2 yr/kWh)

Table 4 Basic data of seven power generation systems (data from the UK) [95]

1.912E + 02 6.241E + 01 6.527E + 02 3.539E + 02

4.451E + 02

4.230E + 1.195E + 2.703E 01 01 + 01 1.120E + 6.000E + 7.382E 01 00 + 01 4.662E + 3.146E + 0.000E 02 01 + 00 1.680E + 1.056E + 8.402E 00 01 + 01 1.680E + 1.058E + 8.402E 00 00 + 01

3.680E + 02

1.094E + 3.670E + 0.000E 02 01 + 00

C7 (%) C8 (person-yrs/TWh) 8.083E + 01

C6 (£/MWh)

7.500E + 1.430E + 5.603E 01 01 + 00

C5 (£/MWh)

8.585E-02

6.618E-02

1.418E-01

5.244E-03

2.276E-01

7.356E-02

1.152E-01

C9 (kg DCB eq./kWh)

1.000E + 02

6.649E + 01

1.000E + 02

8.700E + 01

7.604E + 01

1.000E + 02

8.396E + 01

C10

Coupling LCSA and Multi-criteria Decision Analysis for Energy … 151

152

W. Li et al. 0.1296 0.1291

0.14 0.12

0.1242

0.1051 0.1028

0.1

0.0941

0.0807 0.0851

0.0804 0.0689

0.08 0.06 0.04 0.02 0 C1

C2

C3

C4

C5

C6

C7

C8

C9

C10

Fig. 4 Weight of each criterion

Table 5 Criteria data for each alternative after normalization T1

T2

T3

T4

T5

T6

T7

C1

1.0000

0.9953

0.0000

0.6515

0.9221

0.9049

0.9009

C2

1.0000

0.9967

0.7806

0.2920

0.0000

0.4608

0.6309

C3

1.0000

0.9872

0.4318

0.9702

0.8727

0.0770

0.0000

C4

0.9996

1.0000

0.9423

0.9999

0.9912

0.3955

0.0000

C5

0.8422

0.7681

0.9126

0.9795

0.0000

1.0000

1.0000

C6

0.6285

0.0000

0.6944

0.8613

0.1470

0.7334

1.0000

C7

0.9333

1.0000

0.6783

0.1214

1.0000

0.0000

0.0000

C8

0.0312

0.5177

0.2182

0.0000

1.0000

0.4938

0.6483

C9

0.5055

0.6928

0.0000

1.0000

0.3859

0.7260

0.6375

C10

0.5213

1.0000

0.2850

0.6121

1.0000

0.0000

1.0000

Table 6 Grey correlation coefficients of other criteria with respect Criterion 1 Grey correlation coefficient C2 C1

C3

C4

C5

C6

C7

C8

C9

C10

0.6420 0.6814 0.6646 0.6322 0.5708 0.6222 0.5791 0.6546 0.7309

unfavored. Coal power is assessed to have the lowest sustainability because of its obvious environmental disadvantages. According to the causality diagram for the criteria in Fig. 6, we can see that C 2 , C 5 , and C 6 can greatly affect other criteria. Therefore, if one wants to optimize the seven power generation systems, the first concern should be their ozone layer depletion, capital costs, and operation and maintenance costs. The case study verified that the developed sustainability assessment method by combining LCSA and MCDA can overcome several disadvantages of LCSA by integrating the information regarding the criteria from the environmental, economic,

Coupling LCSA and Multi-criteria Decision Analysis for Energy …

153

Table 7 Total grey relational matrix C1

C2

C1

0

0.6420 0.6821 0.6653 0.6331 0.5718 0.6230 0.5801 0.6556 0.7316

C3

C4

C5

C6

C7

C8

C9

C10

C2

0.6420 0

C3

0.6814 0.6814 0

C4

0.6646 0.6646 0.8332 0

C5

0.6322 0.6322 0.5708 0.6385 0

C6

0.5708 0.5708 0.5086 0.5230 0.7179 0

C7

0.6222 0.6222 0.7945 0.7773 0.5112 0.4907 0

C8

0.5791 0.5791 0.5250 0.5493 0.4602 0.4942 0.5899 0

C9

0.6546 0.6546 0.5706 0.5661 0.6530 0.6664 0.4631 0.6277 0

0.6238 0.6892 0.6736 0.6034 0.6782 0.5987 0.5906 0.5779 0.8332 0.5708 0.5086 0.7945 0.5250 0.5706 0.6912 0.6385 0.5230 0.7773 0.5493 0.5661 0.6258 0.7179 0.5112 0.4602 0.6530 0.5684 0.4907 0.4942 0.6664 0.5927 0.5899 0.4631 0.7066 0.6277 0.6337 0.6036

C10 0.7309 0.7309 0.6912 0.6258 0.5684 0.5927 0.7066 0.6337 0.6036 0 Table 8 Grey incidence matrix after processing C1

C2

C3

C4

C5

C6

C7

C8

C9

C10

C1

0

2.4624 2.7848 2.6493 2.3907 1.8973 2.3091 1.9642 2.5713 3.1828

C2

2.4624 0

C3

2.7791 2.7791 0

C4

2.6435 2.6435 4.0000 0

C5

2.3833 2.3833 1.8898 2.4341 0

C6

1.8893 1.8893 1.3893 1.5047 3.0722 0

C7

2.3027 2.3027 3.6886 3.5504 1.4096 1.2451 0

C8

1.9560 1.9560 1.5207 1.7165 1.0000 1.2729 2.0433 0

C9

2.5637 2.5637 1.8875 1.8518 2.5510 2.6587 1.0229 2.3471 0

2.3160 2.8417 2.7161 2.1512 2.7533 2.1137 2.0483 1.9465 4.0000 1.8898 1.3893 3.6886 1.5207 1.8875 2.8579 2.4341 1.5047 3.5504 1.7165 1.8518 2.3319 3.0722 1.4096 1.0000 2.5510 1.8701 1.2451 1.2729 2.6587 2.0656 2.0433 1.0229 2.9815 2.3471 2.3954 2.1531

C10 3.1771 3.1771 2.8579 2.3319 1.8701 2.0656 2.9815 2.3954 2.1531 0

Fig. 5 Sustainability performance of each alternative

154

W. Li et al.

Fig. 6 Causality diagram of evaluation criteria

and social aspects. In the developed method, an objective grey-DEMATEL approach is employed to determine the weights of criteria, which can significantly reduce the uncertainties of the evaluation process. The use of DEMATEL method identifying the causal relationship among the criteria, accurate identification of key criteria can help decisionmakers to make more realistic and effective decisions. The TOPSIS method that is used to rank the alternatives can also help decisionmakers to better identify the disadvantages of the systems. This method, which combines LCSA with MCDA and uses grey-DEMATEL, not only ensures the objective and effective evaluation results, but also introduces the idea of life cycle, which makes the evaluation more comprehensive and meets the requirements of the society and decision makers for sustainable development. The thought of life cycle is carried through the whole evaluation process, and the MCDA method is used to integrate the evaluation criteria to obtain comprehensive evaluation results, which also make the evaluation results concise and comparable. Consequently, the assessment framework by coupling LCSA with MCDA provides a new way to quantify and compares the sustainability of various energy systems.

5 Conclusions LCSA has recently emerged as a rapidly developing method for the sustainability assessment of complex systems and has attracted more and more attention from policymakers and researchers. However, several limitations, i.e., non-comparability of evaluation results and unclear relationship between guidelines have hindered the application of the LCSA framework. The coupling with MCDA can make the criteria of each dimension of sustainability evaluation better combine, obtain high comprehensive comparable sustainability evaluation results, and make the evaluation process

Coupling LCSA and Multi-criteria Decision Analysis for Energy …

155

clearer and the results more in line with the sustainability requirements of decisionmakers. The MCDA evaluation framework can process the data obtained by LCSA uniformly to obtain a comprehensive sustainability index. While LCSA is used to comprehensively reflect the sustainability performance of alternatives, the application of the MCDA method enables better data integration processing and providing guidance for subsequent sustainability optimization. In this chapter, the structure and applications of LCSA and MCDA are firstly introduced as well as its disadvantages. Then, the recent studies on the combination of MCDA with LCSA are discussed, indicating that several challenges are existing in the application of MCDA in LCSA. Then, a novel sustainability evaluation framework by combining LCSA and MCDA was proposed in this study. Furthermore, the sustainability evaluation of the power generation method is carried out by using greyDLCSA, which contains the idea of life cycle and is relatively objective. The sustainability scores of several commonly used power generation processes are obtained, and the evaluation results are explained. The accuracy of the evaluation method is proved by the analysis of the evaluation results. In addition, there are still many problems in the process of combining LCSA and MCDA, such as how to analyze the relationship between criteria, how to deal with less uncertainty of subjective criteria and so on. In the process of sustainability evaluation, the best alternative is not necessarily the one with the highest sustainability score, which should be judged comprehensively based on the actual situation and regional factors. Therefore, the research on sustainability evaluation, especially the development of a more perfect method combining LCSA and MCDA, is still in a continuous development process.

References 1. Zijp M, Heijungs R, van der Voet E, van de Meent D, Huijbregts M, Hollander A, Posthuma L (2015) An identification key for selecting methods for sustainability assessments. Sustainability 7(3):2490–2512 2. Foolmaun RK, Ramjeawon T (2012) Life cycle sustainability assessments (LCSA) of four disposal scenarios for used polyethylene terephthalate (PET) bottles in Mauritius. Environ Dev Sustain 15(3):783–806 3. Sala S, Farioli F, Zamagni A (2012) Progress in sustainability science: lessons learnt from current methodologies for sustainability assessment: part 1. Int J Life Cycle Assess 18(9):1653–1672 4. Akber MZ, Thaheem MJ, Arshad H (2017) Life cycle sustainability assessment of electricity generation in Pakistan: policy regime for a sustainable energy mix. Energy Policy 111:111– 126 5. Guinée J (2016) Life cycle sustainability assessment: what is it and what are its challenges? In: Clift R, Druckman A (eds) Taking stock of industrial ecology. Springer International Publishing, Cham, pp 45–68 6. Zamagni A (2012) Life cycle sustainability assessment. Int J Life Cycle Assess 17(4):373–376 7. Narula K, Reddy BS (2015) Three blind men and an elephant: the case of energy indices to measure energy security and energy sustainability. Energy 80:148–158

156

W. Li et al.

8. Sala S, Farioli F, Zamagni A (2012) Life cycle sustainability assessment in the context of sustainability science progress (part 2). Int J Life Cycle Assess 18(9):1686–1697 9. Costa D, Quinteiro P, Dias AC (2019) A systematic review of life cycle sustainability assessment: current state, methodological challenges, and implementation issues. Sci Total Environ 686:774–787 10. Seager TP, Linkov I (2008) Coupling multicriteria decision analysis and life cycle assessment for nanomaterials. J Ind Ecol 12(3):282–285 11. Zhou Z, Jiang H, Qin L (2007) Life cycle sustainability assessment of fuels. Fuel 86(1– 2):256–263 12. Kloepffer W (2008) Life cycle sustainability assessment of products. Int J Life Cycle Assess 13(2):89–95 13. Traverso M, Finkbeiner M, Jørgensen A, Schneider L (2012) Life cycle sustainability dashboard. J Ind Ecol 16(5):680–688 14. Ediger VS, ¸ Ho¸sgör E, Sürmeli AN, Tatlıdil H (2007) Fossil fuel sustainability index: an application of resource management. Energy Policy 35(5):2969–2977 15. Wulf C, Werker J, Ball C, Zapp P, Kuckshinrichs W (2019) Review of sustainability assessment approaches based on life cycles. Sustain 11(20) 16. Heijungs R, Huppes G, Guinée JB (2010) Life cycle assessment and sustainability analysis of products, materials and technologies. Toward Sci Framework Sustain Life Cycle Anal Polym Degrad Stab 95(3):422–428 17. Gangadharan P, Zanwar A, Zheng K, Gossage J, Lou HH (2012) Sustainability assessment of polygeneration processes based on syngas derived from coal and natural gas. Comput Chem Eng 39:105–117 18. Grubert E (2017) The need for a preference-based multicriteria prioritization framework in life cycle sustainability assessment. J Ind Ecol 21(6):1522–1535 19. Galán-Martín Á, Guillén-Gosálbez G, Stamford L, Azapagic A (2016) Enhanced data envelopment analysis for sustainability assessment: a novel methodology and application to electricity technologies. Comput Chem Eng 90:188–200 20. Wang J-J, Jing Y-Y, Zhang C-F, Zhao J-H (2009) Review on multi-criteria decision analysis aid in sustainable energy decision-making. Renew Sustain Energy Rev 13(9):2263–2278 21. Mardani A, Zavadskas EK, Streimikiene D, Jusoh A, Nor KMD, Khoshnoudi M (2016) Using fuzzy multiple criteria decision making approaches for evaluating energy saving technologies and solutions in five star hotels: a new hierarchical framework. Energy 117:131–148 22. Volkart K, Weidmann N, Bauer C, Hirschberg S (2017) Multi-criteria decision analysis of energy system transformation pathways: a case study for Switzerland. Energy Policy 106:155– 168 23. Maghrabie HF, Beauregard Y, Schiffauerova A (2019) Multi-criteria decision making problems with unknown weight information under uncertain evaluations. Comput Ind Eng 133:131–138 24. Kumar A, Sah B, Singh AR, Deng Y, He X, Kumar P, Bansal RC (2017) A review of multi criteria decision making (MCDM) towards sustainable renewable energy development. Renew Sustain Energy Rev 69:596–609 25. Polatidis H, Haralambopoulos DA, Munda G, Vreeker R (2006) Selecting an appropriate multi-criteria decision analysis technique for renewable energy planning. Energy Sources Part B 1(2):181–193 26. Mardani A, Jusoh A, Zavadskas EK, Kazemilari M, Ahmad UNU, Khalifah Z (2016) Application of multiple criteria decision making techniques in tourism and hospitality industry: a systematic review. Transform Bus Econ 15(1):192–213 27. Govindan K, Rajendran S, Sarkis J, Murugesan P (2015) Multi criteria decision making approaches for green supplier evaluation and selection: a literature review. J Clean Prod 98:66–83 28. Chen D, Cheng C-Y, Urpelainen J (2016) Support for renewable energy in China: a survey experiment with internet users. J Clean Prod 112:3750–3758

Coupling LCSA and Multi-criteria Decision Analysis for Energy …

157

29. Fang D, Shi S, Yu Q (2018) Evaluation of sustainable energy security and an empirical analysis of China. Sustain 10(5) 30. Jayaraman R, Colapinto C, La Torre D, Malik T (2017) A Weighted goal programming model for planning sustainable development applied to Gulf cooperation council countries. Appl Energy 185:1931–1939 31. Haimes YY, Hall WA (1977) Sensitivity, responsivity, stability and irreversibility as multiple objectives in civil systems. Adv Water Resour 1(2):71–81 32. Ren J, Toniolo S (2018) Life cycle sustainability decision-support framework for ranking of hydrogen production pathways under uncertainties: an interval multi-criteria decision making approach. J Clean Prod 175:222–236 33. Toniolo S, Mazzi A, Niero M, Zuliani F, Scipioni A (2013) Comparative LCA to evaluate how much recycling is environmentally favourable for food packaging. Resour Conserv Recycl 77:61–68 34. Schetke S, Haase D, Kötter T (2012) Towards sustainable settlement growth: a new multicriteria assessment for implementing environmental targets into strategic urban planning. Environ Impact Assess Rev 32(1):195–210 35. Hermann BG, Kroeze C, Jawjit W (2007) Assessing environmental performance by combining life cycle assessment, multi-criteria analysis and environmental performance indicators. J Clean Prod 15(18):1787–1796 36. Thabrew L, Wiek A, Ries R (2009) Environmental decision making in multi-stakeholder contexts: applicability of life cycle thinking in development planning and implementation. J Clean Prod 17(1):67–76 37. Wimmler C, Hejazi G, De Oliveira Fernandes E, Moreira C, Connors S (2015) Multi-criteria decision support methods for renewable energy systems on Islands. J Clean Energy Technol 3:185–195 38. Marler RT, Arora JS (2009) The weighted sum method for multi-objective optimization: new insights. Struct Multidiscip Optim 41(6):853–862 39. Triantaphyllou E, Mann SH (1989) An examination of the effectiveness of multi-dimensional decision-making methods: a decision-making paradox. Decis Support Syst 5(3):303–312 40. Wang M, Liu S, Wang S, Lai KK (2010) A weighted product method for bidding strategies in multi-attribute auctions. J Syst Sci Complexity 23:194–208 41. Winterfeldt D, Fischer (1973) Multi-attribute utility theory: models and assessment procedures 42. Ilangkumaran M, Sasirekha V, Anojkumar L, Raja M (2012) Machine tool selection using AHP and VIKOR methodologies under fuzzy environment. Int J Model Oper Manag 2:409 43. Li W, Ren X, Ding S, Dong L (2020) A multi-criterion decision making for sustainability assessment of hydrogen production technologies based on objective grey relational analysis. Int J Hydrogen Energy 44. Govindan K, Jepsen MB (2016) ELECTRE: a comprehensive literature review on methodologies and applications. Eur J Oper Res 250:1–29 45. Sun C-C (2010) A performance evaluation model by integrating fuzzy AHP and fuzzy TOPSIS methods. Expert Syst Appl 37(12):7745–7754 46. Awasthi A, Chauhan SS, Omrani H (2011) Application of fuzzy TOPSIS in evaluating sustainable transportation systems. Expert Syst Appl 38(10):12270–12280 47. Hesitant fuzzy linguistic VIKOR method and its application in qualitative multiple criteria decision making. IEEE Trans Fuzzy Syst 23 48. Gul M, Celik E, Aydin N, Taskin Gumu A, Guneri AF (2016) A state of the art literature review of VIKOR and its fuzzy extensions on applications. Appl Soft Comput 46:60–89 49. Liu H-C, You J-X, Fan X-J, Chen Y-Z (2014) Site selection in waste management by the VIKOR method using linguistic assessment. Appl Soft Comput 21:453–461 50. Taillandier P, Stinckwich S (2011) Using the PROMETHEE multi-criteria decision making method to define new exploration strategies for rescue robots 51. Amaral TM, Costa APC (2014) Improving decision-making and management of hospital resources: An application of the PROMETHEE II method in an emergency department. Oper Res Health Care 3(1):1–6

158

W. Li et al.

52. Wang J, Zionts S (2008) Negotiating wisely: considerations based on MCDM/MAUT. Eur J Oper Res 188(1):191–205 53. Chatzimouratidis A, Pilavachi P (2009) Sensitivity analysis of technological, economic and sustainability evaluation of power plants using the analytical hierarchy process. Energy Policy 37:788–798 54. De Luca AI, Iofrida N, Leskinen P, Stillitano T, Falcone G, Strano A, Gulisano G (2017) Life cycle tools combined with multi-criteria and participatory methods for agricultural sustainability: insights from a systematic and critical review. Sci Total Environ 595:352–370 55. Tarne P, Lehmann A, Finkbeiner M (2018) Introducing weights to life cycle sustainability assessment—how do decision-makers weight sustainability dimensions? Int J Life Cycle Assess 24(3):530–542 56. Niero M, Kalbar PP (2019) Coupling material circularity indicators and life cycle based indicators: a proposal to advance the assessment of circular economy strategies at the product level. Resour Conserv Recycl 140:305–312 57. Ekener E, Hansson J, Larsson A, Peck P (2018) Developing life cycle sustainability assessment methodology by applying values-based sustainability weighting—tested on biomass based and fossil transportation fuels. J Clean Prod 181:337–351 58. Ness B, Urbel-Piirsalu E, Anderberg S, Olsson L (2007) Categorising tools for sustainability assessment. Ecol Econ 60(3):498–508 59. Guitouni A, Martel J-M (1998) Tentative guidelines to help choosing an appropriate MCDA method. Eur J Oper Res 109(2):501–521 60. Munda G (2005) Measuring sustainability: a multi-criterion framework. Environ Dev Sustain 7(1):117–134 61. Kalbar PP, Birkved M, Nygaard SE, Hauschild M (2017) Weighting and aggregation in life cycle assessment: do present aggregated single scores provide correct decision support? J Ind Ecol 21(6):1591–1600 62. Rowley HV, Peters GM, Lundie S, Moore SJ (2012) Aggregating sustainability indicators: beyond the weighted sum. J Environ Manage 111:24–33 63. Mousavi-Nasab SH, Sotoudeh-Anvari A (2018) A new multi-criteria decision making approach for sustainable material selection problem: a critical study on rank reversal problem. J Clean Prod 182:466–484 64. Mufazzal S, Muzakkir SM (2018) A new multi-criterion decision making (MCDM) method based on proximity indexed value for minimizing rank reversals. Comput Ind Eng 119:427– 438 65. Kalbar P, Karmakar S, Asolekar S (2015) Selection of wastewater treatment alternative: significance of choosing MADM method. Environ Eng Manag J 14:1011–1020 66. Waas T, Hugé J, Block T, Wright T, Benitez-Capistros F, Verbruggen A (2014) Sustainability assessment and indicators: tools in a decision-making strategy for sustainable development. Sustainability 6(9):5512–5534 67. Greco S, Ehrgott M, Figueira J (2016) Multiple criteria decision analysis: state of the art surveys 68. Tziolas E, Bournaris T, Manos B, Nastis S (2018) Life cycle assessment and multi-criteria analysis in agriculture: synergies and insights, pp 289–321 69. Xu D, Li W, Dong L (2020) Chapter 11—A composite life cycle sustainability index for sustainability prioritization of industrial systems. In: Ren J, Toniolo S (eds) Life cycle sustainability assessment for decision-making, Elsevier, pp 225–252 70. Papilo P, Marimin E, Hambali IS (2018) Sitanggang, Sustainability index assessment of palm oil-based bioenergy in Indonesia. J Clean Prod 196:808–820 71. Zardari NH, Ahmed K, Shirazi S, Yusop Z (2015) Weighting methods and their effects on multi-criteria decision making model outcomes in water resources management 72. Ahlroth S, Nilsson M, Finnveden G, Hjelm O, Hochschorner E (2011) Weighting and valuation in selected environmental systems analysis tools—suggestions for further developments. J Clean Prod 19(2–3):145–156

Coupling LCSA and Multi-criteria Decision Analysis for Energy …

159

73. Atilgan B, Azapagic A (2016) An integrated life cycle sustainability assessment of electricity generation in Turkey. Energy Policy 93:168–186 74. Xu D, Li W, Shen W, Dong L (2019) Decision-making for sustainability enhancement of chemical systems under uncertainties: combining the vector-based multiattribute decisionmaking method with weighted multiobjective optimization technique. Ind Eng Chem Res 58(27):12066–12079 75. Miller P, de Barros AG, Kattan L, Wirasinghe SC (2017) Managing uncertainty in the application of composite sustainability indicators to transit analysis. Transp Res Proc 25:4003–4018 76. Mendoza Beltran A, Heijungs R, Guinée J, Tukker A (2015) A pseudo-statistical approach to treat choice uncertainty: the example of partitioning allocation methods. Int J Life Cycle Assess 21 77. Barfod MB, Salling KB (2015) A new composite decision support framework for strategic and sustainable transport appraisals. Transp Res Part A Policy Pract 72:1–15 78. Scholten L, Schuwirth N, Reichert P, Lienert J (2015) Tackling uncertainty in multi-criteria decision analysis—an application to water supply infrastructure planning. Eur J Oper Res 242(1):243–260 79. Sohn JL, Kalbar PP, Birkved M (2017) Life cycle based dynamic assessment coupled with multiple criteria decision analysis: a case study of determining an optimal building insulation level. J Clean Prod 162:449–457 80. Liu KF-R, Ko C-Y, Fan C, Chen C-W (2012) Combining risk assessment, life cycle assessment, and multi-criteria decision analysis to estimate environmental aspects in environmental management system. Int J Life Cycle Assess 17(7):845–862 81. Ren J, Ren X, Dong L, Manzardo A, He C, Pan M (2018) Multi-actor multi-criteria decision making for life cycle sustainability assessment under uncertainties. AIChE J 64 82. Akhtar S, Reza B, Hewage K, Shahriar A, Zargar A, Sadiq R (2014) Life cycle sustainability assessment (LCSA) for selection of sewer pipe materials. Clean Technol Environ Policy 17(4):973–992 83. De Luca A, Falcone G, Stillitano T, Iofrida N, Strano A, Gulisano G (2017) Evaluation of sustainable innovations in olive growing systems: a life cycle sustainability assessment case study in Southern Italy. J Cleaner Product 171 84. Hossaini N, Reza B, Sadiq R, Hewage K (2014) AHP-based life cycle sustainability assessment (LCSA) framework: a case study of 6-storey wood-frame and concrete-frame buildings in Vancouver. J Environ Plann Manage 58 85. Mahbub N, Oyedun AO, Zhang H, Kumar A, Poganietz W-R (2018) A life cycle sustainability assessment (LCSA) of oxymethylene ether as a diesel additive produced from forest biomass. Int J Life Cycle Assess 24(5):881–899 86. Opher T, Friedler E, Shapira A (2018) Comparative life cycle sustainability assessment of urban water reuse at various centralization scales. Int J Life Cycle Assess 24(7):1319–1332 87. Sou W, Chu A, Chiueh P-T (2016) Sustainability assessment and prioritisation of bottom ash management in Macao. Waste Manage Res J Int Solid Wastes Public Cleans Assoc ISWA 34 88. Wang J, Wang Y, Sun Y, Tingley DD, Zhang Y (2017) Life cycle sustainability assessment of fly ash concrete structures. Renew Sustain Energy Rev 80:1162–1174 89. Onat NC, Kucukvar M, Tatari O, Zheng QP (2016) Combined application of multi-criteria optimization and life-cycle sustainability assessment for optimal distribution of alternative passenger cars in U.S. J Cleaner Product 112:291–307 90. Aziz R, Chevakidagarn P, Danteravanich S (2016) Life cycle sustainability assessment of community composting of agricultural and agro industrial wastes 11:57–69 91. Nguyen TA, Kuroda K, Otsuka K (2017) Inclusive impacts assessment for the sustainability of vegetable oil-based biodiesel—part i: linkage between inclusive impact index and life cycle sustainability assessment. J Cleaner Product 66 92. Nzila C, Dewulf J, Spanjers H, Tuigong D, Kiriamiti H, van Langenhove H (2012) Multi criteria sustainability assessment of biogas production in Kenya. Appl Energy 93:496–506

160

W. Li et al.

93. Ostermeyer Y, Wallbaum H, Reuter F (2013) Multidimensional Pareto optimization as an approach for site-specific building refurbishment solutions applicable for life cycle sustainability assessment. Int J Life Cycle Assess 18 94. Ines R, Sobral P, Pecas P, Henriques E (2017) A sustainable business model to fight food waste. J Cleaner Product 177 95. Stamford L, Azapagic A (2014) Life cycle sustainability assessment of UK electricity scenarios to 2070. Energy Sustain Dev 23:194–211 96. Wulf C, Zapp P, Schreiber A, Marx J, Schlör H (2017) Lessons learned from a life cycle sustainability assessment of rare earth permanent magnets. J Ind Ecol 21(6):1578–1590 97. Zortea R, Maciel V, Passuello A (2017) Sustainability assessment of soybean production in Southern Brazil: a life cycle approach. Sustain Product Consump 13 98. Vinyes E, Oliver-Solà J, Ugaya C, Rieradevall J, Gasol C (2013) Application of LCSA to used cooking oil waste management. Int J Life Cycle Assess 18 99. Schau EM, Traverso M, Lehmann A, Finkbeiner M (2011) Life cycle costing in sustainability assessment—a case study of remanufactured alternators. Sustainability 3(11):2268–2288 100. Deshmukh SC (2013) Preference ranking organization method of enrichment evaluation (promethee). Int J Eng Sci Invention 2:28–34 101. Shao J, Taisch M, Ortega-Mier M (2016) A grey-decision-making trial and evaluation laboratory (DEMATEL) analysis on the barriers between environmentally friendly products and consumers: practitioners’ viewpoints on the European automobile industry. J Clean Prod 112:3185–3194 102. Xia X, Govindan K, Zhu Q (2015) Analyzing internal barriers for automotive parts remanufacturers in China using grey-DEMATEL approach. J Clean Prod 87:811–825 103. Govindan K, Khodaverdi R, Vafadarnikjoo A (2015) Intuitionistic fuzzy based DEMATEL method for developing green practices and performances in a green supply chain. Expert Syst Appl 42(20):7207–7220 104. Govindan K, Khodaverdi R, Vafadarnikjoo A (2016) A grey DEMATEL approach to develop third-party logistics provider selection criteria. Indus Manage Data Syst 116(4):690–722 105. Manzardo A, Ren J, Mazzi A, Scipioni A (2012) A grey-based group decision-making methodology for the selection of hydrogen technologies in life cycle sustainability perspective. Int J Hydrogen Energy 37(23):17663–17670 106. Bai C, Sarkis J (2013) A grey-based DEMATEL model for evaluating business process management critical success factors. Int J Prod Econ 146(1):281–292 107. Quezada LE, López-Ospina HA, Palominos PI, Oddershede AM (2018) Identifying causal relationships in strategy maps using ANP and DEMATEL. Comput Ind Eng 118:170–179 108. Su C-M, Horng D-J, Tseng M-L, Chiu ASF, Wu K-J, Chen H-P (2016) Improving sustainable supply chain management using a novel hierarchical grey-DEMATEL approach. J Clean Prod 134:469–481 109. Sakthivel G, Ilangkumaran M, Gaikwad A (2015) A hybrid multi-criteria decision modeling approach for the best biodiesel blend selection based on ANP-TOPSIS analysis. Ain Shams Eng J 6(1):239–256

Multi-criteria Decision Analysis Methods for Sustainability Assessment and Improvement of Energy Systems Under Uncertainties Xusheng Ren, Lichun Dong, and Jingzheng Ren

Abstract As one of the purposes to conduct sustainability assessment of energy systems, providing information on improving system sustainability for decisionmakers is crucial. However, the existing literature fails to provide comprehensive information to improve the sustainability from both short-term and long-term perspectives while considering the uncertainties existing in decision-making process. In order to narrow the research gap, the methodological framework for sustainability assessment of energy systems was established by integrating an improved hierarchical fuzzy Best–Worst Method (BWM), fuzzy Decision-making and Trial Evaluation Laboratory (DEMATEL) method and fuzzy Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) method. In which, the improved hierarchical fuzzy BWM was used to obtain the weights of the criteria, and further to provide the short-term information of improving the sustainability. While the cause-effect relationships among criteria were captured by using the fuzzy DEMATEL method to provide information for the long-term improvement of the system sustainability. The priorities of energy systems were finally acquired by using the fuzzy TOPSIS method. And the validity of the sustainability assessment framework of energy systems was verified using a case study of five combined cooling, heating, and power systems. Keywords Sustainability · Sustainability assessment · Multi-criteria decision-making · DEMATEL · Improved BWM · Fuzzy TOPSIS X. Ren · L. Dong (B) School of Chemistry and Chemical Engineering, National-Municipal Joint Engineering Laboratory for Chemical Process Intensification and Reaction, Key Laboratory of Low-Grade Energy Utilization Technologies & Systems of the Ministry of Education, Chongqing University, Chongqing 400044, China e-mail: [email protected] L. Dong Green Intelligence Environmental School, Yangtze Normal University, Fuling, Chongqing 408100, China J. Ren Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University, Hong Kong SAR, China © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 J. Ren (ed.), Energy Systems Evaluation (Volume 2), Green Energy and Technology, https://doi.org/10.1007/978-3-030-67376-5_7

161

162

X. Ren et al.

1 Introduction As the driving force of social development, energy plays an important role in industry, transportation, living and other fields [1, 2]. However, the excessive consumption of traditional fossil fuels (i.e. coal, oil and natural gas) had brought severe problems of resource depletion, climate change and environmental pollution in the past decades [3, 4]. Such increasingly environmental and social problems aroused by the usage of traditional energy resources had spurred concerns about sustainability and sustainable development, which leaded to the numerous studies of renewable energy systems and sustainability assessment of energy systems [5–8]. However, the sustainability assessment of energy systems is demonstrated to be a complex task [9], within the corresponding environmental, economic, social, technological, and political impacts should be systematically and comprehensively evaluated with the aim of providing information to compare with other systems or to improve the performance of system sustainability [10]. Thus, the multi-criteria decision analysis (MCDA) or multi-criteria decision-making (MCDC) methods [11, 12] as effective tools to tackle the multiple conflicting goals and complex problems had presented huge advantages compared with the single index methods such as thermodynamic method [2], and they had been extensively applied to help the decision-makers/stockholders select the most sustainable or preferable energy systems through eliciting the preferences and modeling the aggregating performances using such preferences [1, 3, 7, 13–16]. For example, Ghenai et al. [4] prioritized four renewable energy technologies by an hybrid MCDM method which consists of an extend Step-wise Weight Assessment Ratio Analysis (SWARA) and Additive Ratio Assessment (ARAS) method for determining the weights of the criteria and prioritizing the above four alternatives, respectively. Maxim [17] ranked 14 electricity generation technologies using a weighted sum multi-attribute utility method based on 10 sustainability indicators from the life cycle perspective. Ren et al. [18] integrated extension theory and analytic hierarchy process (AHP) to analyze the sustainability of ten hydrogen supply chains. Despite the great contributions to the sustainability assessment of energy systems in the above studies, there is a big drawback of the inability to deal with the uncertainties existing in the inventory data and decision makers’ preferences, which will result in the suboptimal or error solutions [19]. Thus, various uncertainty processing techniques, i.e., fuzzy theory [2, 20–24], interval approach [2, 25–28], grey theory [20, 21, 29], were integrated into the above MCDA methods to obtain more accurate results. For example, Ren and Lützen [23] used a MCDM method integrating fuzzy AHP and Dempster-Shafer (DS) theory to prioritize the alternative energy sources for shipping based on sustainability performance, within, the trapezoidal fuzzy numbers were used to express the ambiguity and subjectivity in human judgements. He et al. [26] established a MCDM model under uncertainty using interval analytic network process (ANP) and interval technique for order of preference by similarity to ideal solution (TOPSIS) to rank five alternative combined cooling, heating, and power (CCHP) systems according to their sustainability. While, in the work of Ren and

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

163

Toniolo [28], fuzzy sets and interval numbers were used to expressed the ambiguity in decision-makers’ judgements and the uncertainty of decision-making data, respectively. In which, AHP and decision-making trial and evaluation laboratory (DEMATEL) method were used to determine the weights of criteria, and the ranking results were obtained by evaluation based on distance from average solution (EDAS) method. Ren et al. [30] proposed a two-stage MCDM method based on the fuzzy Best–Worst Method (BWM), fuzzy TOPSIS, and an novel Preference Ranking Linear Programming Method to prioritize eight hydrogen production technologies. As demonstrated above, one of the purposes of conducting sustainability assessment of energy systems is to provide information on improving the system sustainability. The methods with criteria independence hypothesis (for example, AHP) can provide the short-term information to improve system sustainability, such as improving the performances of criteria with higher weights. By contrast, the longterm information can be provided by the DEMATEL method which considers the dependences, interactions or cause-effect relationships among criteria and can identify the crucial cause factors [31]. For example, Ren et al. [32] identified the key driving criteria and cause-effect relationships using DEMATEL method, and in the following work of [18], the weights of the criteria were calculated by AHP. However, there is no study integrating such two approaches under uncertainty to provide a more accurate and comprehensive information for the improvement of sustainability of energy systems. Thus, the aim of this chapter is to establish a MCDM framework under uncertainty for the prioritization the energy systems with the ability to identify the contextual relationships of the criteria, thus providing more integrated information for the sustainable improvement of energy systems. In particular, an improved hierarchical fuzzy BWM was developed to determine the weights of the criteria, while, the cause-effect relationships among criteria was identified by fuzzy DEMATEL method. At last, the fuzzy TOPSIS method was used to prioritize the energy systems according to sustainability performance. The rest of the Chapter is organized as follows: this chapter firstly described the establishment of the proposed MCDM framework for prioritizing the energy systems. Then, the validity of the proposed MCDM framework was verified by a case study of five CCHP systems as well as the main results and discussion. Finally, this chapter is concluded.

2 Method In this section, the MCDM framework was established. In which, the weights of the criteria were calculated using an improved hierarchical fuzzy BWM, with the aim to provide the information of improving the system sustainability in the short term by improving the performance of criteria with high weights; while, the contextual relationships among criteria were determined by the fuzzy DEMATEL method for identifying the cause and crucial criteria, so as to provide the long-term information. At

164

X. Ren et al.

Fig. 1 Methodological framework

last, the fuzzy TOPSIS method was introduced to obtain the ranking results of energy systems according to the sustainability performance. The overall methodological framework can be seen in Fig. 1.

2.1 Hierarchical Fuzzy BWM The BWM was firstly proposed by Rezaei [33] based on a particularly structured comparison method with the advantages of requiring less information and more consistency [34, 35]. While, in order to tackle the ambiguity and subjectivity in human judgements, some fuzzy BWMs were proposed [36–39]. For example, Guo and Zhao [39] proposed a fuzzy BWM with nonlinear constrains and the weights were obtained by graded mean integration representation method. In the model proposed by Hafezalkotob and Hafezalkotob [38], the fuzzy linear constrains were solved by converting them into crisp equivalents. Differently, Karimi et al. [36] proposed a fully fuzzy BWM considering the fuzzy objectives and fuzzy constrains. Amiri et al. [37] established a fuzzy BWM framework by combining the BWM and fuzzy preference programming. While, in this chapter, a linear fuzzy BWM model with fuzzy parameters was established and solved by the method proposed by Jiménez et al. [40], meanwhile, the weights of the criteria and sub-criteria were obtained all at once using a hierarchical model which is suitable to the real-life decisionmaking processes based on the work of Tabatabaei et al. [41]. The model has the following advantages: (1). It can deal with the uncertainty existing in the data inventory and human judgment. (2). The solving method has a strong theoretical basis while maintaining the linearity of the model. (3). It can deal with the hierarchical decision-making process of energy systems more conveniently. The fuzzy BWM can be divided into the following four steps considering the linear model of BWM based on the work of Rezaei [34] and Tabatabaei et al. [41]:

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

165

Step 1: Determining the best and the worst criterion, denotes by C B and C W . In this step, the best and worst criterion which means the most and least desirable/important criterion were selected by decision-maker. Step 2: Determining the Best-to-Others (BO) vector and the Others-to-Worst (OW) vector. Firstly, the BO and OW vectors were obtained in linguistic terms, which were then transformed into the vectors described by triangular fuzzy numbers (TNFs) according to the principles of transformation in Table 1 [38]. Noteworthily, the definition and operational laws of TFNs in this chapter follow the work of Ren et al. [42]. The BO vector in Eq. (1) means the relative preferences of the best criterion C B with respect to all the other criteria, while the OW vector Eq. (2) indicates the relative preferences of all the other criteria with respect to the worst criterion of C W . BO = (a˜ B1 ,a˜ B2 , . . . ,a˜ Bn )

(1)

OW = (a˜ W 1 ,a˜ W 2 , . . . ,a˜ W n )

(2)

U where a˜ B j = (a BL j ,a BMj ,a UB j ), j = 1,2, . . . ,n and a˜ j W = (a LjW ,a M j W ,a j W ), j = 1,2, . . . ,n are TFNs.

Table 1 The linguistic variables and transformation principles for fuzzy BWM [38]

Linguistic scales

Membership function

Just equal (JE)

(1,1,1)

Equally important (EI)

(1,1,3)

Weakly important (WI)

(1,3,5)

Fairly important (FI)

(3,5,7)

Very important (VI)

(5,7,9)

Absolutely important (AI)

(7,9,9)

166

X. Ren et al.

Step 3: Acquiring the optimal criteria weights (w1 ,w2 , . . . ,wn ) by solving the fuzzy linear programming model as shown in Eq. (3). min ξ L  ⎧ w B − a˜ B j w j  ≤ ξ L ⎪ ⎪ ⎪   ⎪ ⎪ w j − a˜ j W wW  ≤ ξ L ⎪ ⎪ ⎨ n  s.t. ⎪ wj = 1 ⎪ ⎪ ⎪ ⎪ j=1 ⎪ ⎪ ⎩ w j ≥ 0, j = 1,2, . . . ,n

(3)

Within, the fuzzy model in Eq. (3) can be converted into an equivalent auxiliary crisp model under a given feasibility degree of α based on the work of Jiménez et al. [40], as shown in Eq. (4). Where, E 1 and E 2 mean the lower and upper limits of the expected interval. min ξ L ⎧ a a (1 − α)E 2 B j + α E 1 B j w j − w B + ξ L ≥ 0 ⎪ ⎪ ⎪  a a ⎪ ⎪ (1 − α)E 2 B j + α E 1 B j w j − w B − ξ L ≤ 0 ⎪ ⎪ ⎪  ⎪ a a ⎪ ⎪ (1 − α)E 2 jw + α E 1 jw ww − w j + ξ L ≥ 0 ⎪ ⎪ ⎪  ⎪ ⎪ (1 − α)E a jw + α E a jw ww − w j − ξ L ≤ 0 ⎪ 2 1 ⎪ ⎨ n  s.t. ⎪ w j =1 ⎪ ⎪ ⎪ ⎪ j=1 ⎪ ⎪  

⎪ aB j L a ⎪ ⎪ E 1 = a B j + a BMj 2,E 2 B j = a BMj + a UB j 2 ⎪ ⎪ ⎪  



a a ⎪ U ⎪ 2,E 2 jw = a M 2 E 1 jw = a Ljw + a M ⎪ jw jw + a jw ⎪ ⎪ ⎩ w j ≥ 0, j = 1,2, . . . ,n

(4)

Then, the hierarchy of decision-making process should be added to the fuzzy model shown in Eq. (3) to better express the real-life problems. Firstly, considering a three-level sustainability assessment system which is the commonly used, in which, sustainability in the top level is the overall objective, which contains n criteria in the second level. And each criterion j ( j ∈ 1, 2, . . . , n) has K j sub-criteria in the third level. Thus, the hierarchical fuzzy BWM can be expressed as follows [41]:

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

min Z = ξ0L +

167

1 L ξ n j j

⎧ ⎧ a a (1 − α)E 2 B j + α E 1 B j w j − w B + ξ0L ≥ 0 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ (1 − α)E a B j + α E a B j w j − w B − ξ L ≤ 0 ⎪ ⎪ 0 2 1 ⎪ ⎪ (A)  a jw a jw ⎪ L ⎪ ⎪ w + α E − w + ξ − α)E (1 ⎪ ⎪ W j 0 ≥0 2 1 ⎪ ⎪ ⎪ ⎪  ⎪ ⎩ a a ⎪ ⎪ (1 − α)E 2 jw + α E 1 jw wW − w j − ξ0L ≤ 0 ⎪ ⎪ ⎪

 ⎧ ⎪ j j ⎪ a Bk a Bk ⎪ j j ⎪ ⎪ ⎪ (1 − α)E 2 + α E 1 wk − w B + ξ jL ≥ 0 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪

 ⎪ ⎪ j j ⎪ ⎪ ⎪ a Bk a Bk ⎪ j j ⎪ ⎨ ⎪ (1 − α)E 2 + α E 1 wk − w B − ξ jL ≤ 0 ⎪ ⎪ ⎪ (L) ⎪  ⎪ ⎪ ⎨ ⎪ j j ⎪ akW akW ⎪ ⎪ (1 − α)E 2 + α E 1 wWj − wkj + ξ jL ≥ 0 s.t. ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎩ a a j j ⎪ ⎪ (1 − α)E 2 jw + α E 1 jw wW − wk − ξ jL ≤ 0 ⎪ ⎪ ⎪ ⎧ j ⎪ j ⎪ ⎪ ⎪ μk = w j wk , j = 1,2, . . . ,n,k = 1,2, . . . ,K j ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ n ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ w j =1,w j ≥ 0, j = 1,2, . . . ,n ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ j=1 ⎪ ⎪ ⎪ (B) K ⎪ ⎪ j ⎪ ⎪  ⎪ ⎪ ⎪ ⎪ j ⎪ ⎪ wk =1 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ k=1 ⎪ ⎪ ⎪ ⎪ ⎩ ⎩ j wk ≥ 0, j = 1,2, . . . ,n,k = 1,2, . . . ,K j

(5)

where the set A among equations in the constraints are used to obtain the relative importance of the criteria in the second level relative to the overall objective (i.e., w j ), while the set L among equations were established to acquire the local weights j of sub-criteria (i.e., wk ). And in set B, the global weights of the sub-criteria were j represented by μk , and other equations are boundary conditions. Step 4: Consistency check. The obtained Z ∗ (i.e., ξ L∗ and ξ jL∗ ) was considered as the consistency indicator based on the work of Rezaei [34], the magnitude of Z ∗ indicates consistency level of such fuzzy preference [38], and the closer Z ∗ to zero shows the better consistency.

168

X. Ren et al.

2.2 Fuzzy DEMATEL Method The fuzzy DEMATEL method was the commonly used tool to indicate the structure of interactions or cause-effect relationships among the criteria in the complex problems under uncertain using a visualization method which is easy to understand, as well as the ability to identify the crucial impact factors of the systems/problems [31, 43]. And it had been applied in product service system [44], energy systems [43, 45–47], public health [48], risk assessment [49] and other fields [50–52]. Meanwhile, the ability of the fuzzy DEMATEL method to find the key impact factors in the longterm prospective was verified through the work of Chou et al. [31]. Thus, the fuzzy DEMATEL method was used to acquire the cause-effect relationships among the criteria and identify the crucial impact criteria of sustainability of energy systems for improving the sustainability of energy systems in long-term prospective, and it can be divided into the following 6 steps based on the work of Chou et al. [31] and Mohammadfam et al. [52]: Step 1: Determining the initial fuzzy direct-relation matrix A˜ expressed by the linguistic terms and the corresponding fuzzy numbers using Table 2 [52].  A˜ = a˜ i, j n×n

(6)

  where a˜ i, j = ai,L j ,ai,Mj ,ai,U j , j = 1,2, · · · ,n is a TFN. Step 2: Obtaining the normalized fuzzy direct-relation matrix D˜ using the following Eqs. (7)–(9) [43]. Where k is the maximum upper limit of the fuzzy element in the i-th row of the initial fuzzy direct-relation matrix, thus the normalized fuzzy directrelation matrix D˜ can be obtained through dividing the initial fuzzy direct-relation matrix by k. ⎛ ⎞ n  k = max⎝ ai,U j ⎠ i

  D˜ = d˜i, j

n×n

Table 2 The linguistic variables and transformation principles for fuzzy DEMATEL [52]

(7)

j=1

  = a˜ i, j n×n k

(8)

Linguistic scales

Membership function

No influence (N)

(0,0,0.25)

Very low influence (VL)

(0,0.25,0.50)

Low influence (L)

(0.25,0.50,0.75)

High influence (H)

(0.50,0.75,1.00)

Very high influence (VH)

(0.75,1.00,1.00)

Multi-criteria Decision Analysis Methods for Sustainability Assessment …



d˜i, j = di,L j ,di,Mj ,di,U j , j = 1,2, . . . ,n

169

(9)

Step 3: Calculating the fuzzy total-relation matrix T˜ by Eqs. (10)–(13) [43].

  where D L = di,L j matrix.

 

T˜ = t˜i, j n×n , t˜i, j = ti,L j ,ti,Mj ,ti,Uj

(10)

−1   T L = ti,L j n×n = D L × I − D L

(11)

−1   T M = ti,Mj n×n = D M × I − D M

(12)

−1   T U = ti,Uj n×n = D U × I − D U

(13)

n×n

  , D M = di,Mj

n×n

  , D U = di,U j

n×n

, and I means the identity

Step 4: Obtaining the significant relation among the  criteria by transforming the  crisp crisp ˜ fuzzy matrix T into a crisp matrix T = ti, j using Eq. (14) and setting n×n

the significance threshold value, which means that the relation was considered as significant when crisp values is bigger than the threshold value [52]. TCrisp =

T L + 2 × T M + TU 4

(14)

Step 5: Calculating Rj and C j . Where, Rj is the sum of i-row which means the effect of i-criterion on other criteria, while C j means the sum of j-row column indicating the effect of other criteria on j-criterion. Ri =

n 

crisp

(15)

crisp

(16)

ti, j

j=1

Cj =

n 

ti, j

i=1

Step 6: Generating the cause-effect relationship map.

170

X. Ren et al.

Then, the values of Ri + C i and Ri − C i (when i = j) can be calculated, in which, Ri + C i means the degree of interactions among criteria which is named “Prominence”, and Ri − C i named “Relation” is used to identify whether the criterion belongs to the cause group or the effect group with the principles that the positive values belong to the cause group and vice versa [31]. Meanwhile, it holds the view that [47, 52]: (1). In the case of lower degree of interactions (Ri + C i less than the threshold value), it means the criteria are more independent and could be removed from the cause-effect diagram. (2). While, the criteria with higher degree of interactions (Ri + C i greater than the threshold value) are the core criteria of improving the system sustainability, and if the criteria are effect, i.e., Ri − C i < 0, it means that they are not the rooted criteria; on the contrary, if the criteria are cause, i.e., Ri − C i > 0, it indicates that the criteria are the core driving criteria which can improve the system sustainability in long term prospective.

2.3 Fuzzy TOPSIS Method In this section, the fuzzy TOPSIS method was described to obtain the final priority of energy systems after acquiring the weight of criteria and analyzing the causeeffect relationships among the criteria as introduced in Sects. 2.1 and 2.2. The fuzzy TOPSIS method used in this chapter can be concluded into eight steps based on the work of Chen [53]. Step 1: Determining the initial decision making matrix expressed by linguistic variables and transforming into the corresponding expression of fuzzy numbers according to Table 3 by requiring the decision-makers to evaluate the performances of the alternatives considering all attributes, i.e., the fuzzy decision-making matrix Eq. (18) considering m alternatives (A1 ,A2 , . . . ,Am ) and n criteria (C1 ,C2 , . . . ,Cn ), and x˜i j = (xiLj ,xiMj ,xiUj ) is a TFN representing the performance of the i-th alternative with respect to j-th criterion. C1 C2 · · · A1 x˜11 x˜12 · · · . A2 x˜21 x˜22 .. .. .. .. . . . . . . Am x˜m1 x˜m2 · · ·

Cn x˜1n x˜2n ,i= 1, 2, . . . ,m; j = 1, 2, . . . ,n .. .

(18)

x˜mn

Table 3 Linguistic variables and transformation principles for the fuzzy TOPSIS method [42] Linguistic Worst/WT Worse/WE Bad/BD Medium/MM Good/GD Better/BR Best/BT variables Fuzzy number

(0,1,1)

(1,1,3)

(1,3,5)

(3,5,7)

(5,7,9)

(7,9,9)

(9,10,10)

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

171

Step 2: Normalizing the initial decision-making matrix by employing Eqs. (19)– (21). In which, B and C means the set of beneficial and cost criteria, and the obtained normalized decision-making matrix is shown in Eq. (20).  y˜i j =

xiLj

xiUj

xiMj



U , U , U = max xiUj , j ∈ B , j ∈ B, xmax U i xmax xmax xmax   L L L xmin xmin xmin L y˜i j = , , = min xiLj , j ∈ C , j ∈ C xmin i xiLj xiMj xiUj

  Y˜ =  y˜i j m×n

(19)

(20) (21)

where Y˜ is the normalized decision-making matrix. Step 3: Obtaining the weights of the criteria w j , j = 1,2, . . . ,n using the method proposed in Sect. 2.1. Step 4: Obtaining the weighted fuzzy normalized decision-making matrix V˜ by incorporating the weights of the criteria into the normalized matrix using Eq. (22).   w1 y˜11 w2 y˜12    w y˜ w y˜     ˜ V = v˜ i j m×n = 1. 21 2. 22 ..  ..   w y˜ w y˜ 1 m1 2 m2

· · · wn y˜1n .. . wn y˜2n .. .. . .

· · · wn y˜mn

     ,i= 1, 2, . . . ,m; j = 1,2, . . . ,n    

(22)

  where v˜ i j = viLj ,viMj ,viUj is still a TFN. Step 5: Determining the fuzzy positive-ideal solution A+ and the fuzzy negative-ideal solution A− according to the Eqs. (23) and (24). 

A+ = v1+ ,v2+ , . . . ,vn+ ,v+j = max viUj , j = 1,2, . . . ,n

(23)



A− = v1− ,v2− , . . . ,vn− ,v−j = min viLj , j = 1,2, . . . ,n

(24)

i

i

Step 6: Calculating the distances to the fuzzy positive-ideal solution di+ and fuzzy negative-ideal solution di− as shown in Eqs. (25) and (26). di+ =

n  j=1

d(˜vi j ,v+j ),i = 1,2, . . . ,m

(25)

172

X. Ren et al.

di− =

n 

d(˜vi j ,v−j ),i = 1,2, . . . ,m

(26)

j=1



 where the distance between two TFNs a˜ = a L ,a M ,a U and b˜ = b L ,b M ,bU was calculated by Eq. (27).  1 L ˜ [(a − b L )2 +(a M − b M )2 +(a U − bU )2 ] d(a, ˜ b) = 3

(27)

Step 7: Obtaining the closeness coefficients CCi which reflects the priority of the alternatives regarding various criteria, thus to give the final ranking results, in which, higher closeness coefficient indicates a better performance of alternatives. CCi =

di+

di− + di−

(28)

3 Case Study In this study, the above proposed model was verified by prioritizing five alternative CCHP systems based on 14 criteria as shown in Fig. 2. In which, SE, MGT, ICE and SOFC means CCHP systems driven by Stirling engine, micro gas turbine, internal combustion engine and solid electrolyte fuel cell, respectively. While, C on means the conventional system which provides the cooling, heating and electricity by purchasing from the utility. The above 14 criteria are classified into four categories (i.e., economic aspect, environmental aspect, technological aspect and social aspect) according to the work of He et al. [26], and another two criteria ( employment opportunities (C 13 ) and contribution to energy security (C 14 )) have been added to the social aspect [30]. As for the detailed configuration and description of five alternative CCHP systems can be seen in the work of He et al. [26].

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

173

Fig. 2 Criteria system and alternative CCHP systems

3.1 Determining the Weights of Criteria The weights of criteria were obtained using the improved fuzzy BWM as described in Sect. 2.1. The best and worst criterion and sub-criterion as well as the corresponding BO and OW vectors should be firstly provided by decision-maker. For example, in the economic aspect, if the decision-maker believes that the best and worst criteria are net present value (C 4 ) and payback period (C 2 ), respectively. Furthermore, the relative importance of the best criterion C 4 over C 1 and C 1 over the worst criterion 1 can be written C 2 are “Weakly important” and “Fairly important”, then, a˜ 1B1 and a˜ 1W as (1,3,5) and (3,5,7). In a similar way, the best and worst criterion and sub-criterion in each aspect as well as the corresponding BO and OW vectors can be determined as shown in Table 4. Subsequently, the three sets of equations in Eq. (5) as described in Step 3 of Sect. 2.1 can be established when the feasibility degree of α takes a fixed value such as 0.8 as shown in Eq. (29), in which, the equations of economic aspect were used to represented set L in order to avoid a heavy description. At last, the model was established and solved by the GAMS software with the Antigone solver. The obtained results can be seen in Table 5 and Fig. 3.

174

X. Ren et al.

Table 4 The BO and OW vectors and corresponding values determined by the decision-maker Aspect

Criteria

B1 : economic

B2 : environmental

B3 : technical

B4 : social

B-aspect

Best-economic (BO)

JE (1,1,1)

WI (1,3,5)

FI (3,5,7)

VI (5,7,9)

Worst-social (OW)

VI (5,7,9)

FI (3,5,7)

WI (1,3,5)

JE (1,1,1)

Criteria

C1

C2

C3

C4

Best-C 4 (BO)

WI (1,3,5)

FI (3,5,7)

EI (1,1,3)

JE (1,1,1)

Worst-C 2 (OW)

FI (3,5,7)

JE (1,1,1)

FI (3,5,7)

FI (3,5,7)

Criteria

C5

C6

C7

C8

Best-C 7 (BO)

WI (1,3,5)

FI (3,5,7)

JE (1,1,1)

AI (7,9,9)

Worst-C 8 (OW)

FI (3,5,7)

WI (1,3,5)

AI (7,9,9)

JE (1,1,1)

Criteria

C9

C 10

C 11

Best-C 9 (BO)

JE (1,1,1)

FI (3,5,7)

VI (5,7,9)

Worst-C 11 (OW)

VI (5,7,9)

EI (1,1,3)

JE (1,1,1)

Criteria

C 12

C 13

C 14

Best-C 12 (BO)

JE (1,1,1)

FI (3,5,7)

WI (1,3,5)

Worst-C 13 (OW)

FI (3,5,7)

JE (1,1,1)

WI (1,3,5)

Economic

Environmental

Technological

Social

Table 5 The weights of the criteria Category

Weights

Criteria

Local weights

Global weights

Economic

0.5368

C1

0.1923

0.1032

C2

0.0641

0.0344

C3

0.3718

0.1996

Environmental

Technological

Social

0.2524

0.1377

0.0731

C4

0.3718

0.1996

C5

0.2460

0.0621

C6

0.1342

0.0339

C7

0.5562

0.1404

C8

0.0637

0.0161

C9

0.7238

0.0997

C 10

0.1599

0.0220

C 11

0.1163

0.0160

C 12

0.6037

0.0441

C 13

0.1282

0.0094

C 14

0.2681

0.0196

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

175

Fig. 3 The priority of criteria under different feasibility degrees α

1 L ξ 4 j=1 j 4

min Z = ξ0L +

⎧ ⎧ 3+5 1+3 ⎪ − 0.8) × w2 − w1 + ξ0L ≥ 0 + 0.8 × (1 ⎪ ⎪ 2 2 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ [0.2 × 6 + 0.8 × 4]w3 − w1 + ξ0L ≥ 0 ⎪ ⎪ ⎪ L ⎪ ⎪ ⎪ (A)⎪ [0.2 × 8 + 0.8 × 6]w4 − w1 + ξ0L ≥ 0 ⎪ ⎪ ⎪ ⎪ ⎪ [0.2 × 6 + 0.8 × 4]w4 − w2 + ξ0 ≥ 0 ⎪ ⎪ ⎩ ⎪ ⎪ ⎪ [0.2 × 4 + 0.8 × 2]w4 − w3 + ξ0L ≥ 0 ⎪ ⎧ ⎪ ⎪ ⎪ [0.2 × 4 + 0.8 × 2]w11 − w41 + ξ1L ≥ 0 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ [0.2 × 6 + 0.8 × 4]w21 − w41 + ξ1L ≥ 0 ⎪ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ [0.2 × 2 + 0.8 × 1]w31 − w41 + ξ1L ≥ 0 ⎪ ⎪ (L) ⎪ ⎪ ⎪ ⎪ [0.2 × 6 + 0.8 × 4]w21 − w11 + ξ1L ≥ 0 ⎪ ⎪ ⎪ ⎪ 1 1 L ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ ⎩ [0.2 × 6 + 0.8 × 4]w2 − w3 + ξ1 ≥ 0 ··· s.t. ⎧ 1 ⎪ 1 1 1 1 1 1 1 ⎪ μ ⎪ ⎪ 1 = w1 w1 ,μ2 = w1 w2 ,μ3 = w1 w3 ,μ4 = w1 w4 ⎪ ⎪ ⎪ ⎪ 2 2 2 2 2 2 2 ⎪ ⎪ μ1 = w2 w1 ,μ2 = w2 w2 ,μ3 = w2 w3 ,μ4 = w2 w42 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ μ31 = w3 w13 ,μ32 = w3 w23 ,μ33 = w3 w33 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ μ4 = w w4 ,μ4 = w w4 ,μ4 = w w4 ⎪ ⎪ ⎪ 4 1 4 2 4 3 ⎪ 1 2 3 ⎪ ⎪ ⎪ ⎨ 4 ⎪ ⎪ ⎪ w j =1,w1 ,w2 , w3 ,w4 ≥ 0 ⎪ (B) ⎪ ⎪ ⎪ j=1 ⎪ ⎪ ⎪ ⎪ 4 ⎪ ⎪ 4 3 3 ⎪ ⎪  1    ⎪ ⎪ ⎪ ⎪ wk =1, wk2 =1, wk3 =1, wk4 =1 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ k=1 k=1 k=1 k=1 ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ w1 ,w1 ,w1 ,w1 ,w2 ,w2 ,w2 ,w2 ≥ 0 ⎪ ⎪ ⎪ ⎩ 13 23 33 44 14 24 3 4 ⎩ w1 ,w2 ,w3 ,w1 ,w2 ,w3 ≥ 0

(29)

176

X. Ren et al.

As can be seen in Table 5, the economic aspect and social aspect are considered as the most and worst important by decision-maker, with total annual cost (C 3 ) and net present value (C 4 ) being the most important sub-criteria, followed by CO2 emission (C 7 ), investments cost (C 1 ) and technology advancement (C 9 ). Thus, it is reasonable to improve the sustainability of the energy system in the short term by reducing annual total cost, increasing the net present value of systems, or improving the performance of other criteria with larger weights such as C 7 , C 1 , C 9 . Subsequently, the sensitivity analysis was also conducted by changing feasibility degree of α from 0 to 1 with a step length of 0.1 to explore the effect of α on the weights of criteria and the values of consistency index as shown in Figs. 3 and 4, respectively. It can be concluded from Fig. 3 that the change of the feasibility degree α has a certain effect on the weights of the criteria and results in the reversal of the ranking among the criteria with similar weights, which proves the effectiveness of the proposed fuzzy BWM model. Meanwhile, it is obvious in Fig. 4 that the increase of feasibility degree of α leads to the decrease of the consistency index. The reason is that the feasible solution area in model 5 expands as α increases which result in the improvement of the optimal value of objective function. At last, it can be seen that all the consistency indexes are less than 0.1 when the value of the feasibility degree of α is greater than 0.7. Although the magnitude of the consistency index is not specified in the literature [35], it is considered to be reasonable that the decision makers’ judgment is consistent if the consistency index is less than 0.1, and it is appropriate to select the feasibility degree of α as 0.8 to calculate the weights of the criteria.

Fig. 4 The values of consistency index under different feasibility degrees α

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

177

3.2 Capturing the Complex Relationships Among Criteria In this section, the above fuzzy DEMATEL method was used to expose the causeeffect relationships among the criteria of energy systems. The decision-maker was firstly requested to determine the initial fuzzy direct-relation matrix expressed by the linguistic terms and the corresponding fuzzy numbers. For example, it could be deduced by decision-maker that the investments cost (C 1 ) should have a very high influence on total annual cost (C 3 ) and no influence on CO emission (C 6 ). Therefore, a˜ 1,3 and a˜ 1,6 in Eq. (6) can be determined by decision-maker, and expressed as “VH (0.75,0.75,1.00)” and “N (0,0,0.25)”, respectively. And the other variables of initial fuzzy direct-relation matrix can also be obtained in a similar way as shown in Tables 9 and 10 in Appendix 1. Then, the fuzzy total-relation matrix T˜ , can be calculated using the above Eqs. (10)–(14) which can be seen in Table 11. According to the Step 4, the crisp total-relation matrix and the significant relationships among criteria can be obtained as shown in Table 12 and Fig. 5, in which the threshold value was set to 0.5. Noteworthy, the threshold value can be set to the mean value or any arbitrarily reasonable value by decision-maker [45]. Subsequently, the results of the fuzzy DEMATEL, i.e., R˜ i + C˜ i , R˜ i − C˜ i , Ri + C i , Ri − C i , can be obtained by conducting the following steps in Sect. 2.2 using MATLAB software, which is shown in Table 6. And the digraph of cause-effect relationships among criteria of energy systems was presented by setting the threshold value to 0.6.in Fig. 6. From the results, it can be concluded that the influence of criteria C 5 , C 6 , C 7 , C 8 , C 10 , C 11 and C 13 are really small and can be removed from the cause-effect diagram, and the criteria of C 9 (Technology advancement), C 1 (Investment cost) and C 3 (Total annal cost) are identified as net cause, while the criteria of C 12 (Social acceptability), C 4 (Net present value), C 2 (Payback period) and C 14 (Contribution to energy security) are classified into effect region. Although the criteria in effect region have high influence and belong to the core criteria of improving system sustainability, they are not the original factors of the problems. Thus, in the long-term prospective, Fig. 5 Diagram of significant relations among criteria

178

X. Ren et al.

Table 6 Results of the fuzzy DEMATEL analysis and weights of criteria ~Ri + ~Ci

~Ri − ~Ci

Ri + C i

Category

Wj

C1

(0.3042, 0.5510, 1.6875)

(−0.3824, 0.3469, 1.0010)

0.7434

Ri − C i 0.3281

Cause

0.1032

C2

(0.2869, 0.5436, 1.7065)

(−0.9292, −0.2370, 0.4903)

0.7401

−0.2282

Effect

0.0344

C3

(0.2573, 0.5300, 1.7320)

(−0.6545, 0.0644, 0.8202)

0.7632

0.0736

Cause

0.1996

C4

(0.2791, 0.5523, 1.7288)

(−0.7849, −0.0612, 0.6648)

0.7781

−0.0606

Effect

0.1996

C5

(0.0860, 0.1669, 1.2650)

(−0.5790, 0.0138, 0.5999)

0.4212

0.0121

Cause

0.0621

C6

(0.0860, 0.1669, 1.2650)

(−0.5790, 0.0138, 0.5999)

0.4212

0.0121

Cause

0.0339

C7

(0.1075, 0.2424, 1.3432)

(−0.5575, 0.0893, 0.6781)

0.4839

0.0748

Cause

0.1404

C8

(0.0860, 0.1455, 1.2328)

(−0.5790,−0.0076, 0.5677)

0.4024

−0.0066

Effect

0.0161

C9

(0.6619, 1.2224, 2.6790)

(−0.3264, 0.6708, 1.6907)

1.4464

0.6764

Cause

0.0997

C 10

(0.0645, 0.1841, 1.2911)

(−0.6373,−0.0462, 0.5892)

0.4309

−0.0351

Effect

0.0220

C 11

(0.0215, 0.1462, 1.2335)

(−0.5838, 0.0442, 0.6282)

0.3869

0.0332

Cause

0.0160

C 12

(0.5966, 1.0187, 2.4896)

(−1.6891,−0.7207, 0.2039)

1.2809

−0.7317

Effect

0.0441

C 13

(0.0430, 0.2366, 1.3607)

(−0.7076,−0.0668, 0.6100)

0.4692

−0.0578

Effect

0.0094

C 14

(0.2103, 0.4751, 1.7100)

(−0.8268,−0.1036, 0.6728)

0.7176

−0.0903

Effect

0.0196

it is a better choice to improve the performance of criteria in cause region, i.e., criteria of C 9 , C 1 and C 3 , for improving the sustainability of energy systems. On the other side, improving the performance of criteria C 3 , C 4 , C 7 , C 1 and C 9 can improve the sustainability of energy systems in the short term as stated in the above section. Therefore, it is rational to improve the sustainability by improving the performance of criteria total annual cost (C 3 ), investments cost (C 1 ) and technology advancement (C 9 ) from both the long-term and short-term prospective.

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

179

Fig. 6 The cause-effect relationships among criteria

3.3 Ranking the Alternatives Finally, the fuzzy TOPSIS method was used to prioritized the proposed five CCHP systems according to the steps introduced in Sect. 2.3. In which, the performances of alternatives regarding each criterion which is expressed by linguistic variables and fuzzy numbers of five CCHP systems were firstly provided by the decision-maker, taken A1 as an example, the performances of A1 with respect to criteria C 1 and C 2 were considered as “Medium” and “bad” by decision-maker, the elements x˜11 and x˜12 can be expressed by fuzzy numbers, and written as (3,5,7) and (1,3,5). Therefore, others elements of initial decision-making matrix can be obtained in a similar way, as shown in Table 7. Then, the initial decision-making matrix was normalized according to Step 2 as shown in Table 13 in Appendix 2, while the weighted fuzzy normalized decision-making matrix was acquired by following Step 3 and Step 4 which was presented in Table 14. Subsequently, di+ , di− and the corresponding integrated closeness coefficients of the five CCHP systems can be calculated by performing Step 5–Step 7 using MATLAB software, as shown in Table 8 and Fig. 7. Obviously, A1 (SE) is regarded as the most sustainable CCHP system, followed by A3 , A4 , A5 and A2 . Subsequently, the sensitivity analysis of weights was conducted to explore the influence of weights on the final priorities of CCHP systems by setting three scenarios including 16 cases, i.e., Scenario 1-Case 0, the weights were obtained by the above fuzzy BWM; Scenario 2-Case 1, equal weights which means each weight of criterion is 1/14; Scenario 2-Case 2–15, oriented-weight, and the oriented weight which is assigned in order to criteria of C 1 , C 2 , …, C 14 is set to be 0.35 and other weights were taken as 0.05. The results were exhibited in Fig. 8. It is apparent that the change of weights has a great influence on the final sustainability priority of the

180

X. Ren et al.

Table 7 The performances of five CCHP systems considering each criterion evaluated by decisionmaker using linguistic variables and fuzzy numbers Criteria

CCHP systems SE (A1 )

MGT (A2 )

ICE (A3 )

SOFC (A4 )

Con (A5 )

C1

MM (3,5,7)

BD (1,3,5)

GD (5,7,9)

WT (0,1,1)

BT (9,10,10)

C2

MM (3,5,7)

MM (3,5,7)

BR (7,9,9)

WE (1,1,3)

BD (1,3,5)

C3

BD (1,3,5)

BD (1,3,5)

MM (3,5,7)

WE (1,1,3)

BR (7,9,9)

C4

BT (9,10,10)

GD (5,7,9)

BT (9,10,10)

WT (0,1,1)

GD (5,7,9)

C5

MM (3,5,7)

MM (3,5,7)

BD (1,3,5)

BT (9,10,10)

WT (0,1,1)

C6

MM (3,5,7)

BD (1,3,5)

BD (1,3,5)

BT (9,10,10)

WT (0,1,1)

C7

MM (3,5,7)

BD (1,3,5)

MM (3,5,7)

GD (5,7,9)

WE (1,1,3)

C8

MM (3,5,7)

MM (3,5,7)

BD (1,3,5)

GD (5,7,9)

GD (5,7,9)

C9

GD (5,7,9)

MM (3,5,7)

BD (1,3,5)

BT (9,10,10)

WT (0,1,1)

C 10

GD (5,7,9)

GD (5,7,9)

WE (1,1,3)

BT (9,10,10)

WT (0,1,1)

C 11

WT (0,1,1)

WE (1,1,3)

GD (5,7,9)

BR (7,9,9)

BT (9,10,10)

C 12

GD (5,7,9)

MM (3,5,7)

WE (1,1,3)

BT (9,10,10)

WE (1,1,3)

C 13

GD (5,7,9)

MM (3,5,7)

MM (3,5,7)

BR (7,9,9)

WE (1,1,3)

C 14

GD (5,7,9)

GD (5,7,9)

MM (3,5,7)

BR (7,9,9)

WT (0,1,1)

Table 8 The results of di+ , di− and the corresponding integrated closeness coefficients of the five CCHP systems SE (A1 )

MGT (A2 )

ICE (A3 )

SOFC (A4 )

Con (A5 )

di+

0.3276

0.4720

0.3426

0.4551

0.4151

di−

0.4987

0.3658

0.4812

0.3591

0.3988

CCi

0.6036

0.4366

0.5841

0.4410

0.4900

five CCHP systems, different decision-makers with various viewpoints will obtain different weights resulting in the variation of ranking results, and it could also explain why the priorities of CCHP systems obtained in this chapter are different from the work of He et al. [26].

4 Conclusion The sustainability assessment framework of energy systems under uncertainty was established by integrating the improved fuzzy BWM, fuzzy DEMATEL method, and fuzzy TOPSIS method with two purposes, i.e., providing an accurate and comprehensive information of improving the system sustainability from both short-term

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

Fig. 7 The integrated closeness coefficients of the five CCHP systems

Fig. 8 The priorities of five CCHP systems in different cases

181

182

X. Ren et al.

and long-term perspectives, and prioritizing the energy system according to sustainability. The validity of the proposed framework was identified through a case study containing five CCHP systems. By analyzing the obtained results, the following conclusions can be drawn: (1) The sustainability of CCHP systems can be improved in the short term by reducing annual total cost and increasing the net present value of systems, while in the long-term perspective, vigorously developing advanced technology might be a better choice. (2) In general, improving the performance of criteria total annual cost, investments cost, and technology advancement is beneficial to the sustainability improvement from both the long-term and short-term prospective. (3) The priorities of the CCHP systems are sensitive to the weights of the criteria, so it is crucial to elicit more accurately the preference of decision-makers on the relative importance of the criteria or to develop the group decision-making method, which is also the work to be done in the future.

Appendix 1 See Tables 9, 10, 11 and 12.

Appendix 2 See Tables 13 and 14.

Table 9 The initial fuzzy direct-relation matrix expressed by the linguistic terms Criteria

C1

C2

C3

C4

C5

C6

C7

C8

C9

C 10

C 11

C 12

C 13

C 14

C1

N

VH

VH

VH

N

N

N

N

VL

N

N

H

VL

VL

C2

N

N

N

N

N

N

N

N

VL

N

N

H

VL

VL

C3

N

H

N

H

N

N

N

N

VL

N

N

H

VL

VL

C4

N

VH

N

N

N

N

N

N

VL

N

N

H

VL

VL

C5

N

N

N

N

N

N

N

N

N

VL

N

H

N

N

C6

N

N

N

N

N

N

N

N

N

VL

N

H

N

N

C7

N

VL

VL

VL

N

N

N

N

N

N

N

VH

N

N

C8

N

N

N

N

N

N

N

N

N

N

N

H

N

N

C9

VH

VH

VH

VH

H

H

H

H

N

M

M

H

M

H

C 10

N

N

N

N

N

N

N

N

N

N

N

H

N

N

C 11

N

N

VL

VL

N

N

N

N

N

VL

N

VL

N

N

C 12

N

N

N

N

N

N

N

N

M

N

N

N

N

H

C 13

N

N

N

N

N

N

N

N

VL

N

N

M

N

N

C 14

N

N

N

N

N

N

N

N

H

N

N

H

N

N

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C1

C2

C3

C4

C5

C6

C7

C8

C1

0, 0, 0.25

0, 0.25, 0.50

0, 0, 0.25

0, 0, 0.25

0.75, 1.00, 1.00

0.50, 0.75, 1.00

0, 0, 0.25

0.75, 1.00, 1.00

C2

0, 0, 0.25

0, 0.25, 0.50

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.75, 1.00, 1.00

C3

0, 0, 0.25

0, 0.25, 0.50

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.50, 0.75, 1.00

0, 0, 0.25

0.75, 1.00, 1.00

C4

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C5

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C6

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C7

Table 10 The initial fuzzy direct-relation matrix expressed by the fuzzy numbers

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C8

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0.25, 0.50

0, 0.25, 0.50

0, 0.25, 0.50

0, 0.25, 0.50

C9

0, 0, 0.25

0, 0, 0.25

0, 0.25, 0.50

0, 0.25, 0.50

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C 10

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C 11

0.50, 0.75, 1.00

0.75, 1.00, 1.00

0.50, 0.75, 1.00

0.50, 0.75, 1.00

0.50, 0.75, 1.00

0.50, 0.75, 1.00

0.50, 0.75, 1.00

0.50, 0.75, 1.00

C 12

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0.25, 0.50

0, 0.25, 0.50

0, 0.25, 0.50

0, 0.25, 0.50

C 13

(continued)

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0.25, 0.50

0, 0.25, 0.50

0, 0.25, 0.50

0, 0.25, 0.50

C 14

Multi-criteria Decision Analysis Methods for Sustainability Assessment … 183

0.75, 1.00, 1.00

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C9

C 10

C 11

C 12

C 13

C 14

C1

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.75, 1.00, 1.00

C2

Table 10 (continued)

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0.25, 0.50

0, 0, 0.25

0.75, 1.00, 1.00

C3

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0.25, 0.50

0, 0, 0.25

0.75, 1.00, 1.00

C4

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.50, 0.75, 1.00

C5

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.50, 0.75, 1.00

C6

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.50, 0.75, 1.00

C7

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.50, 0.75, 1.00

C8

0.50, 0.75, 1.00

0, 0.25, 0.50

0.25, 0.50, 0.75

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

C9

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0.25, 0.50

0, 0, 0.25

0.25, 0.50, 0.75

C 10

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.25, 0.50, 0.75

C 11

0.50, 0.75, 1.00

0.25, 0.50, 0.75

0, 0, 0.25

0, 0.25, 0.50

0.50, 0.75, 1.00

0.50, 0.75, 1.00

C 12

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0, 0, 0.25

0.25, 0.50, 0.75

C 13

0, 0, 0.25

0, 0, 0.25

0.50, 0.75, 1.00

0, 0, 0.25

0, 0, 0.25

0.50, 0.75, 1.00

C 14

184 X. Ren et al.

C1

0.0001

0.0001

0.0001

0.0001

0.0001

0.0001

0.0001

0.0001

0.0602

0.0001

0.0000

0.0013

0.0000

0.0025

C1

0.0025

0.0020

0.0022

TL

C1

C2

C3

C4

C5

C6

C7

C8

C9

C 10

C 11

C 12

C 13

C 14

TM

C1

C2

C3

0.0676

0.0025

0.0947

C2

0.0029

0.0000

0.0015

0.0000

0.0001

0.0703

0.0001

0.0001

0.0001

0.0001

0.0601

0.0425

0.0001

0.0662

C2

0.0025

0.0022

0.0827

C3

0.0026

0.0000

0.0014

0.0000

0.0001

0.0638

0.0001

0.0001

0.0001

0.0001

0.0001

0.0001

0.0001

0.0601

C3

0.0626

0.0023

0.0877

C4

0.0027

0.0000

0.0014

0.0000

0.0001

0.0663

0.0001

0.0001

0.0001

0.0001

0.0001

0.0401

0.0001

0.0625

C4

Table 11 The fuzzy total-relation matrix C5

0.0017

0.0015

0.0019

C5

0.0016

0.0000

0.0009

0.0000

0.0000

0.0401

0.0000

0.0001

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

C6

0.0017

0.0015

0.0019

C6

0.0016

0.0000

0.0009

0.0000

0.0000

0.0401

0.0000

0.0001

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

C7

0.0017

0.0015

0.0019

C7

0.0016

0.0000

0.0009

0.0000

0.0000

0.0401

0.0000

0.0001

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

C8

0.0017

0.0015

0.0019

C8

0.0016

0.0000

0.0009

0.0000

0.0000

0.0401

0.0000

0.0001

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

C9

0.0278

0.0247

0.0311

C9

0.0410

0.0004

0.0217

0.0000

0.0009

0.0029

0.0009

0.0013

0.0009

0.0009

0.0009

0.0009

0.0009

0.0010

C 10

0.0012

0.0011

0.0013

C 10

0.0008

0.0000

0.0004

0.0000

0.0000

0.0201

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

C 11

0.0011

0.0010

0.0012

C 11

0.0008

0.0000

0.0004

0.0000

0.0000

0.0201

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

C 12

0.0730

0.0649

0.0815

C 12

0.0425

0.0201

0.0029

0.0000

0.0401

0.0607

0.0401

0.0602

0.0401

0.0401

0.0425

0.0434

0.0401

0.0477

C13

0.0238

0.0212

0.0266

C13

0.0008

0.0000

0.0004

0.0000

0.0000

0.0201

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

0.0000

C14

(continued)

0.0287

0.0256

0.0321

C14

0.0033

0.0008

0.0410

0.0000

0.0016

0.0425

0.0016

0.0025

0.0016

0.0016

0.0017

0.0018

0.0016

0.0019

Multi-criteria Decision Analysis Methods for Sustainability Assessment … 185

C1

0.0021

0.0002

0.0002

0.0004

0.0002

0.0814

0.0002

0.0002

0.0036

0.0018

0.0051

C1

0.0462

0.0388

0.0436

0.0411

0.0355

0.0355

0.0373

0.0348

TL

C4

C5

C6

C7

C8

C9

C 10

C 11

C 12

C 13

C 14

TM

C1

C2

C3

C4

C5

C6

C7

C8

0.0423

0.0678

0.0432

0.0432

0.1100

0.1166

0.0472

0.1236

C2

0.0064

0.0022

0.0045

0.0031

0.0003

0.1021

0.0003

0.0234

0.0003

0.0003

0.0827

C2

Table 11 (continued)

C3

0.0383

0.0610

0.0391

0.0391

0.0452

0.0479

0.0427

0.1108

C3

0.0056

0.0020

0.0039

0.0202

0.0002

0.0899

0.0002

0.0205

0.0002

0.0002

0.0024

C4

0.0406

0.0647

0.0414

0.0414

0.0479

0.1108

0.0452

0.1175

C4

0.0060

0.0021

0.0042

0.0214

0.0003

0.0953

0.0003

0.0217

0.0003

0.0003

0.0025

C5

0.0348

0.0373

0.0355

0.0355

0.0411

0.0436

0.0388

0.0462

C5

0.0038

0.0013

0.0027

0.0001

0.0002

0.0610

0.0002

0.0003

0.0002

0.0002

0.0016

C6

0.0348

0.0373

0.0355

0.0355

0.0411

0.0436

0.0388

0.0462

C6

0.0038

0.0013

0.0027

0.0001

0.0002

0.0610

0.0002

0.0003

0.0002

0.0002

0.0016

C7

0.0348

0.0373

0.0355

0.0355

0.0411

0.0436

0.0388

0.0462

C7

0.0038

0.0013

0.0027

0.0001

0.0002

0.0610

0.0002

0.0003

0.0002

0.0002

0.0016

C8

0.0348

0.0373

0.0355

0.0355

0.0411

0.0436

0.0388

0.0462

C8

0.0038

0.0013

0.0027

0.0001

0.0002

0.0610

0.0002

0.0003

0.0002

0.0002

0.0016

C9

0.0436

0.0479

0.0444

0.0444

0.0728

0.0772

0.0687

0.0818

C9

0.0637

0.0221

0.0445

0.0020

0.0027

0.0172

0.0027

0.0051

0.0027

0.0027

0.0267

C 10

0.0360

0.0386

0.0567

0.0567

0.0421

0.0446

0.0397

0.0473

C 10

0.0028

0.0010

0.0019

0.0201

0.0001

0.0439

0.0001

0.0002

0.0201

0.0201

0.0012

C 11

0.0340

0.0363

0.0346

0.0346

0.0397

0.0421

0.0374

0.0446

C 11

0.0025

0.0009

0.0018

0.0001

0.0001

0.0407

0.0001

0.0002

0.0001

0.0001

0.0011

C 12

0.1196

0.1278

0.1220

0.1220

0.1360

0.1442

0.1283

0.1528

C 12

0.0670

0.0425

0.0084

0.0242

0.0605

0.1089

0.0605

0.0848

0.0617

0.0617

0.0701

C13

0.0371

0.0410

0.0378

0.0378

0.0646

0.0684

0.0609

0.0725

C13

0.0030

0.0010

0.0021

0.0010

0.0001

0.0481

0.0001

0.0015

0.0001

0.0001

0.0229

C14

(continued)

0.0451

0.0496

0.0460

0.0460

0.0742

0.0786

0.0700

0.0834

C14

0.0083

0.0040

0.0635

0.0025

0.0038

0.0749

0.0038

0.0067

0.0039

0.0039

0.0276

186 X. Ren et al.

C1

0.1204

0.0348

0.0356

0.0398

0.0364

0.0421

TL

C9

C 10

C 11

C 12

C 13

C 14

0.0511

0.0443

0.0484

0.0458

0.0423

0.1464

C2

Table 11 (continued)

C3

0.0462

0.0401

0.0437

0.0592

0.0383

0.1320

C4

0.0490

0.0425

0.0463

0.0627

0.0406

0.1399

C5

0.0421

0.0364

0.0398

0.0356

0.0348

0.1204

C6

0.0421

0.0364

0.0398

0.0356

0.0348

0.1204

C7

0.0421

0.0364

0.0398

0.0356

0.0348

0.1204

C8

0.0421

0.0364

0.0398

0.0356

0.0348

0.1204

C9

0.1087

0.0635

0.0882

0.0439

0.0436

0.0850

C 10

0.0424

0.0373

0.0404

0.0569

0.0360

0.1055

C 11

0.0399

0.0352

0.0380

0.0348

0.0340

0.0987

C 12

0.1318

0.1022

0.0713

0.0847

0.1196

0.2029

C13

0.0436

0.0384

0.0416

0.0388

0.0371

0.1095

C14

0.0537

0.0458

0.1076

0.0448

0.0451

0.1434

Multi-criteria Decision Analysis Methods for Sustainability Assessment … 187

0.0128

0.0107

0.0120

0.0114

0.0090

0.0090

0.0096

0.0088

0.0858

0.0088

0.0090

0.0121

0.0100

0.0137

C1

C2

C3

C4

C5

C6

C7

C8

C9

C 10

C 11

C 12

C 13

C 14

C1

0.0167

0.0122

0.0147

0.0130

0.0107

0.1052

0.0107

0.0287

0.0110

0.0110

0.0839

0.0736

0.0131

0.0948

C2

0.0150

0.0110

0.0132

0.0249

0.0097

0.0939

0.0097

0.0255

0.0099

0.0099

0.0125

0.0132

0.0118

0.0841

C3

Table 12 The crisp total-relation matrix

C4

0.0159

0.0117

0.0140

0.0264

0.0103

0.0992

0.0103

0.0270

0.0105

0.0105

0.0132

0.0690

0.0125

0.0888

C5

0.0128

0.0098

0.0115

0.0090

0.0088

0.0706

0.0088

0.0095

0.0090

0.0090

0.0111

0.0117

0.0105

0.0125

C6

0.0128

0.0098

0.0115

0.0090

0.0088

0.0706

0.0088

0.0095

0.0090

0.0090

0.0111

0.0117

0.0105

0.0125

C7

0.0128

0.0098

0.0115

0.0090

0.0088

0.0706

0.0088

0.0095

0.0090

0.0090

0.0111

0.0117

0.0105

0.0125

C8

0.0128

0.0098

0.0115

0.0090

0.0088

0.0706

0.0088

0.0095

0.0090

0.0090

0.0111

0.0117

0.0105

0.0125

C9

0.0693

0.0270

0.0497

0.0120

0.0124

0.0306

0.0124

0.0149

0.0127

0.0127

0.0318

0.0334

0.0298

0.0363

C 10

0.0122

0.0098

0.0112

0.0243

0.0091

0.0534

0.0091

0.0098

0.0243

0.0243

0.0111

0.0118

0.0105

0.0125

C 11

0.0114

0.0092

0.0105

0.0087

0.0085

0.0500

0.0085

0.0092

0.0087

0.0087

0.0105

0.0111

0.0099

0.0118

C 12

0.0771

0.0518

0.0227

0.0333

0.0702

0.1203

0.0702

0.0894

0.0714

0.0714

0.0797

0.0834

0.0745

0.0909

C13

0.0126

0.0101

0.0116

0.0102

0.0093

0.0564

0.0093

0.0110

0.0095

0.0095

0.0276

0.0290

0.0258

0.0314

C14

0.0184

0.0137

0.0689

0.0124

0.0136

0.0839

0.0136

0.0164

0.0139

0.0139

0.0328

0.0345

0.0307

0.0374

188 X. Ren et al.

0.3000

0.3333

0.1111

0.9000

0.3000

0.3000

0.3333

0.3333

0.5000

0.5000

0.0000

0.5000

0.5556

0.5556

C2

C3

C4

C5

C6

C7

C8

C9

C 10

C 11

C 12

C 13

C 14

0.7778

0.7778

0.7000

0.1000

0.7000

0.7000

0.5556

0.5556

0.5000

0.5000

1.0000

0.3333

0.5556

0.5000

1.0000

1.0000

0.9000

0.1000

0.9000

0.9000

0.7778

0.7778

0.7000

0.7000

1.0000

0.5556

0.7778

0.7000

0.5556

0.3333

0.3000

0.1000

0.5000

0.3000

0.3333

0.1111

0.1000

0.3000

0.5000

0.1111

0.3333

0.1000

xL

C1

MGT

xU

xL

xM

SE

Table 13 The initial decision-making matrix

0.7778

0.5556

0.5000

0.1000

0.7000

0.5000

0.5556

0.3333

0.3000

0.5000

0.7000

0.3333

0.5556

0.3000

xM

1.0000

0.7778

0.7000

0.3000

0.9000

0.7000

0.7778

0.5556

0.5000

0.7000

0.9000

0.5556

0.7778

0.5000

xU

0.3333

0.3333

0.1000

0.5000

0.1000

0.1000

0.1111

0.3333

0.1000

0.1000

0.9000

0.3333

0.7778

0.5000

xL

ICE

0.5556

0.5556

0.1000

0.7000

0.1000

0.3000

0.3333

0.5556

0.3000

0.3000

1.0000

0.5556

1.0000

0.7000

xM

0.7778

0.7778

0.3000

0.9000

0.3000

0.5000

0.5556

0.7778

0.5000

0.5000

1.0000

0.7778

1.0000

0.9000

xU

0.7778

0.7778

0.9000

0.7000

0.9000

0.9000

0.5556

0.5556

0.9000

0.9000

0.0000

0.1111

0.1111

0.0000

xL

SOFC

1.0000

1.0000

1.0000

0.9000

1.0000

1.0000

0.7778

0.7778

1.0000

1.0000

0.1000

0.1111

0.1111

0.1000

xM

1.0000

1.0000

1.0000

0.9000

1.0000

1.0000

1.0000

1.0000

1.0000

1.0000

0.1000

0.3333

0.3333

0.1000

xU

0.0000

0.1111

0.1000

0.9000

0.0000

0.0000

0.5556

0.1111

0.0000

0.0000

0.5000

0.7778

0.1111

0.9000

xL

Con

0.1111

0.1111

0.1000

1.0000

0.1000

0.1000

0.7778

0.1111

0.1000

0.1000

0.7000

1.0000

0.3333

1.0000

xM

0.1111

0.3333

0.3000

1.0000

0.1000

0.1000

1.0000

0.3333

0.1000

0.1000

0.9000

1.0000

0.5556

1.0000

xU

Multi-criteria Decision Analysis Methods for Sustainability Assessment … 189

0.0310

0.0115

0.0222

0.1796

0.0186

0.0102

0.0468

0.0054

0.0499

0.0110

0.0000

0.0221

0.0052

0.0109

C2

C3

C4

C5

C6

C7

C8

C9

C 10

C 11

C 12

C 13

C 14

0.0152

0.0073

0.0309

0.0016

0.0154

0.0698

0.0089

0.0780

0.0170

0.0311

0.1996

0.0665

0.0191

0.0516

0.0196

0.0094

0.0397

0.0016

0.0198

0.0897

0.0125

0.1092

0.0237

0.0435

0.1996

0.1109

0.0268

0.0722

0.0109

0.0031

0.0132

0.0016

0.0110

0.0299

0.0054

0.0156

0.0034

0.0186

0.0998

0.0222

0.0115

0.0103

xL

C1

MGT

xU

xL

xM

SE

0.0152

0.0052

0.0221

0.0016

0.0154

0.0499

0.0089

0.0468

0.0102

0.0311

0.1397

0.0665

0.0191

0.0310

xM

0.0196

0.0073

0.0309

0.0048

0.0198

0.0698

0.0125

0.0780

0.0170

0.0435

0.1796

0.1109

0.0268

0.0516

xU

Table 14 The weighted normalized decision-making matrix

0.0065

0.0031

0.0044

0.0080

0.0022

0.0100

0.0018

0.0468

0.0034

0.0062

0.1796

0.0665

0.0268

0.0516

xL

ICE

0.0109

0.0052

0.0044

0.0112

0.0022

0.0299

0.0054

0.0780

0.0102

0.0186

0.1996

0.1109

0.0344

0.0722

xM

0.0152

0.0073

0.0132

0.0144

0.0066

0.0499

0.0089

0.1092

0.0170

0.0311

0.1996

0.1552

0.0344

0.0929

xU

0.0152

0.0073

0.0397

0.0112

0.0198

0.0897

0.0089

0.0780

0.0305

0.0559

0.0000

0.0222

0.0038

0.0000

xL

SOFC

0.0196

0.0094

0.0441

0.0144

0.0220

0.0997

0.0125

0.1092

0.0339

0.0621

0.0200

0.0222

0.0038

0.0103

xM

0.0196

0.0094

0.0441

0.0144

0.0220

0.0997

0.0161

0.1404

0.0339

0.0621

0.0200

0.0665

0.0115

0.0103

xU

0.0000

0.0010

0.0044

0.0144

0.0000

0.0000

0.0089

0.0156

0.0000

0.0000

0.0998

0.1552

0.0038

0.0929

xL

Con

0.0022

0.0010

0.0044

0.0160

0.0022

0.0100

0.0125

0.0156

0.0034

0.0062

0.1397

0.1996

0.0115

0.1032

xM

0.0022

0.0031

0.0132

0.0160

0.0022

0.0100

0.0161

0.0468

0.0034

0.0062

0.1796

0.1996

0.0191

0.1032

xU

190 X. Ren et al.

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

191

References 1. Yuan J, Li Y, Luo X, Zhang Z, Ruan Y, Zhou Q (2020) A new hybrid multi-criteria decisionmaking approach for developing integrated energy systems in industrial parks. J Cleaner Product 270 2. Ren J (2018) Multi-criteria decision making for the prioritization of energy systems under uncertainties after life cycle sustainability assessment. Sustain Product Consump 16:45–57 3. Liu Y, Du J-l (2020) A multi criteria decision support framework for renewable energy storage technology selection. J Cleaner Product 277 4. Ghenai C, Albawab M, Bettayeb M (2020) Sustainability indicators for renewable energy systems using multi-criteria decision-making model and extended SWARA/ARAS hybrid method. Renew Energy 146:580–597 5. Xu Y, Yan C, Liu H, Wang J, Yang Z, Jiang Y (2020) Smart energy systems: a critical review on design and operation optimization. Sustain Cities Soc 62 6. Robertson Munro F, Cairney P (2020) A systematic review of energy systems: the role of policymaking in sustainable transitions. Renew Sustain Energy Rev 119 7. Shao M, Han Z, Sun J, Xiao C, Zhang S, Zhao Y (2020) A review of multi-criteria decision making applications for renewable energy site selection. Renew Energy 157:377–403 8. Ilbahar E, Cebi S, Kahraman C (2019) A state-of-the-art review on multi-attribute renewable energy decision making. Energy Strategy Rev 25:18–33 9. Ren J, Xu D, Cao H, Wei Sa, Dong L, Goodsite ME (2016) Sustainability decision support framework for industrial system prioritization. AIChE J 62:108–130 10. Thies C, Kieckhäfer K, Spengler TS, Sodhi MS (2019) Operations research for sustainability assessment of products: a review. Eur J Oper Res 274:1–21 11. W˛atróbski J, Jankowski J, Ziemba P, Karczmarczyk A, Zioło M (2019) Generalised framework for multi-criteria method selection. Omega 86:107–124 12. Cinelli M, Kadzi´nski M, Gonzalez M, Słowi´nski R (2020) How to support the application of multiple criteria decision analysis? Let us start with a comprehensive taxonomy. Omega 96 13. Witt T, Dumeier M, Geldermann J (2020) Combining scenario planning, energy system analysis, and multi-criteria analysis to develop and evaluate energy scenarios. J Cleaner Product 242 14. Ridha E, Nolting L, Praktiknjo A (2020) Complexity profiles: a large-scale review of energy system models in terms of complexity. Energy Strategy Rev 30 15. Ren J, Ren X, Shen W, Man Y, Lin R, Liu Y et al (2020) Industrial system prioritization using the sustainability-interval-index conceptual framework with life-cycle considerations. AIChE J 66 16. Nsafon BEK, Butu HM, Owolabi AB, Roh JW, Suh D, Huh J-S (2020) Integrating multicriteria analysis with PDCA cycle for sustainable energy planning in Africa: application to hybrid mini-grid system in Cameroon. Sustain Energy Technol Assessments 37 17. Maxim A (2014) Sustainability assessment of electricity generation technologies using weighted multi-criteria decision analysis. Energy Policy 65:284–297 18. Ren J, Manzardo A, Toniolo S, Scipioni A (2013) Sustainability of hydrogen supply chain. Part II: prioritizing and classifying the sustainability of hydrogen supply chains based on the combination of extension theory and AHP. Int J Hydrogen Energy 38:13845–13855 19. Pilpola S, Lund PD (2020) Analyzing the effects of uncertainties on the modelling of low-carbon energy system pathways. Energy 201 20. Ebrahimi M, Keshavarz A (2012) Prime mover selection for a residential micro-CCHP by using two multi-criteria decision-making methods. Energy Buildi 55:322–331 21. Jing Y-Y, Bai H, Wang J-J (2012) A fuzzy multi-criteria decision-making model for CCHP systems driven by different energy sources. Energy Policy 42:286–296 22. Ren J, Liang H (2017) Measuring the sustainability of marine fuels: a fuzzy group multi-criteria decision making approach. Transp Res Part D Transp Environ 54:12–29

192

X. Ren et al.

23. Ren J, Lützen M (2017) Selection of sustainable alternative energy source for shipping: multicriteria decision making under incomplete information. Renew Sustain Energy Rev 74:1003– 1019 24. Ren J, Dong L (2018) Evaluation of electricity supply sustainability and security: multi-criteria decision analysis approach. J Clean Prod 172:438–453 25. Streimikiene D, Baležentis T (2013) Multi-criteria assessment of small scale CHP technologies in buildings. Renew Sustain Energy Rev 26:183–189 26. He C, Zhang Q, Ren J, Li Z (2017) Combined cooling heating and power systems: sustainability assessment under uncertainties. Energy 139:755–766 27. Ren J (2018) Life cycle aggregated sustainability index for the prioritization of industrial systems under data uncertainties. Comput Chem Eng 113:253–263 28. Ren J, Toniolo S (2018) Life cycle sustainability decision-support framework for ranking of hydrogen production pathways under uncertainties: an interval multi-criteria decision making approach. J Clean Prod 175:222–236 29. Yang K, Ding Y, Zhu N, Yang F, Wang Q (2018) Multi-criteria integrated evaluation of distributed energy system for community energy planning based on improved grey incidence approach: a case study in Tianjin. Appl Energy 229:352–363 30. Ren X, Li W, Ding S, Dong L (2020) Sustainability assessment and decision making of hydrogen production technologies: a novel two-stage multi-criteria decision making method. Int J Hydrogen Energy. https://doi.org/10.1016/j.ijhydene.2019.12.134 31. Chou Y-C, Sun C-C, Yen H-Y (2012) Evaluating the criteria for human resource for science and technology (HRST) based on an integrated fuzzy AHP and fuzzy DEMATEL approach. Appl Soft Comput 12:64–71 32. Ren J, Manzardo A, Toniolo S, Scipioni A (2013) Sustainability of hydrogen supply chain. Part I: identification of critical criteria and cause–effect analysis for enhancing the sustainability using DEMATEL. Int J Hydrogen Energy 38:14159–14171 33. Rezaei J (2015) Best-worst multi-criteria decision-making method. Omega 53:49–57 34. Rezaei J (2016) Best-worst multi-criteria decision-making method: Some properties and a linear model. Omega 64:126–130 35. Mi X, Tang M, Liao H, Shen W, Lev B (2019) The state-of-the-art survey on integrations and applications of the best worst method in decision making: why, what, what for and what’s next? Omega 87:205–225 36. Karimi H, Sadeghi-Dastaki M, Javan M (2020) A fully fuzzy best–worst multi attribute decision making method with triangular fuzzy number: a case study of maintenance assessment in the hospitals. Appl Soft Comput 86 37. Amiri M, Hashemi-Tabatabaei M, Ghahremanloo M, Keshavarz-Ghorabaee M, Zavadskas EK, Antucheviciene J (2020) A new fuzzy approach based on BWM and fuzzy preference programming for hospital performance evaluation: a case study. Appl Soft Comput 92 38. Hafezalkotob A, Hafezalkotob A (2017) A novel approach for combination of individual and group decisions based on fuzzy best-worst method. Appl Soft Comput 59:316–325 39. Guo S, Zhao H (2017) Fuzzy best-worst multi-criteria decision-making method and its applications. Knowl-Based Syst 121:23–31 40. Jiménez M, Arenas M, Bilbao A, Rodrı´guez MV (2007) Linear programming with fuzzy parameters: An interactive method resolution. Euro J Oper Res 177:1599–609 41. Tabatabaei MH, Amiri M, Ghahremanloo M, Keshavarz-Ghorabaee M, Zavadskas EK, Antucheviciene J (2019) Hierarchical decision-making using a new mathematical model based on the best-worst method. Int J Comput Commun Control 14(6):669–684 42. Ren J, Gao S, Tan S, Dong L, Scipioni A, Mazzi A (2015) Role prioritization of hydrogen production technologies for promoting hydrogen economy in the current state of China. Renew Sustain Energy Rev 41:1217–1229 43. Ren J (2018) New energy vehicle in China for sustainable development: analysis of success factors and strategic implications. Transp Res Part D Transp Environ 59:268–288 44. Zhang X, Ming X, Yin D (2020) Application of industrial big data for smart manufacturing in product service system based on system engineering using fuzzy DEMATEL. J Cleaner Product 265

Multi-criteria Decision Analysis Methods for Sustainability Assessment …

193

45. Xu C, Wu Y, Dai S (2020) What are the critical barriers to the development of hydrogen refueling stations in China? A modified fuzzy DEMATEL approach. Energy Policy 142 46. Addae BA, Zhang L, Zhou P, Wang F (2019) Analyzing barriers of smart energy city in Accra with two-step fuzzy DEMATEL. Cities 89:218–227 47. Ren J, Sovacool BK (2014) Quantifying, measuring, and strategizing energy security: determining the most meaningful dimensions and metrics. Energy 76:838–849 48. Mahmoudi S, Jalali A, Ahmadi M, Abasi P, Salari N Identifying critical success factors in heart failure self-care using fuzzy DEMATEL method. Appl Soft Comput 84 49. Ahmadi O, Mortazavi SB, Mahabadi HA, Hosseinpouri M (2020) Development of a dynamic quantitative risk assessment methodology using fuzzy DEMATEL-BN and leading indicators. Process Saf Environ Prot 142:15–44 50. Qi R, Li S, Qu L, Sun L, Gong C (2020) Critical factors to green mining construction in China: a two-step fuzzy DEMATEL analysis of state-owned coal mining enterprises. J Cleaner Product 273 51. Farooque M, Jain V, Zhang A, Li Z (2020) Fuzzy DEMATEL analysis of barriers to blockchainbased life cycle assessment in China. Comput Indus Eng 147 52. Mohammadfam I, Mirzaei Aliabadi M, Soltanian AR, Tabibzadeh M, Mahdinia M (2019) Investigating interactions among vital variables affecting situation awareness based on Fuzzy DEMATEL method. Int J Indus Ergonomics 74 53. Chen C (2000) Extensions of the TOPSIS for group decision-making under fuzzy environment. Fuzzy Sets Syst 114:1–9

An Integrated Hesitant Fuzzy Decision Model for Sustainable Wind Farm Site Selection: The Case Study in the Central Anatolian Region of Turkey Beyzanur Cayir Ervural

Abstract Assessment of the ideal location for power plants is an important issue followed by urban planners and energy investors in order to identify appropriate conditions depending on sustainability principles and different dynamics. Wind energy is one of the fastest-growing power sources in the world due to many advantages. The study aims to select an optimum location for new wind farms installation in the Central Anatolia Region of Turkey, which has unutilized great wind power potential, considering the hesitations/concerns inherent such decisions. For this purpose, we conduct a methodology based on an integrated analytic hierarchy process (AHP) and technique for order preference by similarity to ideal solution (TOPSIS) under a hesitant fuzzy environment for the assessment of the multi-criteria renewable power plant site selection problem to make optimal/the best use of available resources. In this study, we investigate five alternative locations, four main criteria, and thirteen sub-criteria considering the judgments of three experts from the energy market. Firstly, AHP methodology was applied to obtain criteria weights, and then, hesitant fuzzy TOPSIS technique was implemented to identify ideal areas for a wind farm. The results showed that Aksaray is the most important location among alternatives, while Yozgat is the least favored priority. A sensitivity analysis was conducted to understand whether the developed model behavior is consistent and applicable. We discuss the obtained results for the progress of long-term wind energy strategies that will help development of the region in a sustainable and green energy framework. Keywords Site selection · Wind farm · Hesitant fuzzy TOPSIS · AHP · Decision-making

B. Cayir Ervural (B) Konya Food and Agriculture University, Konya, Turkey e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 J. Ren (ed.), Energy Systems Evaluation (Volume 2), Green Energy and Technology, https://doi.org/10.1007/978-3-030-67376-5_8

195

196

B. Cayir Ervural

1 Introduction The increasing needs of humanity force them to find new alternative solutions to meet the requirements that arise daily in different ways. In parallel with the population expansion, industrial growth and technological developments primarily cause the depletion of natural resources as well as energy resources. One of the most striking reality is that if the resources are utilized in this way, all energy sources will be exhausted in the near future [1]. In light of this fact, all resources should be properly evaluated, their effectiveness should be analyzed and assessed in-depth, then alternative resources should be investigated, and strategic decisions should be developed instantly [2]. It is very advantageous to install wind farms in places with high wind energy potential because wind farms are clean, economical, and long-term investment tools. Overall, the payback period of wind farms is usually five years and attracts most investors to generate electricity from wind power. However, the most critical issue is the identification of the optimal location of wind farms. Power plant site selection is a substantial research field since the determination of appropriate regions is crucial for achieving a balance of successful energy production/consumption, meeting the needs smoothly, and providing sustainable development [3]. Due to its versatile nature, it is clear that many determinants must be taken into account for the selection of a suitable location for wind farms. Wind farms are widespread on the coasts of the Marmara Sea and in the Aegean Region of Turkey, where the wind potential is high. The wind power potential is measured and determined by Ministry of Energy and Natural Resources, especially considering the velocity of the wind, density of the wind, and direction of the wind power. It is observed that wind farms are generally clustered in the coastal areas where the potential is high. Therefore, we aim to evaluate the Central Anatolia Region of Turkey, which has extensive, unused large lands in order to establish wind farms in terms of social-environmental-economic aspects for specifying the most suitable site. There are some limitations due to various conflicts for a proper land-based wind farm in Turkey [4]. Therefore, such a multi-faceted and complex decision-making problem necessitates the application of the multi-criteria decisionmaking tools to investigate optimum alternative sites. In the case of missing crisp data, fuzzy logic applications are quite helpful. The fuzzy set approach is particularly favored in human-based modeling problems because it has the potential to generate more practical and reasonable results in the assessment of ambiguous, deficient, and limited situations. Ordinary fuzzy sets are not enough to represent the modeling of decision problems where multiple sources of ambiguity arise together [5]. In order to accomplish the defined state, numerous extensions of fuzzy sets have been presented in the literature. The application of hesitant fuzzy set (HFS) assessments makes experts’ decisions more reliable and descriptive in the decision-making process. Since the HFS is concerned with uncertainty, it can be correctly defined in terms of experts’ views. In a fuzzy environment, it is hard to make the right decision based on the

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

197

growing complexity of energy management decisions. The hesitant fuzzy technique for order preference by similarity to ideal solution (TOPSIS) is a consistent and reliable method to deal with such inaccurate and vague information environment, principally in cases where managers remain hesitant due to the uncertainty of verbal evaluations of human thinking judgments. The study aims to assess unutilized wind energy potential participation in the regional economy. In this way, it will be ensured that idle resources are used/reassessed from a green energy perspective, especially from wind energy sources. Currently, there are few wind farms in the relevant region, but new wind farms need to be established in line with the increasing energy demand. The most suitable power plant location should be determined according to mathematical models and scientific methods. Due to the nature of energy problems, the evaluation of too many factors, the existence of expert opinions from different fields, and the uncertainties in these opinions necessitate fuzzy multi-criteria decision-making applications. The study aims to select an optimum site for a new wind farm in the Central Anatolia Region of Turkey, with considering the hesitations/concerns inherent such decisions. For this purpose, we adopt a methodology utilizing the analytic hierarchy process (AHP) weighted hesitant fuzzy TOPSIS for the evaluation of the multi-criteria renewable power plant site selection problem. According to the literature review, it is the first application of the wind farm location selection problem, which is analyzed by conducting an integrated two-stage methodology based on the analytic hierarchy process (AHP) and fuzzy TOPSIS under a hesitant fuzzy environment in the Central Anatolia region of Turkey. Multi-criteria evaluations of power plant locations are available in the literature utilizing various methodologies. Although many studies have been conducted on this subject, there are still plenty of opportunities to contribute to the current literature. Therefore, the study aims to contribute to the current literature in the following ways. First, this study uses a novel integrated AHP weighted hesitant fuzzy TOPSIS approach to assessing wind farm power plant location/site selection problem from a multidimensional view specifying the influential environmental and managerial determinants at a regional level using the model of Turkey. To the best of our knowledge, the identified problem has not been issued before using hesitant fuzzy sets in the literature with the defined conditions. Second, the study investigates different parts/provincials of the Central Anatolia Region in Turkey in terms of different wind energy potential with the updated data and provides provincial/local comparisons concerning energy efficiency and energy quality. The rest of the paper is organized as follows. The next section explains brief information about the wind power situation in the world. Section 3 presents a review of earlier research on the power plant site selection problem and an overview of the methods which concentered on the fuzzy set theory. Section 4 offers the background information on the employed methodology, which is based on the hesitant fuzzy TOPSIS approach. Section 5 provides the application of the model and the obtained findings. Section 6 discusses the results with a sensitivity analysis, while Sect. 7 gives concluding remarks and future research directions.

198

B. Cayir Ervural

2 Wind Energy in the World Wind energy has a significant share in electricity generation with increasing renewable energy investments. In 2017, wind energy covered an estimated 11.6% of the European Union’s annual electricity consumption, and one of the remarkable issues is that Denmark meets 43.4% of its annual electricity consumption with wind power. Wind power is rapidly becoming a competitive and cost-effective technology worldwide. It is very advantageous to install wind farms in places with high wind energy potential. Wind farms are installed in two different ways. The first one established on lands, which is called “on-shore wind power,” and the other established at seas, which is known as “off-shore wind power.” Due to the heavy cost burden of the off-shore power plants, it is not common in Turkey. The rapid price drop in both off-shore and on-shore wind power has made it the most cost-effective opportunity for new energy generation capacity in the emerging energy market. Figure 1 shows cumulative wind power capacity by country in 2019, according to Global Wind Power Statistics. According to the Global Wind Energy Council, world wind energy capacity is approximately 539 GW, and China is the leader in the wind power market with new installations (added 19.7 GW in 2017, for a total installed capacity 188.4 GW) and followed by USA (7.0 GW), Germany (6.1 GW), the United Kingdom (4.3 GW), and India (4.1 GW) [7]. At the end of the year, the leading countries in total wind energy capacity per capita are Denmark, Ireland, Sweden, Germany, and Portugal [7]. Turkey is listed in the top ten countries in terms of capacity building with 0.8 GW and a total of 6.9 GW capacity. China 16%

2%

United States

2% 2%

Germany 36%

India Spain

3% 4%

United Kingdom France

4%

Brazil

6%

Canada 9%

16%

Fig. 1 Cumulative wind capacity in 2019 [6]

Italy Rest of the world

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

199

In order to reduce greenhouse gas emissions and to prevent global warming, there have been several attempts, and international, serious measures are taken, such as increasing the share of renewable energy with a focus on solar and wind energy. As a result of the United Nations’ climate change agreement, the Paris agreement was adopted in 2015. Turkey is also trying to shift to renewable energy sources like other countries to ensure compliance with the Paris agreement. For this reason, investments are made by taking into consideration the renewable energy potential of the country. Wind farms are essential to meet the electricity demand due to the wind power potential of the country. It is planned to increase the share of the wind power to at least 20 GW to achieve the energy goals of 2023.

3 Literature Review Assessment of the ideal site selection for power plants is a momentous issue followed by scholars and investors to identify the optimum conditions for the installation of power plants under environmental, technical, economic, and social perspectives. Several studies have conducted multi-criteria decision-making (MCDM) methods to assess optimum site selection of wind farms. The studies can be classified as geographic information system (GIS)-based, MCDM-based, and integrated fuzzy application-based researches. Table 1 summarizes the relevant literature from an overview. According to the literature review, the site selection problems focus more on MCDM methods and their combination with GIS software. Although fuzzy approaches were used in the studies, it has been observed that a hesitant fuzzy approach was very rarely used. In this context, it is observed that the hesitant fuzzy sets theory based on the TOPSIS approach for optimal wind farm selection was not issued in the literature on a real application of Central Anatolia region case in Turkey, which was the first implementation district in terms of the stated conditions.

4 Background Information 4.1 Fuzzy Set Theory Fuzzy set theory, proposed by Zadeh [48] to overcome problems involving uncertainty and ambiguity, employs linguistic terms to express decision maker’s preferences, and then, verbal statements turn into triangular fuzzy numbers. The main concepts of fuzzy sets are as follows.

200

B. Cayir Ervural

Table 1 Literature review of site selection for wind farms Author(s)

Goal

Method

Beskese et al. [7]

Wind turbine assessment

Hesitant fuzzy Hesitant fuzzy AHP is AHP-TOPSIS method employed to overcome the vagueness, which arises during criteria prioritization, and then, ranking of the alternatives is evaluated using the hesitant fuzzy TOPSIS method

Dhiman and Deb [8]

Selection of wind farm Fuzzy TOPSIS and sites located in Fuzzy COPRAS Massachusetts

Both methods offer the best alternative A3

Argin et al. [1]

Exploring the off-shore Multi-criteria site wind energy in Turkey selection with WindPRO software

Bozcaada, Bandırma, Gökceada, Inebolu, and Samandag coasts were found to be the most suitable places for the development of off-shore wind farms

Dhunny et al. [9]

Determination of optimal wind, solar, and hybrid wind-solar farming sites

Fuzzy logic system modeling

Le Morne and La Laura-Malenga were determined as potential locations for the hybrid farming

Dincer and Yuksel [47]

Assessment of renewable energy investments

Hesitant fuzzy DEMATEL, Hesitant fuzzy TOPSIS

The results show that solar energy is the most favorable one, followed by wind energy as a second renewable investment alternative

Konstantinos et al. [10]

Selecting wind farm AHP and TOPSIS installation locations in Eastern Macedonia and Thrace region, Greece

Suitable locations are determined using both methods considering social criteria

Pambudi and Nananukul [11]

Wind turbine site selection in Indonesia

Five appropriate places for the installation of wind turbines in Indonesia are South Sumatra, Papua, West Papua, Maluku, and East Nusa Tenggara provinces

Hierarchical fuzzy data envelopment analysis, hesitant fuzzy linguistic term sets

Findings

(continued)

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

201

Table 1 (continued) Author(s)

Goal

Method

Findings

Ayodele et al. [3]

Wind farm site selection in Nigeria

A multi-criteria GIS-based model

The northern part of Nigeria is in favor of a wind farm due to good wind resources

Ayodele et al. [3]

Model for wind farm A GIS-based model site selection in Nigeria employing interval type-2-Fuzzy AHP

The states of Kano, Sokoto, Plateau, Katsina, Jigawa, Kaduna, and Bauchi were obtained as several perfect places for wind farm

Cali et al. [12]

Off-shore wind farm locations in Turkey

Bozcaada is the best investment alternative while Bandirma is the worst of the three studied locations

Kim et al. [13]

Site selection of A decision-making off-shore wind farms in support tool-based Korea GIS

The study presents a decision-making tool to choose the most appropriate sites for off-shore wind farms under social, environmental, and economic factors

Chaouachi et al. [14]

Selection of off-shore wind farms in the Baltic States

Off-shore wind site assessment was implemented according to the AHP approach per each Baltic States

Vasileiou et al. [15]

Site selection of hybrid GIS-based Combining AHP and off-shore wind and multi-criteria decision GIS to identify the wave energy systems analysis most proper marine areas in Greece

Wu et al. [16]

Site selection for wind farms in China

2-tuple linguistic setting

Noorollahi et al. [17]

Wind farm site selection in western Iran

A GIS-based The study provided multi-criteria decision convenient conditions support system for wind power energy in Markazi province

Techno-economic feasibility analysis

AHP with GIS data

The proposed decision context provides a comprehensive optimal location problem

(continued)

202

B. Cayir Ervural

Table 1 (continued) Author(s)

Goal

Method

Findings

Wu et al. [18]

Site selection for off-shore wind power station (OWPS) in China

ELECTRE-III under intuitionistic fuzzy environment

The proposed intuitionistic fuzzy ELECTRE-III method is more effective in OWPS site selection

Atici et al. [2]

Site selection for the wind power plant in Turkey

A GIS-based multiple Three MCDA criteria decision methods (ELECTRE analysis III, ELECTRE-TRI, and SMAA- TRI) are employed for various aims and with various concerns

Fetanat and Khorasaninejad [19]

Off-shore wind farm site selection in Iran

Hybrid MCDM approach

Jun et al. [20]

Macro-site selection of ELECTRE-II wind/solar hybrid power station in China

The obtained results show that Erlian haote, Zhangjiakou and Yumen are very convenient for building the wind/solar hybrid power station

Satkin et al. [21]

Site selection for wind-compressed air energy plants in Iran

ArcGIS

The study identifies wind-CAES power plant sites in Iran according to three different categories

Wu et al. [22]

Choice site of solar-wind hybrid power station in China

AHP

The decision context of SWHPS site selection is built based on the AHP because of practicability

Sánchez-Lozano et al. Determination of GIS and [23] potential places for ELECTRE-TRI, on-shore wind farms in Lexicographic order Spain

Hybrid MCDM method based on the fuzzy ANP, fuzzy decision DEMATEL, and fuzzy ELECTRE approaches. In order to select the best locations for power station of OWF

The methods help for selecting the best places to host on-shore wind farms in the Region of Murcia (continued)

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

203

Table 1 (continued) Author(s)

Goal

Haaren and Fthenakis GIS-based wind farm [24] site selection in New York State

Method

Findings

Spatial multi-criteria analysis (SMCA)

The spatial multi-criteria analysis tool provided the use of GIS for site selection in New York State

Definition 1 (Fuzzy Sets) Let X is a set with a generic element of X symbolized by x, which is X = {x}. Then, fuzzy set is described as in Eq. (1): A = { x, μ A (x)|x  X }

(1)

where μ A : X → [0, 1] is the membership function of the fuzzy set A, μ A (x) ∈ [0, 1] is the degree of membership of the element x to the set A. Definition 2 (Triangular Fuzzy Number) A triangular fuzzy number can be characterized as a triplet (a, b, c); the membership function of the fuzzy number F(x) is defined as in Eq. (2): ⎧ x−a ⎪ ⎨ b−a , a ≤ x ≤ b c−x F(x) = c−b , b ≤ x ≤ c ⎪ ⎩ 0, otherwise

(2)

4.2 The Hesitant Fuzzy Sets (HFS) Hesitant fuzzy sets, one of the new extensions of the fuzzy sets, were proposed by Torra in [17], and it is utilized for modeling vagueness that emerged by hesitation in the degrees of membership functions. Let X be a reference set; an HFS on X is with regard to a function named as h that, when implemented to X, returns a subset of [0, 1]. Then, hesitant fuzzy set is defined as in Eq. (3) [17]: E = { x, h E (x)|x  X }

(3)

where h E (x) is a hesitant fuzzy element and which is a set of values between [0, 1], which indicates the membership values of the component x  X to the set E. Definition 3 (Triangular Fuzzy Hesitant Fuzzy Sets) Triangular fuzzy hesitant fuzzy sets defined in the following Eq. (4) [25]:

204

B. Cayir Ervural

    E˜ = x, f˜E(x) x  X ˜

(4)

where f˜E(x) shows a set of triangular fuzzy numbers and means the potential ˜ ˜ membership values of an element x  X to a set of E. Definition 4 (Hesitant Arithmetic Operations) The basic arithmetic operations on these elements are given as follows (Eqs. 5–10) [17, 11]:  γh

hλ =

(5)

γ ∈h

λh =



1 − (1 − γ )λ



(6)

γ ∈h



h1 ∪ h2 =

max{γ1 , γ2 }

(7)

min{γ1 , γ2 }

(8)

{γ1 + γ2 − γ1 × γ2 }

(9)

γ1 ∈h 1 ,γ2 ∈h 2 ,



h1 ∩ h2 =

γ1 ∈h 1 ,γ2 ∈h 2 ,

h1 ⊕ h2 =

γ1 ∈h 1 ,γ2 ∈h 2 ,

h1 ⊗ h2 =



{γ1 γ2 }

(10)

γ1 ∈h 1 ,γ2 ∈h 2 ,

4.3 Hesitant Fuzzy TOPSIS Hesitant fuzzy set manages the membership degree of a component to a determined set expressed by a few different potential values. HFS, one of the latest extensions of fuzzy sets, is applied for modeling ambiguity that happened by hesitation, which emerges to assign membership degrees of components. TOPSIS approach was essentially put forward by Hwang and Yoon [26]. It is one of the most widely used multicriteria decision-making techniques due to its practicality and efficiency. Hesitant fuzzy TOPSIS [23, 21] is a reliable and effective method to overcome such an inaccurate and unclear data environment, especially in cases where decision-makers remain hesitant. The stages of hesitant fuzzy TOPSIS technique are specified as follows: Step 1: Determine the positive ideal solution and negative ideal solution. The positive and negative ideal solutions are calculated as follows: 

A∗ = h ∗1 , h ∗2 , h ∗3 , . . . , h ∗n

(11)

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

205

where 

m h i j = ∪γl j ∈hl j,..., γm j ∈h m j , max γl j , . . . , γm j j = 1, 2, . . . , n h ∗j = ∪i=1 

− − − A− = h − 1 , h2 , h3 , . . . , hn

(12)

where 

m h −j = ∩i=1 h i j = ∩γl j ∈hl j,..., γm j ∈h m j , min γl j , . . . , γm j j = 1, 2, . . . , n

(13)

Step 2: Compute the positive and negative solution distances. In this study, a weighted hesitant normalized Hamming distance [27] is employed. The distance of each alternative from D ∗ and D − is computed as follows in Eqs. (14) and (15): Di∗ =

n

w j h i j − h ∗j

(14)

w j h i j − h −j

(15)

j=1

Di−

=

n j=1

where w j states the crisp weight of the jth criterion calculated by AHP. Step 3: Obtain the closeness coefficient (Ci ). The following Eq. displays the calculation of the closeness coefficient. Ci =

Di−

Di− + Di+

(16)

Step 4: Rank the options. According to closeness coefficients obtained from step 3, the option with the highest Ci value will be the most appropriate alternative.

5 Application of the Hesitant Fuzzy TOPSIS Model 5.1 Problem Definition Wind turbines cannot be installed everywhere due to many factors such as settlement areas, population density, historical and archeological values, airports, harbors, rivers, and topographic reasons, etc. Overall, Turkey has a high wind energy potential due to its geographical location, but North Aegean, Marmara, Eastern Mediterranean, and Southeast Anatolia have the highest wind energy potential.

206 Table 2 Wind potential of seven regions in Turkey [29]

B. Cayir Ervural Region

Annual average wind speed (m/s)

Annual average wind density (W/m2 )

Marmara

3.3

51.9

Southeast Anatolia

2.7

29.3

Aegean

2.6

23.5

Mediterranean

2.5

21.4

Black Sea

2.4

21.3

Central Anatolia

2.5

20.1

East Anatolia

2.1

13.2

Turkey Average

2.5

24.0

Annual average wind speed and wind density values for seven main regions of Turkey were presented in Table 2, and Ministry of Energy and Natural Resources [28] published a wind energy atlas which shows a yearly average wind speed distribution of 50 m height. It is observed that the wind energy potential in the Central Anatolia region is sufficient to establish a wind power plant when compared to other regions. The Central Anatolia region consisting of thirteen provinces which are, namely Aksaray, Ankara, Çankırı, Eski¸sehir, Karaman, Kayseri, Kırıkkale, Kır¸sehir, Konya, Nev¸sehir, Ni˘gde, Sivas, and Yozgat, illustrated in Fig. 2, and more details are given in Fig. 3. Five sites have been selected in order to evaluate the locations that have high wind energy potential but remain idle in the Central Anatolia region. The related alternatives determined as Konya, Karaman, Aksaray, Yozgat, and Ni˘gde. In this study, we employ one of the recently proposed decision analysis processes that combine HFSs to the TOPSIS approach for choosing the best location for wind farms under the verbal assessment structure. We evaluate the options from different angles before determining the best choice. Through a detailed literature review and

Fig. 2 Central Anatolia region in Turkey

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

207

Fig. 3 Central Anatolia region with provinces

interviews with the energy policy-makers, various important criteria and sub-criteria are investigated. For the application of the study, we investigate five alternative locations, four criteria, and thirteen sub-criteria considering the judgments of three experts from the energy market.

5.2 Defining the Evaluation Criteria The evaluation criteria of the fuzzy model are defined as follows, and the hierarchical model for wind farm site selection is shown in Fig. 4. Technical criteria Average wind speed (C1): Means to wind speed at 30, 50, and 100 m3 altitude. This criterion is measured by the average speed over. It is calculated by the effect of annual wind speed changes representing long-term variations [30, 24, 15]. Average wind power density (C2): It shows how much wind energy is available in a place for wind turbine production. Defines the kinetic power created by the wind in the unit (W/m2 ) [31]. Distance to power lines (C3): Wind farms should be established close to the electricity grid. Long power lines between wind farms and the power grid lead to energy losses and large infrastructure costs [30, 24].

208

B. Cayir Ervural

Fig. 4 A hierarchical model for wind farm site selection

Environmental criteria Impact on the ecosystem (C4): It measures the wind turbines influence on the environment in terms of protection of ecological balance and the damage on the quality of the plants’ ecosystem due to negative effects of flora and fauna and biodiversity such as the paths of migration for birds [16]. Protected area (C5): Protected area criterion comprises such as national parks, wildlife refuges, protected areas, the historical heritage, and natural national monuments [22]. Distance to airports (C6): Wind turbines may cause disturbances in the radar signal and would require a significant distance around airports [24]. Visual impact (C7): Wind turbines have some adverse effects on areas such as noise, visual pollution, or blade shadow effects [16].

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

209

Economical criteria Land cost (C8): The requirement of a large district makes land cost as significant as capital cost and it can change region to region [32, 18]. Operation and maintenance cost (C9): It considers costs of the power plant in the procedure of operation and maintenance [31]. Incentives (C10): Long-term incentives are essential issue to implement quick wind turbine installation to promote wind power in terms of feed-in tariffs or quotes/permits. Incentives provide speed up wind power projects and support market participants to maximize their efficiency by falling wind integration expenses [29]. Social criteria Social acceptability (C11): In general, wind turbines/farms are accepted but sometimes can be met with strong local opposition. There should be taken supportive actions of the related region to handle the process smoothly. The public can oppose such projects because of environmental or social effects [31, 20]. Distance from the residential area (C12): It is a fundamental criterion in wind farm positioning. Reasonable distances from settlements should be characterized to avoid noise, disturbance, and natural surroundings should be defined. Wind farm locations should be within an acceptable distance from residential areas to minimize transmission losses [19, 24]. Land use (C13): Improving the existing land area for wind farm development is a critical issue for power plant installation. Infeasible wind farm areas based on land use and geological constraints (topographic elevation and land use distribution, land with major gradients, and steep slopes) cause various problems for infrastructure works. Plain areas are advantageous with minimal mountain ranges that are well suited for the development of wind farms [19, 33]. Wind farms may not be easily installed everywhere due to several essential restrictions. The identified limitations should be considered in a general perspective when investigating the optimal site location for wind power. Because ignoring any of the problematic components leads to waste of time, money, energy, and effort. Therefore, complex terrains for wind farming should be efficiently planned in the process of selecting the best locations for wind farms by collecting data. For instance, when evaluating appropriate sites for wind power distance to transmission lines affect the cost of turbine installation and besides the closeness of the road and the slope of the land which affect construction and maintenance costs. Also, the capacity factor is one of the critical issues to be considered in the electricity generation process, which is closely related to the wind speed and the elevation of the region, which all affect the performance of the wind turbine. Turkey has a severe potential for wind power, and more particularly, the estimated wind potential exceeds 48 GW the locations having wind speed is more than 7 m/s in 50 m of altitude. However, it is not possible to install wind turbines in all places due to the specified constraints clarified in the study.

210

B. Cayir Ervural

For this reason, an approach that evaluates different dynamics under uncertainty with a holistic and scientific approach is needed in order to determine potential areas with reasonable and adequateway.

5.3 Application of the Model In this study, the weights of the criteria were obtained by the evaluations of experienced experts in the energy management process. For this reason, surveys covering criteria and renewable energy options were conducted and evaluated by three experts with verbal evaluations. As a result of expert assessments, pairwise comparison matrices (Table 3) and the criteria weights (Table 4) are constructed. According to the expert’s opinions, a pairwise comparison matrix is created to evaluate the four main criteria. The importance weights of sub-criteria calculations are obtained using the same procedure. Table 4 shows the expert evaluations of the sub-criteria with respect to the main criteria. The importance weights of sub-criteria with respect to the main criteria are determined according to the local priorities of the sub-criteria. The priorities of each sub-criteria depend on the goal is called as global priorities and computed by multiplying the local priorities with weights of the main criteria. Local and global priorities of the sub-criteria are given in Table 5. The consistency ratio (CR) was calculated by performing the consistency analysis of all pairwise comparison matrices. While performing all the pairwise comparisons, the consistency value is investigated by employing eigenvalue, λmax , to compute the consistency index, CI as follows: C I = (λmax − n)/(n − 1), where n is the matrix size, and the consistency ratio was calculated according to CI and random consistency ratio (RI) based on matrix size referenced from [34]. The obtained CR value is expected to be less than 0.10. It means the judgments are consistent, and the emerged weights can be used. Table 6 shows the transformation rules of linguistic variables. In the first stage of hesitant fuzzy TOPSIS, positive and negative ideal solutions are defined. For this purpose, maximum and minimum membership values for each criterion are obtained. Table 3 Main criteria matrix Technical

Environmental

Economic

Social

Priority vector

Technical

1

2

3

5

0.482

Environmental

1/2

1

2

3

0.272

Economic

1/3

1/2

1

2

0.158

Social

1/5

1/3

1/2

1

0.088

= 1.000

λmax = 4.014, CI = 0.0048, RI = 0.90, CR = 0.005 < 0.10 OK

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

211

Table 4 Matrices for evaluating the sub-criteria under technical, environmental, economic, and social criteria Technical

C1

C2

C3

Priority vector

Average wind speed (C1)

1

1

2

0.411

Average wind power density (C2)

1

1

1

0.328

Distance to power lines(C3)

1/2

1

1

0.261

= 1.000

λmax = 3.053, CI = 0.0268, RI = 0.58, CR = 0.046 < 0.10 OK Environmental

C4 C5

C6

C7

Priority vector

Impact on ecosystem(C4)

1

1

1

2

0.282

Protected area (C5)

1

1

2

2

0.337

Airports (C6)

1

1/2

1

2

0.240

Visual impact (C7)

1/2 1/2

1/2

1

0.141

= 1.000

λmax = 4.06, CI = 0.0202, RI = 0.90, CR = 0.022 < 0.10 OK Economical

C8

C9

C10 Priority vector

Land cost (C8)

1

2

1

0.411

Operation and maintenance cost (C9)

1/2

1

1

0.261

Incentives (C10)

1

1

1

0.328

= 1.000

λmax = 3.053, CI = 0.0268, RI = 0.58, CR = 0.046 < 0.10 OK Social

C11 C12 C13 Priority vector

Social acceptability (C11)

1

2

2

0.500

Distance from the residential area (C12)

1/2

1

1

0.250

Land use (C13)

1/2

1

1

0.250

= 1.000

λmax = 3.00, CI = 0.00, RI = 0.58, CR = 0.00 < 0.10 OK Table 5 Relative priorities of criteria and sub-criteria Main criteria

Priorities

Sub-Criteria

Local priorities

Global priorities

Technical

0.482

C1

0.411

0.198

C2

0.328

0.158

C3

0.261

0.126

C4

0.282

0.077

C5

0.337

0.092

C6

0.240

0.065

Environmental

Economic

Social

0.272

0.158

0.088

C7

0.141

0.038

C8

0.411

0.065

C9

0.261

0.041

C10

0.328

0.052

C11

0.500

0.044

C12

0.250

0.022

C13

0.250

0.022

212

B. Cayir Ervural

Table 6 Transformation rules of linguistic variables [35]

Linguistic variable

Triangular fuzzy number

Very low (VL)

(0,0,1)

Low (L)

(0,1,3)

Medium low (ML)

(1,3,5)

Fair (F)

(3,5,7)

Medium high (MH)

(5,7,9)

High (H)

(7,9,10)

Very High (VH)

(9,10,10)

The separation measure of each alternative from the ideal solution is computed according to Eqs. (14) and (15). And then, the relative closeness value for each alternative are calculated. Table 7 shows the hesitant fuzzy decision matrix, and the obtained values for each alternative and the result of ranks are given in Table 8. Table 7 Hesitant fuzzy decision matrix C3

C4

C5

A1

{0.2 0.3 0.5} {0.4}

C1

C2

{0.4 0.5}

{0.4 0.5 0.6 0.8}

{0.2 0.3 0.6} {0.6 0.7}

A2

{0.3 0.4 0.7} {0.5 0.6 0.7}

{0.3 0.5 0.8}

{0.5 0.6 0.7} {0.1 0.2 0.4} {0.5 0.7 0.8}

A3

{0.3 0.5 0.6 0.8}

{0.5 0.7}

{0.7 0.8 0.9} {0.3 0.5}

A4

{0.4 0.5 0.7} {0.4 0.5}

{0.2 0.4}

{0.6 0.7}

A5

{0.1 0.2 0.4} {0.3 0.4}

{0.4 0.6}

{0.2 0.4 0.5} {0.2 0.4 0.5 0.6}

C7

C8

C9

C10

C11

C12

C13

A1

{0.2 0.3}

{0.4 0.6 0.8}

{0.1 0.2 0.3}

{0.6 0.7}

{0.2 0.4 0.5 0.6}

{0.5}

{0.1 0.3 0.5}

A2

{0.1 0.4 0.5} {0.3 0.5 0.6}

{0.2 0.3} {0.4}

{0.3 0.4}

{0.7 0.9} {0.3 0.4 0.7}

A3

{0.2 0.3 0.5} {0.4}

{0.3 0.5} {0.5 0.6}

{0.3 0.4 0.6} {0.4 0.6 0.7}

A4

{0.3 0.5 0.6 0.7}

{0.2 0.4} {0.6 0.7}

{0.3 0.3}

A5

{0.3 0.4 0.5} {0.5 0.7 0.9}

{0.3 0.5 0.6}

{0.3 0.3 0.4} {0.3 0.5 0.6}

{0.2 0.3 0.4}

{0.3 0.4 0.6}

{0.8}

C6

{0.4 0.5 0.7}

{0.1 0.3 0.4} {0.3 0.4} {0.5}

{0.4 0.5 0.6 0.7}

{0.3 0.4} {0.1 0.3 0.4} {0.4 0.7 0.8}

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

213

Table 8 Rank of alternatives Alternatives

D∗i

D− i

Ci

Rank

0.329

0.246

0.429

3

A1

Karaman

A2

Konya

0.279

0.296

0.516

2

A3

Aksaray

0.269

0.306

0.532

1

A4

Nigde

0.327

0.244

0.421

4

A5

Yozgat

0.337

0.238

0.414

5

In this study, a weighted hesitant normalized Hamming distance is employed, which was proposed by Zhang and Wei [27].  1  h 1σ (i) − h 2σ (i)  h1 − h2 = l i=1 l

(17)

where h 1 , h 2 are HFEs, and l is the number of elements in an HFE and identified as the length of HFE. According to closeness coefficients, the option with the highest C i value will be the most appropriate alternative. Option A3 , with the highest value (0.532), is the best suitable site, among others. It is followed by A2 , A1 , and A4 , with the values of 0.516, 0.429, and 0.421, respectively. However, option A5 is the least favorable place for a wind turbine location selection problem.

6 Results and Discussion Due to some uncertainties and hesitations of decision-makers’ views on wind farm site selection problem which inherent a multidimensional structure, the hesitant fuzzy approaches help to analyze the problem rationally and overcome the apparent factor uncertainty. Identification of the best location for a wind farm is essential to realize correct investment decisions and to obtain efficient energy production under diverse and multiple restrictions. According to the obtained ranking values presented in Table 8, it can be concluded that Aksaray (A2) ranks first with the highest priority rate 0.532. Konya (A3 ) came as the second-highest priority with a ranking value of 0.516. Karaman (A4) is the third highest with a priority rank value of 0.429, and the fourth one is Nigde (A1), with a priority rank value of 0.421. Yozgat (A5) has the lowest priority, with a value of 0.414. All results from the proposed approach support diverse official reports, such as the ministry’s infrastructure studies, the government’s feasibility reports, and energy

214

B. Cayir Ervural

policies [28, 36]. The study offers a significant advantage, especially for the effectively utilization of unused wind energy potential and investors need to consider the analysis results before building a power plant. Wind farms are most heavily concentrated in the Marmara and Aegean regions in Turkey. During this study, the absence of wind farms in the selected provinces was taken into account. Thirteen provinces in the Central Anatolia Region were evaluated in terms of their wind energy potential and power plant installability and then the number of provinces were reduced to the top five. As mentioned before, although the geographical differences between provinces is not severe, some factors such as slope, closeness to the transformer center, and the prevalence of residential and protected areas affect the decision to establish a power plant. In this respect, the study is expected to guide decision-makers and investors in energy management. Sensitivity analysis is a basic method to reveal the behavior of the model created by the system and decide the feasibility of the model. The main criteria weights were equally distributed among the sub-criteria. Using modified values of criteria weights, we try to measure the changes in the result. In particular, five cases were investigated conducting a sensitivity analysis, but the results remained mostly the same. The derived criteria weights are given in Table 9. The obtained sensitivity results are demonstrated in Fig. 5. According to the rankings, third, second, first, and fourth options are generally in the same priority order except for fifth. The fifth alternative rank has been changed depending on criteria weights. In case 4, the weight of economic criteria has been increased, and the fifth alternative has ranked first. These findings indicate that the hesitant fuzzy approach is sensitive to alterations in the weights of the criteria. Table 9 Sensitivity analysis for criteria weights Technical Environmental Economic Social Ranking Case Current 0 situation

0.482

0.272

0.158

0.088

A3 > A2 > A1 > A4 > A5

Case All equal 1

0.250

0.250

0.250

0.250

A3 > A5 > A2 > A1 > A4

Case Technical 0.400 2 criteria weight increased

0.200

0.200

0.200

A3 > A2 > A5 > A1 > A4

Case Environmental 0.200 3 criteria weight increased

0.400

0.200

0.200

A3 > A2 > A5 > A1 > A4

Case Economic 0.200 4 criteria weight increased

0.200

0.400

0.200

A5 > A3 > A2 > A1 > A4

Case Social criteria 5 weight increased

0.200

0.200

0.400

A3 > A2 > A5 > A1 > A4

0.200

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

215

0.55 0.50 0.45 0.40 0.35 A1 0.30

A2

A3

A4

A5

A1

case 0 0.429

case 1 0.419

case 2 0.410

case 3 0.434

case 4 0.421

case 5 0.408

A2

0.516

0.488

0.514

0.489

0.438

0.510

A3

0.532

0.517

0.519

0.533

0.489

0.525

A4

0.421

0.372

0.385

0.393

0.375

0.335

A5

0.414

0.495

0.459

0.482

0.546

0.494

Fig. 5 Sensitivity analysis results when the weights of criteria are changed

7 Conclusion In parallel with the increasing energy need, turning to renewable energy resources rather than fossil resources is a more attractive and advantageous strategy worldwide. Wind energy has become one of the most widely used energy types due to its many advantages. Therefore, site selection decision for wind farms emerges as an important issue and is considered as the first step in the efficient energy generation process [37] given the social-economic impact on the community and environmental issues. The existence of many factors and various stakeholder’s interests that complicate the problem to be considered in the wind farm location problem requires a systematic approach and a holistic understanding. In addition, the emerging uncertainties and hesitant nature of decision-makers require the use of hesitant fuzzy approaches. Turkey has great wind energy potential as well as other renewable energy sources. In this study, we aimed to evaluate the inclusion of existing but not used energy resources into the energy generation system. The results show that Aksaray is the most advantageous position in terms of installation of the wind farm, and then Konya, Karaman, Ni˘gde, and finally Yozgat in the subsequent priorities in the Central Anatolia Region in Turkey. The sensitivity analysis confirms that the obtained results of the hesitant fuzzy model are consistent. The applied two-staged integrated hesitant fuzzy model provides flexibility and reality due to covering decision-makers’ intuitive on the complex problem nature under fuzzy conditions. The findings of the study guide urban planners and energy policy-makers to better specify optimum locations for wind farms.

216

B. Cayir Ervural

The application of the proposed method has some limitations due to the collected empirical data just related to the Central Anatolia Region in Turkey. To tackle computational tasks during sensitivity analysis, algorithm-based methods may be practical and more systematic. By integrating with other approaches, the method can be strengthened with several stages of implementation. For future research, the study can be expanded taking into account the internal dependence of the criteria, and their interactions on the model can be considered by combining other hierarchically based MCDM methods such as analytic network process. Moreover, the study can be expanded by adding new criteria, new alternatives, and constraints to the model that integrates with various new fuzzy decision-making techniques such as neutrosophic fuzzy set, Pythagorean fuzzy set, and spherical fuzzy sets.

References 1. News E (2017) Energy resources will be exhausted in 40 years: expert, energy news, ET energyworld. Energy Econ Times. https://energy.economictimes.indiatimes.com/news/coal/energyresources-will-be-exhausted-in-40-years-expert/57846890 2. Cayir Ervural B, Evren R (2019) A policy proposal for effective energy management. Springer, Cham. pp 297–306. https://doi.org/10.1007/978-3-030-03317-0_25 3. Ervural BC, Ervural B (2018) Improvement of grey prediction models and their usage for energy demand forecasting. J Intel Fuzzy Syst 34(4). https://doi.org/10.3233/JIFS-17794 4. Cayir Ervural B (2020) A combined methodology for evaluation of electricity distribution companies in Turkey. J Intell Fuzzy Syst 38(1):1059–1069. https://doi.org/10.3233/JIFS179468 5. Beg I, Rashid T (2013) TOPSIS for hesitant fuzzy linguistic term sets. Int J Intell Syst 28(12):1162–1171. https://doi.org/10.1002/int.21623 6. Statistics GWP (2020) Global wind power statistics—wind energy international platform. https://library.wwindea.org/global-statistics/ 7. Zervos A (2018) Renewables 2018 global status report. https://www.ren21.net/gsr-2018/ 8. Cali U, Erdogan N, Kucuksari S, Argin M (2018) TECHNO-ECONOMIC analysis of high potential offshore wind farm locations in Turkey. Energy Strategy Rev 22:325–336. https:// doi.org/10.1016/j.esr.2018.10.007 9. Kim CK, Jang S, Kim TY (2018) Site selection for offshore wind farms in the southwest coast of South Korea. Renew Energy 120:151–162. https://doi.org/10.1016/j.renene.2017.12.081 10. Satkin M, Noorollahi Y, Abbaspour M, Yousefi H (2014) Multi criteria site selection model for wind-compressed air energy storage power plants in Iran. Renew Sustain Energy Rev 2:579–590. Elsevier Ltd. https://doi.org/10.1016/j.rser.2014.01.054 11. Xia M, Xu Z (2011) Hesitant fuzzy information aggregation in decision making. Int J Approximate Reasoning 52(3):395–407. https://doi.org/10.1016/j.ijar.2010.09.002 12. Beskese A, Camci A, Temur GT, Erturk E (2020) Wind turbine evaluation using the hesitant fuzzy AHP-TOPSIS method with a case in Turkey. J Intell Fuzzy Syst 38(1):997–1011. https:// doi.org/10.3233/JIFS-179464 13. Jun D, Tian-Tian F, Yi-Sheng Y, Yu M (2014) Macro-site selection of wind/solar hybrid power station based on ELECTRE-II. In: Renewable and sustainable energy reviews, vol 35. Elsevier Ltd. pp 194–204. https://doi.org/10.1016/j.rser.2014.04.005

An Integrated Hesitant Fuzzy Decision Model for Sustainable …

217

14. Konstantinos I, Georgios T, Garyfalos A (2019) A decision support system methodology for selecting wind farm installation locations using AHP and TOPSIS: case study in Eastern Macedonia and Thrace region, Greece. Energy Policy 132:232–246. https://doi.org/10.1016/j.enpol. 2019.05.020 15. Krewitt W, Nitsch J (2003) The potential for electricity generation from on-shore wind energy under the constraints of nature conservation: a case study for two regions in Germany. Renew Energy 28(10):1645–1655. https://doi.org/10.1016/S0960-1481(03)00008-9 16. Dai K, Bergot A, Liang C, Xiang WN, Huang Z (2015) Environmental issues associated with wind energy—a review. In: Renewable energy, vol 75. Elsevier Ltd. pp 911–921. https://doi. org/10.1016/j.renene.2014.10.074 17. Torra V (2010) Hesitant fuzzy sets. Int J Intell Syst 25(6):n/a–n/a. https://doi.org/10.1002/int. 20418 18. Azizi A, Malekmohammadi B, Jafari HR, Nasiri H, Amini Parsa V (2014) Land suitability assessment for wind power plant site selection using ANP-DEMATEL in a GIS environment: case study of Ardabil province, Iran. Environ Monit Assess 186(10):6695–6709. https://doi. org/10.1007/s10661-014-3883-6 19. Noorollahi Y, Yousefi H, Mohammadi M (2016) Multi-criteria decision support system for wind farm site selection using GIS. Sustain Energy Technol Assessments 13:38–50. https:// doi.org/10.1016/j.seta.2015.11.007 20. Fetanat A, Khorasaninejad E (2015) A novel hybrid MCDM approach for offshore wind farm site selection: a case study of Iran. Ocean Coast Manag 109:17–28. https://doi.org/10.1016/j. ocecoaman.2015.02.005 21. Onar SC, Oztaysi B, Kahraman C (2014) Strategic decision selection using hesitant fuzzy TOPSIS and interval type-2 fuzzy AHP: a case study. Int J Comput Intell Syst 7(5):1002–1021. https://doi.org/10.1080/18756891.2014.964011 22. Schallenberg-Rodríguez J, Notario-del Pino J (2014) Evaluation of on-shore wind technoeconomical potential in regions and islands. Appl Energy 124:117–129. https://doi.org/10. 1016/j.apenergy.2014.02.050 23. Liu H, Rodríguez RM (2014) A fuzzy envelope for hesitant fuzzy linguistic term set and its application to multicriteria decision making. Inform Sci 258(Supplement C):220–238. https:// doi.org/10.1016/j.ins.2013.07.027 24. Baseer MA, Rehman S, Meyer JP, Alam MM (2017) GIS-based site suitability analysis for wind farm development in Saudi Arabia. Energy 141:1166–1176. https://doi.org/10.1016/j.ene rgy.2017.10.016 25. Zhao X, Lin R, Wei G (2014) Hesitant triangular fuzzy information aggregation based on Einstein operations and their application to multiple attribute decision making. Expert Syst Appl 41(4 PART 1):1086–1094. https://doi.org/10.1016/j.eswa.2013.07.104 26. Hwang C-L, Yoon K (1981) Concluding remarks. In: Multiple attribute decision making. Springer, Berlin Heidelberg. pp 207–225. https://doi.org/10.1007/978-3-642-48318-9_5 27. Zhang N, Wei G (2013) Extension of VIKOR method for decision making problem based on hesitant fuzzy set. Appl Math Model 37(7):4938–4947. https://doi.org/10.1016/j.apm.2012. 10.002 28. YEGM (2017) General directorate of renewable energy-wind energy potential atlas. https:// www.eie.gov.tr/YEKrepa/REPA-duyuru_01.html 29. Erdogdu, E. (2009). On the wind energy in Turkey. Renew Sustain Energy Rev 13(6–7):1361– 1371. Pergamon. https://doi.org/10.1016/j.rser.2008.09.003 30. Wu Y, Geng S, Xu H, Zhang H (2014) Study of decision framework of wind farm project plan selection under intuitionistic fuzzy set and fuzzy measure environment. Energy Convers Manage 87:274–284. https://doi.org/10.1016/j.enconman.2014.07.001 31. Wu Y, Zhang J, Yuan J, Geng S, Zhang H (2016) Study of decision framework of offshore wind power station site selection based on ELECTRE-III under intuitionistic fuzzy environment: a case of China. Energy Convers Manage 113:66–81. https://doi.org/10.1016/j.enconman.2016. 01.020

218

B. Cayir Ervural

32. Van Haaren R, Fthenakis V (2011) GIS-based wind farm site selection using spatial multicriteria analysis (SMCA): evaluating the case for New York state. Renew Sustain Energy Rev 15(7):3332–3340. https://doi.org/10.1016/j.rser.2011.04.010 33. Atici KB, Simsek AB, Ulucan A, Tosun MU (2015) A GIS-based multiple criteria decision analysis approach for wind power plant site selection. Utilities Policy 37:86–96. https://doi. org/10.1016/j.jup.2015.06.001 34. Saaty RW (1987) The analytic hierarchy process-what it is and how it is used. Math Model 9(3–5):161–176. https://doi.org/10.1016/0270-0255(87)90473-8 35. Chen C-T (2000) Extensions of the TOPSIS for group decision-making under fuzzy environment. Fuzzy Sets Syst 114(1):1–9. https://doi.org/10.1016/S0165-0114(97)00377-1 36. IRENA (2017) Data and statistics—IRENA REsource. https://resourceirena.irena.org/gateway/ dashboard/?topic=15&subTopic=42 37. Cayir Ervural B, Zaim S, Delen D (2018) A two-stage analytical approach to assess sustainable energy efficiency. Energy 164:822–836. https://doi.org/10.1016/J.ENERGY.2018.08.213 38. Dhiman HS, Deb D (2020) Fuzzy TOPSIS and fuzzy COPRAS based multi-criteria decision making for hybrid wind farms. Energy 202:117755. https://doi.org/10.1016/j.energy.2020. 117755 39. Argin M, Yerci V, Erdogan N, Kucuksari S, Cali U (2019) Exploring the offshore wind energy potential of Turkey based on multi-criteria site selection. Energy Strategy Rev 23:33–46. https:// doi.org/10.1016/j.esr.2018.12.005 40. Dhunny AZ, Doorga JRS, Allam Z, Lollchund MR, Boojhawon R (2019) Identification of optimal wind, solar and hybrid wind-solar farming sites using fuzzy logic modelling. Energy 188:116056. https://doi.org/10.1016/j.energy.2019.116056 41. Pambudi G, Nananukul N (2019) A hierarchical fuzzy data envelopment analysis for wind turbine site selection in Indonesia. Energy Reports 5:1041–1047. https://doi.org/10.1016/j. egyr.2019.08.002 42. Ayodele TR, Ogunjuyigbe ASO, Odigie O, Munda JL (2018) A multi-criteria GIS based model for wind farm site selection using interval type-2 fuzzy analytic hierarchy process: the case study of Nigeria. Appl Energy 228:1853–1869. https://doi.org/10.1016/j.apenergy.2018.07.051 43. Chaouachi A, Covrig CF, Ardelean M (2017) Multi-criteria selection of offshore wind farms: case study for the Baltic states. Energy Policy 103:179–192. https://doi.org/10.1016/j.enpol. 2017.01.018 44. Vasileiou M, Loukogeorgaki E, Vagiona DG (2017) GIS-based multi-criteria decision analysis for site selection of hybrid offshore wind and wave energy systems in Greece. In: Renewable and sustainable energy reviews, vol 73. Elsevier Ltd. pp 745–757. https://doi.org/10.1016/j. rser.2017.01.161 45. Wu Y, Chen K, Zeng B, Yang M, Li L, Zhang H (2017) A cloud decision framework in pure 2-tuple linguistic setting and its application for low-speed wind farm site selection. J Clean Prod 142:2154–2165. https://doi.org/10.1016/j.jclepro.2016.11.067 46. Sánchez-Lozano JM, Teruel-Solano J, Soto-Elvira PL, Socorro García-Cascales M (2013) Geographical information systems (GIS) and multi-criteria decision making (MCDM) methods for the evaluation of solar farms locations: case study in south-eastern Spain. Renew Sustain Energy Rev 24:544–556. https://doi.org/10.1016/j.rser.2013.03.019 47. Dinçer H, Yüksel S (2019) Multidimensional evaluation of global investments on the renewable energy with the integrated fuzzy decision-making model under the hesitancy. Int J Energy Res 43(5):1775–1784. https://doi.org/10.1002/er.4400 48. Zadeh LA (1965) Information and control. Fuzzy sets 8(3):338–353

Rough Set-Based Multi-Criteria Decision Analysis Methods in Sustainability Assessment of Photovoltaic Projects Jing Li and Wenyan Song

Abstract For a photovoltaic (PV) project, there are multiple stages during its life cycle, including investment, construction, and operation. To promote the sustainability of PV projects, it is essential to use appropriate methods to handle the key decision-making issues at different life-cycle stages. Site selection is an important step in the investment phase, which has a significant effect on the potential electricity generation and the socio-economic and environmental benefits. In this context, various factors need to be considered to select a suitable location for a sustainable PV project, for example, solar resources, transportation condition, geology, and influences of PV project on the local environment and society, etc. Therefore, site selection of PV power stations could be seen as a multiple-criteria decision-making (MCDM) process. A number of MCDM methods have been used for site selection; however, many of them do not consider the judgments’ vagueness, and suppose experts are rational, which are not consistent with the reality. In this chapter, an integrated location selection model for PV power stations is constructed. It utilizes rough sets and prospect theory (PT) improving the limitations of technique for order performance by similarity to ideal solution (TOPSIS) method. Rough sets can handle the imprecise and vague information flexibly, without no preset assumptions; PT could effectively manipulate the bounded rationality. In the end, this hybrid framework is applied to a real case to demonstrate its effectiveness and feasibility.

W. Song (B) School of Economics and Management, Beihang University, Beijing 100191, China J. Li College of Economics and Management, Taiyuan University of Technology, Taiyuan 030024, China e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 J. Ren (ed.), Energy Systems Evaluation (Volume 2), Green Energy and Technology, https://doi.org/10.1007/978-3-030-67376-5_9

219

220

J. Li and W. Song

1 Introduction Solar energy, one of the renewable energy, has been explored and utilized by many countries. One of its important applications is photovoltaic (PV) power generation. To achieve ecological civilization and sustainable development, China has successively introduced a series of policies to promote photovoltaic projects. In this context, newly added photovoltaic installed capacity has been increased from 0.01 gigawatt (GW) in 2006 to 30.11 GW in 2019, increased by 3011 times (China Electric Power Yearbook, 2007–2020). Since 2013, China has the greatest PV installed capacity in the world. Although the PV projects are developing rapidly in China, there are still a lot of problems need to be solved. For example, the technical level in the PV field is not mature; energy policies, PV electricity price, and the PV power generation system are not stable, which make investors face a lot of uncertain information in the decision making of implementing PV projects. In addition, the sustainable development of PV projects is not paid much attention. In spite of no pollutant emissions during power generation, environmental issues, such as dust, light, and domestic wastewater pollution, are not taken into consideration. The PV power generation projects involve a systematic decision-making process including a number of decision issues at the different life-cycle stages. It starts from investment decision making and ends with operational decisions. Therefore, it is essential to establish a complete and effective decision-making framework for PV projects, which can help managers comprehensively evaluate the decision issues, make accurate judgments, and promote the sustainable development. In this chapter, sustainable site selection, a key decision problem in the phase of investment, will be taken as an example to evaluate the sustainability of PV power generation projects. Site selection is a multi-criteria decision-making (MCDM) process, which needs to consider a lot of factors, such as climate, solar energy, topography, transport condition, etc. [3]. In the previous research, some criteria have been used to evaluate the alternative locations of PV power generation projects. For instance, Tahri et al. [23] classify the criteria into four types, i.e., climate, land use, orography and location, and discovery that climate is the most influential indicator for the site evaluation of PV power generation projects. Uyan [30] constructs an indicator system that includes economic and environmental aspects of the sites. Usually, solar energy, economic factors, and environmental factors are frequently adopted by industry and academia to assess the alternative sites of the PV projects. However, for the sustainable development of PV projects, social criteria are also important, which are often neglected by scholars. For example, local public acceptance is an important social criterion for site evaluation [4]. This is because that PV power projects will hardly conducive to implementation if the local public are opposed to them. In Western China, solar energy is rich and a lot of PV power generation projects are constructed in this region, whereas many people are ethnic minorities. They have religious beliefs and may conflict with some projects. Thus, before implementation of the projects, it is necessary to investigate the public’s willingness to PV power generation. To sum up, solar energy, economic factors, environmental factors, and social factors should

Rough Set-Based Multi-Criteria Decision Analysis Methods …

221

all be considered at the same time to assess the sites, which would maintain the PV power project sustainable. Different criteria of site selection of PV power projects are summarized as follows: Resource criteria Sunshine duration [1, 4, 24]: It means that how long the duration of sunshine under certain conditions. The PV power plant should be built in places with longer sunshine duration. Solar radiation [5, 25]: It is the power per unit area, in the form of electromagnetic radiation, which is emitted from the sun to the ground. The stronger the solar radiation, the higher potential that PV power plants generate electricity. Economic criteria Cost [17]: Expenses that spending on a project, which includes land acquisition costs, construction cost, transportation cost, operation cost, labor cost, etc. Location [14, 30]: The meaning is whether it is convenient to access to the PV power plant. If the plant is far from the road and the transmission grid, it will affect the project cost. Orography [1, 26]: It describes the characteristics of topography. The PV plants should be built in a place with a gentle terrain and no shelter to the direction of the sun. It can lower construction costs and generate more electricity. Environmental criteria Land use [7, 23]: There are different types of land in China, such as cultivated land, garden land, woodland, residential area, industrial land, etc. PV power projects can only be built on industrial land. To protect ecological environment, they cannot be constructed on cultivated land, garden land, and woodland. Environmental influence [11]: PV projects have influences on the local environment. For example, there are no emissions of pollutants during power generation, and there may be dust pollution and soil erosion in the process of constructing PV plants. Social criteria Public support [2]: In China, different places may have different custom. The desire of the local people for PV power plants must be considered and investigated during site selection. Policy support [24]: The development of PV power project mainly relies on the policies provided by the government. PV power plants should be built in places where the government can offer policy support, for instance, subsidies and tax incentives. Impact on the local economy [7, 11]: Implementing PV power projects has impacts on the local economy, such as increasing residents’ income, jobs, and fiscal revenue. In this study, the criteria above would be utilized to evaluate the sites to improve the sustainability of PV power projects. From the above analysis, site selection of PV power project can be regarded as a MCDM process. In the past, a lot of researches apply MCDM approaches to

222

J. Li and W. Song

select suitable sites of energy projects including PV power farms. For instance, Jun et al. [15] evaluate seven locations of new energy power plants through ELECTRE-II method. Tahri et al. [23] adopt analytical hierarchy process (AHP) approach to select solar farm stations. Although a number of MCDM methods have been utilized for site selection of energy projects, they have limitations when applied in reality. An important one is that they assume experts’ judgments are certain. However, decision environment often contains some uncertain information, which would affect experts’ judgments. Thus, the evaluations offered by experts are vague and imprecise. Traditional MCDM methods do not consider vague information inherent in the process of experts’ judgments. In order to cope with the imprecise information, some studies integrate fuzzy set theory into MCDM methods. Fuzzy set theory is a useful technique to manipulate the uncertain information, which permits experts combining incomplete and unquantifiable information into MCDM models [8]. Lee et al. [17] incorporate this theory into analytic network process (ANP) and VIšekriterijumsko kompromisno rangiranje (VIKOR) for location selection of PV power plants. Zoghi et al. [33] apply fuzzy logic and MCDM method to evaluate solar farms. Noorollahi et al. [20] optimize land suitability analysis of PV projects through using fuzzy AHP. Nzotcha et al. [21] propose an integrated approach, which utilizes MCDM methods and fuzzy logicbased scoring systems to select the best alternative of pumped hydro-energy storage candidate sites. Zhang et al. [32] construct an extended preference ranking organization method for enrichment evaluation (PROMETHEE) methodology to evaluate alternative sites for ocean thermal energy conversion projects under intuitionistic fuzzy environment. Combing fuzzy set theory and MCDM methods can quantify the vagueness in the experts’ judgments and enhance the subjective evaluations, meanwhile, assist managers choose suitable solar farms. However, this theory also have limitations, for example, it needs priori assumptions (i.e., membership functions and data distribution), which would lead to relatively fixed intervals to express vagueness [31]. Corresponding to fuzzy set theory, rough set theory can also manipulate imprecision. No priori information is required in rough sets, which makes the vague intervals more flexible. Thus, in this study, rough set theory will be utilized to express vagueness. Finally, most of the traditional MCDM approaches potentially assume that experts are totally rational to provide their evaluations. They are rational choice model. In reality, the personal decisions are often affected by psychological characters. Bounded rationality proposed by Simon [27] can explain it. Further, Kahneman and Tversky [16] discovery that there are deviations between the actual decision-making behaviors and the predictions generated by rational choice model. Experts are not totally rational, and their risk attitude is not fixed; it will change with the environment [19]. The decisions are not made according to the final wealth. In this respect, bounded rationality will be taken into account in this chapter to make the location selection model of solar farms more reasonable and accurate. According to the above analysis, the main purpose of this study is to develop a hybrid site selection model to promote the sustainable development of PV power projects. This developed model will integrate the rough set theory, prospect theory

Rough Set-Based Multi-Criteria Decision Analysis Methods …

223

(PT), and technique for order preference by similarity to ideal solution (TOPSIS) method, where rough sets are utilized to handle vague information, PT is adopted to manipulate bounded rationality, and TOPSIS method is utilized to prioritize alternative sites of PV power plants. Then, the proposed model is adopted to a real case to demonstrate its effectiveness and feasibility. The rest of the paper is organized as follows: Section 2 introduces the basics of rough sets, prospect theory, and TOPSIS method. Section 3 presents the developed model of site selection. In Sect. 4, a real case study is conducted, where the proposed method is adopted to obtain the ranking order of PV power stations. Finally, conclusions are remarked in Sect. 5.

2 Preliminaries 2.1 Rough Sets Rough set theory, proposed by Pawlak [22], is a useful tool to effectively deal with inaccurate information without prior assumptions. Using a pair of concepts (i.e., upper approximation and lower approximation), the inaccurate information in the judgment process could be mined and processed. Assuming that there is a judgment set S = {si |i = 1, 2,..., k } provided by k experts. The distance of the judgment set S can be defined as d = maxi si − mini si . Therefore, the lower and the upper approximations are listed as follows:    Apr(si ) = ∪ sj ∈ S sj ≤ si

(1)

   Apr(si ) = ∪ sj ∈ S sj ≥ si

(2)

where Apr(si ) and Apr(si ) represent the lower approximation and the upper approximation, respectively. Then, the limits of rough number can be obtained by siL

   sj sj ∈ Apr(si ) = p

(3)

siU

   = q sj sj ∈ Apr(si )

(4)

where siL and siU represent the limits. The superscripts L and U denote lower and upper, respectively. p and q represent the number of elements in the approximations. Thus, the vagueness of judgment si can be expressed by a rough interval RI (si ), which is presented as follows:

224

J. Li and W. Song

RI (si ) = siL , siU

(5)

The distance of the above rough boundary region is DRI (si ) = siU − siL

(6)

2.2 Prospect Theory Prospect theory (PT), proposed by Kahneman and Tversky [16], can be used to manipulate bounded rationality. In reality, individual persons are not totally rational. Their decisions are affected by bounded rationality. The basics of PT are introduced, which can describe human choice behaviors through two classic functions, namely value function and probability weighting function. A commonly used form of value functions is listed by  σ a≥0 a , f (a) = −λ( − a)ς , a < 0

(7)

where a is the amount of change between the actual value and the expected value. If a is positive, it means the risk return of the project; if it is negative, the risk loss occurs. The curves’ convexity of gain and loss is reflected by σ and ς , where both of the values are in the interval [0, 1], indicating experts’ sensitivity degree to risk. Usually, σ = ς = 0.88 [16]. λ represents the degree of loss aversion, λ = 2.25 [16]. Another function is probability weighting function. w+ (b) = bδ /(bδ + (1 − b)δ )1/δ

(8)

w− (b) = bθ /(bθ + (1 − b)θ )1/θ

(9)

where w+ (b) and w− (b) are the weighting functions when the probability is b. The superscripts + and – means the functions are for the gain and loss, respectively. δ and θ represent the experts’ attitude toward risk gain and risk loss, respectively. In this paper, the values of the two coefficients are 0.6 and 0.72, respectively.

2.3 The TOPSIS Method Hwang and Yoon [13] develop TOPSIS approach. It is one of multi-criteria decisionmaking (MCDM) methods and has been utilized by a number of scholars [7, 10, 18,

Rough Set-Based Multi-Criteria Decision Analysis Methods …

225

28, 29]. The approach can prioritize the alternatives based on the distances from each alternative to the positive ideal solution (PIS) and the negative ideal solution (NIS). The best alternative will be chosen if it is farthest from the NIS and closest to the PIS. The calculation process is presented as follows: Step 1: Construct a decision-making matrix Suppose there are m alternatives to be evaluated in relative to n criteria. Then, the decision-making matrix can be constructed by ⎡

z11 · · · ⎢ .. . . M =⎣ . . zm1 · · ·

⎤ z1n .. ⎥ . ⎦

(10)

zmn

where M is the decision matrix. Its element zij is the evaluation of the ith alternative under the jth criterion (i = 1,2,…,m; j = 1,2,…,n). Step 2: Normalize the decision-making matrix Then, the matrix M is normalized by 

zij = 

zij

m i=1

(11) (zij )2



where zij is the normalized form of zij . Step 3: Compute the weighted form of the normalized decision-making matrix Assume that wj is the weight of the jth criterion (j = 1, 2,…, n). Thus, the weighted form of the normalized evaluation can be obtained by vij = wj zij

(12)

where vij is the weighted value. Step 4: Identify the PIS and the NIS  PIS =  NIS =

vj+ = maxm i=1 (vij ), (j ∈ B) vj+ = minm i=1 (vij ), (j ∈ C)

(13)

vj− = minm i=1 (vij ), (j ∈ B) − vj = maxm i=1 (vij ), (j ∈ C)

(14)

where vj+ and vj− are the PIS and the NIS, respectively. Set B contains benefit indicators, while set C includes cost indicators. Step 4: Calculate the distances

226

J. Li and W. Song

  2  n  + 2 vij − vj+ si = 

(15)

j=1

  2  n  − 2 vij − vj− si = 

(16)

j=1

where si+ and si− are the Euclidean distances from each alternative to the PIS and the NIS. Step 5: Obtain the closeness coefficient of each alternative by cci =

si+

si− + si−

(17)

where cci is the closeness coefficient of the ith alternative. It is the basis to rank the alternatives. The bigger the value, the better the alternative.

3 The Proposed Method Due to the limited number of funds and personnel, decision makers need to evaluate potential alternatives, determine their priority, and invest the limited resources on the most suitable alternative. However, the traditional TOPSIS method often ignores the vagueness, cognitive limitations, and bounded rationality of decision-makers’ judgments under uncertain environment. Therefore, rough sets and PT theory are introduced into the conventional TOPSIS approach, which can flexibly deal with the vagueness and subjectivity of decision-makers’ judgments, and can effectively handle the bounded rationality of experts. The proposed model includes two stages, i.e., determination of criteria importance and alternatives evaluation. The calculations are presented as follows:

3.1 Determination of Criteria Importance The criteria importance is directly assessed by experts based on their professional background and work experience.

Rough Set-Based Multi-Criteria Decision Analysis Methods …

227

Step 1: Determine the original importance of the criteria. The manager invites experienced field experts to form an expert committee to evaluate the importance of indicators through a 1–10 point system. A score of 1 indicates that the criterion is the least important for alternatives selection. A score of 10 shows that the criterion is the most important for alternatives evaluation. With the increase of the score, the criteria importance becomes higher. Suppose that the importance evaluation vector of all experts for the jth criterion is ςj .   ζj = ζj1 , ζj2 , . . . , ζjk , . . . , ζjd , k = 1, 2, . . . , d ; j = 1, 2, . . . , n

(18)

where ςjk is the original evaluation of the jth criterion importance given by the kth expert. According to Eq. (18), it can be seen that there are d experts forming an expert committee. Step 2: Obtain the vague importance through rough sets. Rough set theory is utilized to deal with vagueness of experts’ judgments. Besides, experts may have varying cognitive ambiguity due to the differences in the experience and knowledge. In order to describe the ambiguity, a parameter, variable precision α, is introduced in the process of converting accurate values into rough numbers. Therefore, the rough numbers are called variable precision rough numbers. The original criteria assessment is transformed into the form of rough interval. The specific conversion process is listed as follows:         Apr α ζjk = ∪ ζjl ∈ ζj ζjl ≤ ζjk , ζjk − ζjl ≤ αg     = ∪ ζjl ∈ ζj ζjk − αg ≤ ζjl ≤ ζjk

(19)

        ζjk = ∪ ζjl ∈ ζj ζjl ≥ ζjk , ζjk − ζjl ≤ αg     = ∪ ζjl ∈ ζj ζjk ≤ ζjl ≤ ζjk + αg

(20)

Apr

α

where g = maxk ςjk − mink ςjk is the distance of vagueness of the importance evaluation vector ςj . The variable precision rough importance of the jth criterion provided by the kth expert is obtained as follows: ζjkL

=

ζjkU =

  p

    ζjl ζjl ∈ Apr α ζjk

(21)

  q

   α  ζjl ζjl ∈ Apr ζjk

(22)

where ςjkL and ςjkU are the limits of vague importance. The variable precision rough importance of ςjk is expressed as

228

J. Li and W. Song

    V PRN α ζjk = ζjkL , ζjkU (α ∈ [0, 1])

(23)

The distance of the rough boundary region is   IBRα ζjk = ζjkU − ζjkL .

(24)

where it describes the vague degree. The parameter α represents the cognitive ambiguity, which changes in the interval [0,1]. It can be divided into three situations: (1) If α = 0, there is no cognitive uncertainty. All assessments given by experts are completely deterministic. In this case, the variable precision rough number is definite and accurate. (2) If α = 1, it means that the cognition of all experts is completely uncertain. The evaluation of all experts determines the vagueness. (3) If α ∈ (0, 1), it means that there is incomplete cognitive ambiguity in expert assessment. The vague distance is determined by α × g. In this case, the number of elements in the upper and the lower approximation sets is less than that in the case of complete cognitive vagueness (α = 1). In this article, the parameter values are assigned to 0, 0.5, and 1, respectively, which are used to describe the different cognitive ambiguities. Besides, Eqs. (25)–(27) the arithmetic  present    operations of variable precision rough numbers. V PRN α ζj1 and V PRN α ζj2 are two vague evaluations of the jth criterion.       1U 2L 2U 1L 2L 1U 2U V PRN α ζ1j + V PRN α ζ2j = ζ1L j , ζj ]+[ζj , ζj ]=[ζj + ζj , ζj + ζj     1U 1L 1U t × V PRN α ζ1j = t × ζ1L (tis a nonzero constant) j , ζj ]=[tζ j , tζ j

(25) (26)

      1U 2L 2U 1L 2L 1U 2U V PRN α ζ1j × V PRN α ζ2j = ζ1L j , ζj ]×[ζj , ζj ]=[ζj × ζj , ζj × ζj (27) Step 3: Integrate the vague importance given by all experts. According to formulas (28)-(30), criteria’s vague evaluations given by all experts are aggregated to obtain the group rough importance. V PRN α (ζ j )= [ζ Lj , ζ U j ]  ζLj =

d

 ζU j =

d

 

ζkL j

(28)



ζkU j

(29)  (30)

Rough Set-Based Multi-Criteria Decision Analysis Methods …

229

where ςjL and ςjU are the lower and the upper limits of the jth criterion’s group importance.

3.2 Evaluation of Alternatives Using the Extended TOPSIS Method In this subsection, an extended TOPSIS method will be built to obtain the ranking order of the alternatives. This method integrates the advantages of variable precision rough set theory and PT, which can effectively solve the problems of ambiguity, subjectivity, and bounded rationality in expert judgments. The specific calculation process is presented as follows. Step 1: Construct a decision-making matrix. Assuming there are m alternatives, it is necessary to comprehensively evaluate their performance under n criteria. The same expert committee in the stage of determining criteria importance is also invited to evaluate the alternatives. In the same way, experts use scores, 1–10, to assess the performance of alternatives under each criterion. Therefore, the decision matrix M k provided by the kth expert is listed as ⎡ ⎢ ⎢ Mk = ⎢ ⎣

k k z11 z12 k k z21 z22 ... ... k k zm1 zm2

k ... z1n k ... z2n ... ... k ... zmn

⎤ ⎥ ⎥ ⎥ ⎦

(31)

where zijk is the kth expert’s evaluation for the ith alternative under the jth criterion (i = 1,2,…,m; j = 1,2,…,n). Similar to the calculations in Sect. 3.1, the original judgments are converted into the vague form, which are presented as follows:     V PRN α zijk = zijkL , zijkU (α ∈ [0, 1])

(32)

  where zijkL and zijkU are limits of V PRN α zijk . Then, all experts’ variable precision rough evaluations are aggregated into the group rough evaluation, which is expressed as V PRN α (z ij )= [z Lij , z U ij ]  zijL =

d

 zijU = d

 

zijkL

(33)



zijkU

(34)  (35)

230

J. Li and W. Song

where V PRN α (z ij ) is the group rough evaluation. zijL and zijU are its limits. The vague group decision-making matrix is shown as follows: ⎡ ⎢ ⎢ V PRα = ⎢ ⎣

L U [z L11 , z U 11 ] [z 12 , z 12 ] L L U [z 21 , z 21 ] [z 22 , z U 22 ] ... ... L U [z Lm1 , z U m1 ] [z m2 , z m2 ]

... [z L1n , z U 1n ] L ... [z 2n , z U 2n ] ... ... ... [z Lmn , z U mn ]

⎤ ⎥ ⎥ ⎥ ⎦

(36)

Step 2: Normalize the decision-making matrix. A normalization method is used to make the criteria of different units comparable, which is shown as follows: 



L U U U m L U zijL = zijL /maxm i=1 {max[z ij , z ij ]}, zij = zij /maxi=1 {max[z ij , z ij ]}

(37)

    where zijL , zijU is the normalized interval. Step 3: Determine the ideal solutions by  PIS = 





L m U zj+ = [maxm i=1 (z ij ), maxi=1 (z ij )], (j ∈ B)  L m U zj+ = [minm i=1 (z ij ), mini=1 (z ij )], (j ∈ C) 

(38)



L m U z − = [minm i=1 (z ij ), mini=1 (z ij )], (j ∈ B) NIS = −j  L m m zj = [maxi=1 (z ij ), maxi=1 (z U ij )], (j ∈ C)

(39)

where zj+ is the PIS and zj− represents the NIS. Sets B and C contain benefit indicators and cost indicators, respectively. The benefit indicator means the larger the better, while the cost indicator means the smaller the better. Then, using the Euclidean distance operator [12], the distances between each candidate PV power plant location and the ideal solutions are calculated through the following formulas: ⎧ ⎨ 2 (z L −z U + )2 +(z U −z L+ )2 , (j ∈ B) ij ij j j + sij =  + 2 L U L U 2 ⎩ (z −z ) +(z −z + )2 , (j ∈ C) ij ij j j ⎧ ⎨ 2 (z L −z U − )2 +(z U −z L− )2 , (j ∈ B) ij ij j j − sij =  ⎩ 2 (z L −z L− )2 +(z U −z U − )2 , (j ∈ C) ij ij j j +

+

−

−

(40)

(41)

where zjL and zjU are the limits of zj+ ; zjL and zjU are the limits of zj− . sij+ and sij− are the distances from the ith location to ideal locations under the jth indicator. Step 4: Compute the closeness coefficient improved by PT. When the distance between each location and ideal solutions is obtained, the closeness coefficient can be computed to rank the candidate locations in the TOPSIS approach. However, bounded rationality is not considered in the method. In reality,

Rough Set-Based Multi-Criteria Decision Analysis Methods …

231

human decision-making behavior is not totally rational. In this step, PT is used to modify the closeness coefficient. Based on the basics of PT, the improved calculation process of the closeness coefficient is computed as follows: In order to obtain the weighting function, the criteria’s variable precision rough importance is first transformed into a definite value according to the following equations. (1)

(2)

Normalize     ζ˜jL = ζjL − minj ζjL / maxj ζjU − minj ζjL

(42)

    ζ˜jU = ζjU − minj ζjL / maxj ζjU − minj ζjL

(43)

Define a parameter       τj = ζ˜jL × 1 − ζ˜jL + ζ˜jU × ζ˜jU / 1 − ζ˜jL + ζ˜jU

(3)

(44)

Compute the definite form of rough importance.   L ζ˜j = minj ζjL + τj × maxj ζU − min ζ j j j

(45)

where ςj is the definite form of vague importance. The normalized definite importance is calculated by ζ˜j = ζ˜j /maxj ζ˜j

(46)

    The weighting functions of ςj , i.e., w+ ςj and w− ςj are computed by w+ (ζ˜j ) = ζ˜jδ /(ζ˜jδ + (1 − ζ˜j )δ )1/δ

(47)

 w− (ζ˜j ) = ζ˜jθ /(ζ˜jθ + (1 − ζ˜j )θ )1/θ

(48)

If the positive ideal solution is identified as the reference point, each alternative will have a loss (i.e., −sij+ ) relative to the PIS for experts. If the negative ideal solution is considered as the reference point, then each alternative will generate benefit (i.e., sij− ) relative to the NIS. Hence, the values of weighted value function for the risk profits and risk losses are calculated according to the following formulas. vi+ =

n  j=1

 w+ (ζ˜j )(sij− )σ

(49)

232

J. Li and W. Song

vi− = −λ

n 

 w− (ζ˜j )[ − ( − sij+ )ς

(50)

j=1

where vi+ and vi− are the values of weighted value function for the risk profits and risk losses, respectively. The closeness coefficient cci taking bounded rationality into account is obtained by v+ cci =  −  i + v  + v i

(51)

i

The greater the closeness coefficient, the more suitable the ith alternative. Apparently, we can select the optimal alternative based on the closeness coefficients.

4 Case Study 4.1 Sustainable Evaluation of PV Projects In this section, the proposed method is utilized by a power generation company for site selection of PV power projects in China. This firm is selecting the best construction site for a PV power project. Currently, four alternative sites have been initially identified, namely site A, site B, site C, and site D. The development of economy, energy planning, and climate in the four regions are not exactly the same. Specifically, site C and site D are rich in solar resources and are residential areas of ethnic minority in China. Therefore, the construction of PV power plants in these areas must consider social risks, such as local public support. In addition, we invite a number of experts to form an expert committee to evaluate the candidate sites of PV power station. Two managers, an investor, and a related field scholar are included in the committee. They all have at least ten years work experience in the PV field. The experts undertake two duties, i.e., determining the criteria importance and evaluating the performance of candidate locations under criteria. Before officially starting the evaluation work, the author organizes experts to discuss and understand various criteria in depth. They believe that social criteria are very important for evaluating the location of PV power plants. The evaluation criteria have been identified in Sect. 1. Then, the questionnaire is designed and sent to the experts. They provide their original evaluations according to the alternative site information. Utilizing the raw data, the closeness coefficients are obtained and the PV power sites are prioritized through applying the proposed approach, which can be seen in Table 1.

Rough Set-Based Multi-Criteria Decision Analysis Methods …

233

Table 1 Ranking order of the alternative sites [9] Alternative sites

The proposed method (i.e., Rough_PT Rough TOPSIS) TOPSIS

Fuzzy_PT TOPSIS

Traditional TOPSIS

α=1

α = 0.5

α=0

Site A

2

3

2

1

2

3

Site B

4

4

4

4

4

2

Site C

3

2

3

3

3

1

Site D

1

1

1

2

1

4

4.2 Results and Discussions First of all, compared with the previous location selection models of PV power plants, the proposed method considers the local social indicators, which are often ignored in the previous research. In this respect, this chapter constructs a comprehensive criteria evaluation system, including resource criteria, economic criteria, environmental criteria, and social criteria to ensure that the proposed model can comprehensively examine all aspects of the candidate sites of PV power plants and provide useful information for managers. In addition, to verify the effectiveness of the proposed approach (i.e., rough_PT TOPSIS), it is compared with other site selection methods. The traditional TOPSIS method, TOPSIS approach based on fuzzy sets and prospect theory (i.e., Fuzzy_PT TOPSIS), and TOPSIS method based on rough sets (i.e., rough TOPSIS) are also employed to the same case to obtain the site ranking of PV power plants. To facilitate comparison and discussion, the values of variable precision α, 0, 0.5, and 1 are adopted. The ranking results of PV power plants produced by different methods are shown in Fig. 1. In the following, the proposed method is compared with other methods to verify its advancement. First, the proposed site selection approach is compared with the traditional TOPSIS methodology. Rough_PT TOPSIS (α = 1) is selected as the proposed approach. The rankings of the alternative sites generated by the two methodologies are different (see Fig. 1). Site D has the rank of 1 in the proposed method and ranks the fourth in the other method. Site C ranks the third in the former approach and the first in the latter. The main reason for the different site rankings is crisp numbers are used in the conventional TOPSIS method for evaluating the criteria importance and locations performance. It does not consider the subjectivity and ambiguity of the experts, while vague information expressed by rough sets is manipulated in the proposed approach. Second, the proposed location selection model is compared with the Fuzzy_PT TOPSIS method. It can be seen from Fig. 1 that when the variable precision α is 1 and 0, the location ranking generated by the Rough_PT TOPSIS method is the same as the result of the Fuzzy_PT TOPSIS method. However, when α = 0.5, the two methods produce different site ranks of PV power projects. For instance, the rank of site A is three in the developed method and two in the other approach. An important

234

J. Li and W. Song

5 4

Rank

3 2 1 0 Site A

Site B

Site C

Site D

Rough_PT TOPSIS α=1

Rough_PT TOPSIS α=0.5

Rough_PT TOPSIS α=0

Rough TOPSIS

Fuzzy_PT TOPSIS

Traditional TOPSIS

Fig. 1 Ranking order of PV power sites in different methods [9]

reason for the ranking differences is that ways of handling ambiguity are different between rough sets and fuzzy sets in the decision-making process. Rough sets are more flexibly and accurately in dealing with the subjectivity of experts without prior assumptions, while fuzzy sets require a predefined membership function, which leads to a fixed interval. Figure 2 presents the experts’ evaluations in the form of rough numbers and fuzzy numbers. The original scores provided by the four experts are 3, 5, 8, and 10, respectively. Obviously, the intervals produced by different mechanisms are different. The boundary intervals of variable precision rough evaluations of the four experts are various, while those of fuzzy evaluations are all the same, fixed at 2.

Expert 1 Rough number (α=0)

Expert 2 Rough number (α=0.5)

Fig. 2 Experts’ evaluations in different types [9]

Expert 3 Rough number (α=1)

Expert 4 Fuzzy number

Rough Set-Based Multi-Criteria Decision Analysis Methods …

235

Besides, another critical reason is that the proposed rough method takes the cognitive ambiguity of experts into account, which is represented by variable precision α. It makes the site ranking results more general. The greater the value of α, the greater the boundary distance of rough intervals. According to Fig. 2, the rough intervals get greater with the increase of α. Third, the proposed location selection method of PV power stations is compared with the rough TOPSIS method. In this comparison, Rough_PT TOPSIS (α = 1) is chosen as the proposed approach. It can be seen in Fig. 1 that the compared techniques generate the identical ranks for site B and site C, which are 4 and 3, respectively. The rankings of site A and site D are different. Specifically, site A ranks the second in the developed approach, while the first in the other method. The differences produced by the two methods are mainly because bounded rationality of experts is considered through PT in the proposed approach. The priority of alternative sites is assessed through computing the risk gain and risk loss relative to the PIS and NIS. Meanwhile, risk attitudes of experts toward risk gains and risk losses are also not fixed. By contrast, the other TOPSIS method assumes that experts are completely rational, and the sites of PV power stations are ranked by the distances relative to reference points without considering risk attitude. Through Fig. 3, the basis of calculating final results between the compared approaches is shown. In the rough TOPSIS method, the distances from the performance scores of PV power plant locations to the PIS and the NIS are the basis to calculate final results. By contrast, risk losses and risk gains between each alternative site and the PIS and NIS are regarded as basis to obtain Prospect function value

Loss

Gain

Value of loss and gain

Fig. 3 Prospect function values of risk gains and risk losses of alternative sites [9]

236

J. Li and W. Song

closeness coefficients in the proposed approach. Specifically, risk losses will occur in all locations for experts when the positive ideal solution is selected as the reference point; risk gains will occur in all PV power sites for experts when the negative ideal solution is considered as the reference point. Finally, it can be seen that the curve slope in the red area is greater than that in the green area, which reveals people are more sensitive to risk losses. To sum up, the advantages of the proposed site selection model of PV power plants are described as follows: (1)

(2)

(3)

This method uses variable precision rough set theory to deal with ambiguity during determining the criteria weights and evaluating PV power sites. It does not require much prior assumptions, while fuzzy-based TOPSIS method needs to preset the membership function, which adds the method’s subjectivity. A variable precision parameter is defined in this study to flexibly deal with the cognitive ambiguity of expert judgment. Both the evaluation set and cognitive vagueness determine the vague intervals. Thus, the variable precision rough numbers can mine the uncertainty of experts more flexibly, which provides managers a more reasonable site priority of PV power plants. PT is introduced into the proposed model to deal with the bounded rationality of experts. In practice, the rationality of decision makers is usually limited, and their decision-making behaviors often deviate from the predictions. Psychological factors of the experts and their risk attitudes are considered in the proposed approach.

5 Conclusions In this chapter, an integrated site selection framework for PV power plants is constructed. First, a criteria system is established to comprehensively assess the all aspects of the candidate locations to promote the sustainability of PV projects. At the same time, an extended TOPSIS methodology is built for ranking the alternative sites. It combines the advantages of the traditional TOPSIS method, rough numbers, and PT, which could manipulate vague judgments and risk attitude. Then, the developed method is applied to a real case, where a 10 MW PV power station needs to select site in China. The main contributions of this site selection model are presented as follows: (1) Social criteria, together with solar resource, economy, and environment, are included in the criteria evaluation system to ensure sustainable development of projects. (2) Variable precision rough numbers are introduced to accurately and flexibly express the imprecise information inherent in experts’ decision making. (3) Prospect theory is adopted in the approach to effectively manipulate bounded rationality and dynamic risk attitude of experts, which is closer to the actual situation.

Rough Set-Based Multi-Criteria Decision Analysis Methods …

237

References 1. Aydin NY, Kentel E, Duzgun HS (2013) GIS-based site selection methodology for hybrid renewable energy systems: a case study from western Turkey. Energy Convers Manage 70:90– 106 2. Chang CT (2015) Multi-choice goal programming model for the optimal location of renewable energy facilities. Renew Sustain Energy Rev 41:379–389 3. Carrión JA, Estrella AE, Dols FA, Toro MZ, Rodríguez M, Ridao AR (2008) Environmental decision-support systems for evaluating the carrying capacity of land areas: optimal site selection for grid-connected photovoltaic power plants. Renew Sustain Energy Rev 12(9):2358–2380 4. Carrera DG, Mack A (2010) Sustainability assessment of energy technologies via social indicators: results of a survey among European energy experts. Energy Policy 38(2):1030–1039 5. Charabi Y, Gastli A (2011) PV site suitability analysis using GIS-based spatial fuzzy multicriteria evaluation. Renew Energy 36(9):2554–2561 6. China Electric Power Yearbook, 2007–2020. 7. Choudhary D, Shankar R (2012) An STEEP-fuzzy AHP-TOPSIS framework for evaluation and selection of thermal power plant location: a case study from India. Energy 42(1):510–521 8. Da˘gdeviren M, Yavuz S, Kılınç N (2009) Weapon selection using the AHP and TOPSIS methods under fuzzy environment. Expert Syst Appl 36(4):8143–8151 9. Fang H, Li J, Song W (2018) Sustainable site selection for photovoltaic power plant: an integrated approach based on prospect theory. Energy Convers Manage 174:755–768 10. Fang H, Li J, Song W (2020) Failure mode and effects analysis: an integrated approach based on rough set theory and prospect theory. Soft Comput 24(9):6673–6685 11. Gamboa G, Munda G (2007) The problem of windfarm location: a social multi-criteria evaluation framework. Energy Policy 35(3):1564–1583 12. Hennig C, Meila M, Murtagh F, Rocci R (eds) (2015) Handbook of cluster analysis. CRC Press 13. Hwang CL, Yoon K (1981) Multiple attribute decision making: methods and applications. Springer, New York 14. Janke JR (2010) Multicriteria GIS modeling of wind and solar farms in Colorado. Renew Energy 35(10):2228–2234 15. Jun D, Tian-tian F, Yi-sheng Y, Yu M (2014) Macro-site selection of wind/solar hybrid power station based on ELECTRE-II. Renew Sustain Energy Rev 35:194–204 16. Kahneman D, Tversky A (1979) Prospect theory: an analysis of decision under risk. Econometrica: J Econometric Soc, 263–291 17. Lee AH, Kang HY, Liou YJ (2017) A hybrid multiple-criteria decision-making approach for photovoltaic solar plant location selection. Sustainability 9(2):184 18. Li J, Fang H, Song W (2019) Sustainable supplier selection based on SSCM practices: A rough cloud TOPSIS approach. J Clean Prod 222:606–621 19. Liu P, Jin F, Zhang X, Su Y, Wang M (2011) Research on the multi-attribute decision-making under risk with interval probability based on prospect theory and the uncertain linguistic variables. Knowl-Based Syst 24(4):554–561 20. Noorollahi E, Fadai D, Akbarpour Shirazi M et al (2016) Land suitability analysis for solar farms exploitation using GIS and fuzzy analytic hierarchy process (FAHP)—a case study of Iran[J]. Energies 9(8):643 21. Nzotcha U, Kenfack J, Manjia MB (2019) Integrated multi-criteria decision making methodology for pumped hydro-energy storage plant site selection from a sustainable development perspective with an application. Renew Sustain Energy Rev 112:930–947 22. Pawlak Z (1982) Rough sets. Int J Comput Inform Sci 11(5):341–356 23. Tahri M, Hakdaoui M, Maanan M (2015) The evaluation of solar farm locations applying geographic information system and multi-criteria decision-making methods: case study in southern Morocco. Renew Sustain Energy Rev 51:1354–1362 24. Wu Y, Geng S, Zhang H, Gao M (2014) Decision framework of solar thermal power plant site selection based on linguistic Choquet operator. Appl Energy 136:303–311

238

J. Li and W. Song

25. Sánchez-Lozano JM, García-Cascales MS, Lamata MT (2015) Evaluation of suitable locations for the installation of solar thermoelectric power plants. Comput Ind Eng 87:343–355 26. Sánchez-Lozano JM, Teruel-Solano J, Soto-Elvira PL, García-Cascales MS (2013) Geographical Information Systems (GIS) and Multi-Criteria Decision Making (MCDM) methods for the evaluation of solar farms locations: case study in south-eastern Spain. Renew Sustain Energy Rev 24:544–556 27. Simon HA (1971) Administrative behavior-a study of decision making processes in administrative organization. Macmillan Publishing Co, lnc, New York 28. Song W, Ming X, Wu Z, Zhu B (2013) Failure modes and effects analysis using integrated weight-based fuzzy TOPSIS. Int J Comput Integr Manuf 26(12):1172–1186 29. Song W, Ming X, Wu Z, Zhu B (2014) A rough TOPSIS approach for failure mode and effects analysis in uncertain environments. Qual Reliab Eng Int 30(4):473–486 30. Uyan M (2013) GIS-based solar farms site selection using analytic hierarchy process (AHP) in Karapinar region, Konya/Turkey. Renew Sustain Energy Rev 28:11–17 31. Zhai LY, Khoo LP, Zhong ZW (2008) A rough set enhanced fuzzy approach to quality function deployment. Int J Advan Manuf Technol 37(5–6):613–624 32. Zhang J, Xu C, Song Z, Huang Y, Wu Y (2019) Decision framework for ocean thermal energy plant site selection from a sustainability perspective: the case of China. J Clean Prod 225:771– 784 33. Zoghi M, Ehsani AH, Sadat M, Javad Amiri M, Karimi S (2017) Optimization solar site selection by fuzzy logic model and weighted linear combination method in arid and semi-arid region: a case study Isfahan-IRAN. Renew Sustain Energy Rev 68:986–996

Multicriteria-Oriented Optimization of Building Energy Performances: The Annex 72 IEA-EBC Experience Francesco Montana, Sonia Longo, Harpa Birgisdottir, Maurizio Cellura, Rolf Frischknecht, Francesco Guarino, Benedek Kiss, Bruno Peuportier, Thomas Recht, Eleonora Riva Sanseverino, and Zsuzsa Szalay Abstract This chapter describes the research experience of the International Energy Agency - Energy in Buildings and Communities Programme Annex 72 members on the application of multi-objective optimization processes for the selection of design or retrofit actions that allow for improving different aspects (energy, environmental, economic, etc.) of buildings in a life cycle perspective. Thirteen case studies were examined focussing on methodologies, applications and results and deriving generic conclusions and guidelines for building designers and decision-makers. Keywords Optimization · Life cycle · Buildings · Energy performance · Environmental impacts

1 Introduction The building sector is one of the most impacting on the energy demand and on the environment in developed countries, together with industry and transports [1]. This is well represented in Fig. 1, where the trend of primary energy consumption in final F. Montana (B) · S. Longo (B) · M. Cellura · F. Guarino · E. Riva Sanseverino Department of Engineering, University of Palermo, Palermo, Italy e-mail: [email protected] S. Longo e-mail: [email protected] H. Birgisdottir Danish Building Research Institute, Aalborg University Copenhagen, Copenhagen, Denmark R. Frischknecht Treeze Ltd., Uster, Switzerland B. Kiss · Z. Szalay Department of Construction Materials and Technologies, Budapest University of Technology and Economics, Budapest, Hungary B. Peuportier · T. Recht Center for Energy Efficiency of Systems, MINES ParisTech, PSL Research University, Paris, France © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 J. Ren (ed.), Energy Systems Evaluation (Volume 2), Green Energy and Technology, https://doi.org/10.1007/978-3-030-67376-5_10

239

240

M. Francesco et al.

Fig. 1 Primary energy consumption trend in final uses in EU [1]

uses in European Union (EU) is shown, with the building sector being included in households and commercial categories. Further detail is provided in Fig. 2, where the share in 2017 is reported. 35% 30% 25% 20% 15% 10% 5% 0% Households

Commercial & public services

Industry

Transports

Fig. 2 Primary energy consumption in final uses in EU in 2017 [1]

Other Sectors

Multicriteria-Oriented Optimization of Building Energy …

241

A big share of this consumption can be ascribed to thermal uses, mainly due to poor buildings envelope and inefficiencies in heating, ventilation and air-conditioning (HVAC) plants [2]. With the aim of reducing the energy consumption of the building sector, EU emitted a series of energy performance of building directives, introducing the topic of nearly zero-energy building (nZEB) [3] and promoting deep renovations in the existing building stock [4]. The design process of a nZEB, and in general of a low-energy building, is generally composed by the following steps: • Building thermal loads reduction through interventions on the envelope transmittance; • Passive strategies employment, attributing part of the thermal loads removal to natural phenomena; • Installation of efficient HVAC systems, as low temperature heating or high temperature cooling systems; • Integration of renewable energy system technologies to cover as much as possible the residual loads. Furthermore, according to EU directives, nZEBs should also be cost-efficient as well as energy-efficient, thus requiring to compare a set of interventions on the building to identify the optimal combination of strategies to implement. The economic point of view is also a key issue for the investors, usually aiming at reaching their result with the minimum disbursement. Furthermore, the building inhabitants would prefer to enjoy a comfortable dwelling, aspect that in some cases can be hardly ensured together with the energy efficiency and the cost-optimality. For what above, the adoption of a multicriteria approach is often required in the low-energy buildings sector. In detail, one of the most suitable approaches is to integrate the preliminary building design (or refurbishment) phase in an optimization problem, allowing to rapidly compare many alternatives and to identify the most adapt interventions. In order to take into account for the different points of view of policy makers, investors and inhabitants, a multi-objective approach is also recommended. A further step forward may also be done considering that the operating energy reduction risks to be accompanied with a high rising in the energy used to build the envelope and equipment components (embodied energy), e.g. synthetic insulation materials or building automation systems, and also in the embodied environmental impacts as greenhouse gas emissions. In order to take into account this additional aspect, the multi-objective optimization of buildings needs a holistic approach integrating the energy analysis, the economic evaluation, the life cycle thinking, i.e. the application of the life cycle assessment (LCA) methodology [5–7], and other types of analyses. This chapter illustrates and reviews the contribution of the International Energy Agency - Energy in Buildings and Communities (IEA-EBC) Programme Annex 72 members to the life cycle multi-objective optimization of buildings performance, comparing methodologies, applications and results and deriving generic conclusions and guidelines from a collection of case studies.

242

M. Francesco et al.

2 The IEA-EBC Annex 72 The IEA-EBC Annex 72 focuses on the assessment of the primary energy demand, greenhouse gas emissions and environmental impacts of buildings during production, construction, use (including repair and replacement) and end of life (dismantling), i.e. during the entire life cycle of buildings [8]. Annex 72 comprises five main Subtasks (1—Harmonized methodology guidelines, 2—Building assessment tools, 3—Case studies, 4—Building sector LCA databases, 5—Dissemination) and aims at establishing a common methodology guideline to assess the life cycle energy and environmental impacts caused by buildings, at establishing methods for the development of specific environmental benchmarks for different types of buildings, at deriving guidelines and tools for building design and planning for the stakeholders (architects, planners, researchers, etc.), at developing or examining case studies focussed to some research issues and for deriving empirical benchmarks, and at developing national or regional databases with life cycle assessment data for the construction sector. Within the Subtask 3, Activity 3.3 deals with the following problem: a specific design or retrofit action can allow for reducing the energy impact and some environmental or economic impacts of the building, but it can cause the increase of other environmental impacts as well (e.g. the use of electricity from PV instead of electricity from a country-specific grid allows for reducing the impact on global warming potential but causes an increase of the impact on land use). The research issue is to avoid the shift of impacts from one impact category to another and to identify, among a group of solutions (e.g. different thicknesses of insulation or different types of insulation), the best ones (e.g. the best thickness of insulation or the best type of insulation) that allow for obtaining an “optimum” among the examined energy and environmental indexes, also taking into account other aspects such as the economic one. The optimization techniques can be applied to solve the above issue. In this context, the goal of Activity 3.3 is to examine selected case studies in order to classify and characterize different approaches aimed at the optimization of the life cycle energy, environmental, economic and other performance of new buildings and renovation projects and to assess the potential of optimization strategies in order to provide some guidelines for building design and decision-making.

3 Review of the Annex 72 Case Studies 3.1 Introduction Within the context of the Annex 72, a set of thirteen case studies related to the optimization of life cycle performance of buildings was collected and reviewed.

Multicriteria-Oriented Optimization of Building Energy …

243

The approaches and the outcomes of these studies were analysed and compared in this chapter, in order to identify a generic framework and to draw guidelines for the scientific community. In this section, a brief review of the analysed case studies is provided, reporting some details on each aspect of a typical optimization study.

3.2 Optimization Software, Approaches and Algorithms The studies reviewed in this work can be categorized according to different aspects. Regarding the aim of the study, most of these thirteen works analysed the refurbishment of existing buildings [9–15], while three of them focussed on the early design stage of residential buildings [16, 17], and one just examined the optimal concrete quantity and quality, disregarding the operating phase and the energy demand for air conditioning [18]. Only one of the examined buildings was optimized to reach the plus-energy level [16], while in most of the other studies the installation of renewables and the annual energy balance were not taken into account. Considering the developers of the studies, six of them were performed by the research group of the Bauhaus-University Weimar [9–12, 18], four studies by the group from the University of Palermo [13–15], two studies came from the Budapest University of Technology and Economics [17] and the last work was developed by the MINES ParisTech researchers [16]. Most of the studies [9–12, 17, 18] were developed in Rhinoceros CAD environment [19], since this tool has a user-friendly framework and allows for integrating many aspects of the building design in a unique tool through many plug-ins available on its library. Two studies [13, 14] were modelled on SketchUp 3D CAD [20], using other tools for the optimization; one did not require a CAD modelling phase [15] and one was entirely developed in Pleiades [21], a French tool able to manage the 3D modelling, the energy simulation, the optimization process and the life cycle assessment [16]. In the reviewed studies, the use phase energy demand of the building was assessed mainly through a dynamic building performance simulation (BPS) software, with EnergyPlus [22] being the most popular [9–11, 13, 14], since it allows for the connection with both Rhinoceros and SketchUp, while four studies [12, 15, 17] adopted energy calculations based on the quasi-stationary seasonal method described on the European standard EN ISO 13790 [23] or on the German standard DIN V 18599 [24]. Many building performance optimization (BPO) tools and algorithms were employed in these studies: MOBO Multi-Objective Building Optimization tool [25, 26] was chosen in three studies [13–15], three Rhinoceros plug-ins for the optimization, namely Galapagos [9, 10, 18], Octopus [11, 17] and GOAT [11, 12], and MATLAB programming environment [14] were used in the other studies. The first outcome of this review is the great variability in the adoption of software tools, since the researchers were often forced to combine many software and plug-ins

244

M. Francesco et al.

to perform their studies, with Pleiades being the unique commercial tool for the integration of all these aspects in the same platform. The same outcome was previously observed by Gilles et al. in 2017, discussing on the difficulties in the interactions between building performance simulators and life cycle assessment studies [27]. Analysing the optimization approach, six studies adopted a single-objective optimization approach [9–12, 17, 18], while the remaining seven employed multiobjective algorithms [11, 13–17]. Nevertheless, some single-objective optimization studies also assessed additional life cycle impact indicators, although they were not optimized [9, 10, 18]. Moreover, some studies also evaluated the thicknesses of insulation materials minimizing many life cycle impact indicators, comparing the optimal retrofit actions [12, 17]. Almost all the studies adopted heuristic algorithms, mainly genetics, with the only exception of two case studies that adopted a dual step approach and employed the branch-and-bound algorithm in the second step [14]. This is a common aspect, since most of the optimization problems in engineering are highly nonlinear, preventing the adoption of exact deterministic algorithms based on derivatives evaluation, and heuristic and meta-heuristic algorithms are usually preferred. Moreover, the evaluation of one or more objective functions through a simulation-based optimization almost obliges the adoption of a heuristic approach. Some studies involved the single-objective genetic algorithm available on Galapagos tool [9, 10, 18] and the optimization plug-in of Rhinoceros Grasshopper modelling environment, while other researchers preferred the CRS2 [11, 12] or HypE [17] algorithms, also working on Rhinoceros. Unfortunately, according to a review performed by Wortmann [28] on the algorithms available on Rhinoceros, the genetic algorithm of Galapagos plug-in is considered the worst performing one and even the CRS2 algorithm was disregarded in this study, while the author recommends RBFOpt algorithm available on the Opossum plug-in. The multi-objective algorithms employed were NSGA-II [13, 16], since it is one of the most popular in scientific literature [29], and also HypE [17] and Omni-Optimizer [14, 15]. A recap of the main information provided so far is given in Table 1, while a visual representation of the studies developed is shown in Fig. 3, with the pins indicating how many studies were developed in each location and the logos indicating the adopted software tools. The importance of selecting the most adapted algorithm was briefly discussed in [15], where the four multi-objective algorithms available in MOBO were compared, also changing the algorithm parameters, in order to identify the most suitable one. Only NSGA-II, aNSGA-II and Omni-Optimizer were illustrated, since the Random Search algorithm did not provide any feasible solution. The performance of these algorithms can be compared through the visualization of the resulting bi-dimensional Pareto Fronts, shown in Figs. 4 and 5, and through the number of the resulting feasible solutions and optimal compromise solutions (Pareto front), as reported in Table 2. It is evident that, although 200 generations were set, the aNSGA-II algorithm was not able to reach the Pareto front, both with 16 and 40 individuals. Omni-Optimizer instead identified good solutions with 16 individuals and 500 generations, but the best

Developed by

MINES ParisTech

Bauhaus-University Weimar

Bauhaus-University Weimar

Bauhaus-University Weimar

Bauhaus-University Weimar/Fraunhofer Institute

Bauhaus-University Weimar/Fraunhofer Institute

Bauhaus-University Weimar

Budapest University of Technology and Economics

Aim

Design of a plus-energy house

Refurbishment

Refurbishment

Design of a garage

Refurbishment

Refurbishment

Refurbishment

Design

Grasshopper. Quasi-steady state approach based on ISO 13790

Grasshopper. Quasi-steady state approach based on DIN V 18599

EnergyPlus

EnergyPlus

Non necessary

EnergyPlus

EnergyPlus

Pleiades (COMFIE module)

BPS

Table 1 Recap of the main features of the reviewed studies

Octopus plug-in for Grasshopper (Rhinoceros)

GOAT plug-in for Grasshopper (Rhinoceros)

Octopus plug-in for Grasshopper (Rhinoceros)

GOAT plug-in for Grasshopper (Rhinoceros)

Galapagos plug-in for Grasshopper (Rhinoceros)

Galapagos plug-in for Grasshopper (Rhinoceros)

Galapagos plug-in for Grasshopper (Rhinoceros)

AMAPOLA (developed in Python)

BPO

Single-objective

Single-objective

Multi-objective

Single-objective

Single-objective

Single-objective

Single-objective

Multi-objective

Approach

[18]

[9]

[10]

[16]

Reference

[11]

HypE genetic algorithm

(continued)

[17]

CRS2 evolutionary [12] algorithm

Genetic algorithm

CRS2 evolutionary [11] algorithm

Evolutionary algorithm

Evolutionary algorithm

Evolutionary algorithm

NSGA-II genetic algorithm

Algorithm

Multicriteria-Oriented Optimization of Building Energy … 245

Developed by

Budapest University of Technology and Economics

University of Palermo

University of Palermo

University of Palermo/Aalborg University

University of Palermo

Aim

Design

Refurbishment

Refurbishment

Refurbishment

Refurbishment

Table 1 (continued)

MOBO; MATLAB

MOBO; MATLAB

Octopus plug-in for Grasshopper (Rhinoceros)

BPO

EnergyPlus

MOBO; MATLAB

Be18. Quasi-steady state MOBO approach based on ISO 13790

EnergyPlus

EnergyPlus

Grasshopper. Quasi-steady state approach based on ISO 13790

BPS

Multi-objective

Multi-objective

Multi-objective

Multi-objective

Multi-objective

Approach

Omni-Optimizer genetic algorithm; branch-and-bound MILP algorithm

Omni-Optimizer genetic algorithm

NSGA-II genetic algorithm; branch-and-bound MILP algorithm

NSGA-II genetic algorithm; branch-and-bound MILP algorithm

HypE genetic algorithm

Algorithm

[14]

[15]

[13, 14]

[13, 14]

[17]

Reference

246 M. Francesco et al.

Multicriteria-Oriented Optimization of Building Energy …

247

MOBO©

MOBO© Fig. 3 Visual representation of the studies reviewed in this work

Embodied Global Warming Potential [kg CO2_eq]

400000

aNSGA II 16-200 aNSGA II 40-200 NSGA II 40-200 Omni-Optimizer 40-200 Omni-Optimizer 16-500

350000 300000 250000 200000 150000 100000 50000 0 2500

3500

4500

5500

6500

7500

Use Phase Final Energy Demand [MWh] Fig. 4 Comparison of bi-dimensional Pareto fronts from 5 multi-objective algorithms used in [15] relating use phase final energy demand against embodied GWP

248

M. Francesco et al.

Investment Cost [DKK]

7000000

aNSGA II 16-200 aNSGA II 40-200 NSGA II 40-200 Omni-Optimizer 40-200 Omni-Optimizer 16-500

6000000

5000000

4000000

3000000

2000000 2500

3500

4500

5500

6500

7500

Use Phase Final Energy Demand [MWh] Fig. 5 Comparison of bi-dimensional Pareto fronts from 5 multi-objective algorithms used in [15] relating use phase final energy demand against investment cost

Table 2 Comparison of 5 multi-objective algorithms used in [15] according to the number of alternatives set for the optimization and the resulting number of feasible and optimal solutions Alternatives

Feasible solutions (with duplicates)

Feasible solutions (no duplicates)

Solutions in the Pareto front

aNSGA-II 16-200 3200

1492

768

252

aNSGA-II 40-200 8000

3325

2267

379

NSGA-II 40-200

8000

4619

2434

381

Omni-Optimizer 40-200

8000

4402

2986

491

Omni-Optimizer 16-500

8000

4948

2150

267

results were obtained in the optimization with 40 individuals and 200 generations. The NSGA-II algorithm solutions were sparser than the others. A study on the parameter tuning in NSGA-II algorithm was conducted in [30], based on the hypervolume indicator (also called metric) which is a measure of both intensification and diversification of the optimization. On the basis of a reference variant, best compromises two pairs of parameters were explored: population size (P) and number of evaluations, and crossover (C) and mutations (M) rates (in %). For an equivalent number of evaluations after initialization, a large population size (P2000, M15 and C80 in Fig. 6) reduces the number of generations, degrading the research intensification. On the contrary, a small population size (P20, M15 and C80) speeds up the intensification at the beginning of optimization but quickly reaches its

Multicriteria-Oriented Optimization of Building Energy …

249

Fig. 6 Hypervolume evolution versus the number of model evaluations (after initialization) with 5 parameter settings [30]

limit by lack of diversification. Compared to a quasi-random search reproduced with a mutation rate of 100% and a crossover rate of 0% (P200, M100 and C0), the optimization by artificial evolution through NSGA-II with a standard tuning (P200, M15 and C80) is much more efficient. In this case study, the diversity produced by the crossover operator is sufficient to effectively conduct the research, leading to setting the mutation rate at 0% (P200, M0 and C100).

3.3 Objective Functions and Variables 3.3.1

Objective Functions

The objective functions minimized in the reviewed studies may be grouped in three categories: use phase performance, LCA embodied impacts and economic aspects. Almost all of the studies evaluated the use phase performance, since it is the most widely adopted criterion for the assessment of the energy efficiency. All the studies optimized the use phase of the building employing the global warming potential (GWP) as objective function, suggesting that this indicator appears as more important than the use phase energy demand, even if usually the GWP and energy consumption are dependent each other. Some studies, both single and multi-objective, also assessed other LCA indicators. In detail, although some works [12, 17] described singleobjective optimizations, many indicators were minimized separately in order to show the dependency of the optimal interventions on the objective function. Other studies, based on a two-step approach, optimized the use phase final energy demand in a first run where only the building envelope components were employed as variables, while

250

M. Francesco et al.

in the second step this value was used to identify the optimal equipment minimizing the use phase primary energy, GWP and operating costs. The unique work neglecting the use phase is the study by Hollberg et al. [18], where a case study on the design of a bike garage was considered; in this case only the embodied impacts were assessed. The LCA impact assessment indicators employed in the reviewed studies are among the most commonly used in LCA studies on buildings [31, 32]. Apart from the GWP, the energy demand of the building was assessed through the cumulative energy demand (CED) [13–15, 17], or, as an alternative, the indicators from the CML 2001 method, namely the total renewable primary energy (RPE) and non-renewable primary energy (NRPE) Use [9–12, 18]. Other indicators are the ozone layer depletion potential (ODP) [9, 12, 17, 18], the acidification potential (AP) [9, 12, 17, 18], the eutrophication potential (EP) [9, 12, 17, 18], the abiotic resource depletion potential (ADP) [12, 18] and the photochemical oxidation creation potential (POCP) [9, 12, 17, 18]. The calculation of less common indicators, as exergy [33] or emergy [34], was disregarded, although their employment may provide further useful outcomes. Regarding the economic aspects, it is well known that these are among the first considerations that an investor or a building owner assesses before the beginning of a building design or renovation. Furthermore, with specific reference to the nearly zero-energy buildings defined by the European Energy Performance of Buildings Directive, it is suggested to compare many sets of building interventions to identify the cost-optimal combination [35]. Nevertheless, seven out of thirteen of the studies assessed the economic aspects of the interventions [11, 13–16]. In detail, three of these works compared the construction cost to the GWP in a multi-objective optimization study [11, 16], two studies included both investment and operating costs [13] while the remaining two also accounted for the maintenance costs [14, 15]. Including embodied and operating terms for both costs and impacts in some multi-objective studies allowed for identifying that some indicators might be nonconflicting. In detail, embodied primary energy and embodied GWP [15] and CED and GWP [17] couples showed a quasi-linear relationship. For example, Kiss and Szalay highlighted that CED, GWP and POCP tend to the same direction in [17], as was shown by Montana et al. in [13–15] for embodied energy, embodied GWP and investment cost. Moreover, minimizing the Life Cycle Cost (LCC), CED and GWP in the same optimization led the space of objective functions to become a cloud of solutions with the Pareto Front being concentrated at the base of this cloud, although the proportionality relation between the functions is not exactly linear [15]. On the opposite, investment costs and embodied impacts were confirmed to be conflicting with operating energy demand [13–15] and with life cycle GWP [16].

3.3.2

Variables Categories

The variables usually assessed in the optimization of buildings energy performance can be grouped in five main categories:

Multicriteria-Oriented Optimization of Building Energy …

251

• • • •

Early design parameters, as the building orientation or the number of floors; Opaque envelope components, as walls materials and thicknesses; Transparent envelope components, as windows glazing, frame or surface; Heating, ventilation and air-conditioning (HVAC) equipment features, as the rated size or the operation schedule of boilers or the share of heating provided by heat pumps and boilers; • Renewable energy sources (RES) systems features, as photovoltaic (PV) and solar collectors’ surfaces. The “opaque components” category is the most commonly considered, and each of the studies assessed the optimal material or the optimal thickness at least for one envelope component. More in detail, the insulation material and thickness are the most popular variables, but also the use of concrete and bricks was optimized in some studies [11, 13, 18]. The assessment of the best HVAC was also common, although it was changed only parametrically and out of the optimization process in a couple of studies [9, 10]. The heating system is the predominant topic, since most of the studies were performed in cold climates, while space cooling or ventilation technologies were hardly included in the LCA performance optimization [13, 16, 17]. Furthermore, the embodied impacts of the equipment were sometimes neglected [10]. Early design parameters were included only in three studies, with two of these assessing the optimal number of floors [17] and the third one the optimal position of the supporting columns of a garage [18]. This is due to the fact that the renovation of existing buildings is more common than design of new ones. Since in optimization studies many variables are involved, assessing the influence of each one is hard. Furthermore, it is well known that the building physics is highly nonlinear, implying that the influence of a combination of interventions is different than the sum of the individual effects. Nevertheless, embodied impacts can be well approximated to be linear. For example, doubling the insulation thickness means more or less doubling the related embodied impacts, but the influence on the building heating and cooling loads cannot be predicted and specific calculations are needed. Anyway, having an indication on the influence of each intervention may help both designers and researchers in selecting the alternatives in a more conscious way. For this reason, Hollberg et al. performed many parametric analyses on the optimal insulation thickness with many insulation materials, heating systems and also time horizons [9, 10]. Although some aspects, as the embodied impacts in the heating systems, were neglected, this kind of results can really help future researchers in selecting a reasonable range of variables in their optimization studies, thus reducing the computational time and avoiding the risk of assessing useless or non-convenient values of variables. A recap of the main information provided on objective functions and variables is given in Table 3.

252

M. Francesco et al.

Table 3 Recap of the main features of the reviewed studies Aim

Objective functions

Variables

Reference

Design of a plus-energy house

– Construction cost – GWP

– Insulation thickness of walls, floor and roof – Windows area and glazing – Ventilation system – PV modules

[16]

Refurbishment

– GWP NRPE was also assessed

– Insulation thickness [10] Service life, heating system and insulation material were changed parametrically

Refurbishment

– GWP RPE, NRPE, ODP, AP, EP, and POCP were also assessed

– Insulation material and [9] thickness Heating system and energy mix were changed parametrically

Design of a garage

– GWP PE, NRPE, ODP, AP, EP, POCP, ADP (both mineral and fossil fuels) were also assessed

– Position of the columns – Slab thickness – Concrete quality

[18]

Refurbishment

Construction cost/GWP ratio

– Insulation material and thickness – Cladding material – Heating system

[11]

Refurbishment

– Construction cost – GWP NRPE was also assessed

– Insulation material and thickness – Cladding material – Heating system

[11]

Refurbishment

– – – – – – – –

RPE NRPE GWP ODP AP EP POCP ADPE

– Insulation material and thickness – Windows glazing – Heating system

[12]

Design

– – – – – –

GWP AP ODP POCP EP CED

– Number of storeys – Insulation material and thickness of walls and roof – Window areas – Fixtures shading and glazing type – HVAC system

[17]

(continued)

Multicriteria-Oriented Optimization of Building Energy …

253

Table 3 (continued) Aim

Objective functions

Variables

Reference

Design

– – – – – –

– Number of storeys – Insulation material and thickness of walls and roof – Window areas – Fixtures shading and glazing type – HVAC system

[17]

Refurbishment

– Construction, operating and maintenance costs – GWP – CED

– Additional insulation [13, 14] material and thickness – Additional thermal mass – HVAC system – Electricity production systems – Electricity and thermal storages

Refurbishment

– Construction, operating and maintenance costs – GWP – CED

– Additional insulation [13, 14] material and thickness – Additional thermal mass – HVAC system – Electricity production systems – Electricity and thermal storages

Refurbishment

– Construction, operating and maintenance costs – GWP – CED

– Additional insulation material and thickness – Cladding replacement – Transparent materials glazing and frames – HVAC system – PV modules

Refurbishment

– Construction, operating and maintenance costs – GWP – CED

– Additional insulation [14] material and thickness for each orientation – Additional thermal mass for each orientation – Windows materials glazing and frames for each orientation – HVAC system – Electricity production systems – Electricity and thermal storages

GWP AP ODP POCP EP CED

[15]

254

3.3.3

M. Francesco et al.

Constraints Setting and Management

The constraints are relations used to limit the number of solutions obtained in an optimization study. In the space of solutions, they can be identified as lines or planes at the boundary. As an example, the absolute minimum value of the graph at the left in Fig. 7 is about −6.3, but, if the space of solutions is constrained by the plane z = −3.5, the feasible minimum of the problem becomes the circumference resulting from the intersection between the curve and the horizontal plane, as in the graph at the right in Fig. 7. The constraints can be well managed in linear programming, since the optimum is located in one of the vertexes identified by the constraints, thus allowing to employ exact methods as the simplex algorithm to identify the best value of a problem. Nevertheless, despite the flexibility of heuristic and genetic algorithms, the solution of constrained optimization problems is not an easy task. The most common approach is the adoption of penalty functions, converting the problem to an unconstrained optimization problem, although other approaches were developed in the literature [36] and in [16, 30]. In the reviewed papers, penalty functions were imposed to adopt a specific combination of insulation and cladding materials in [11] or to set a minimum distance between the supporting cement columns in [18]. The adoption of penalty functions requires defining meta-parameters for the penalty function. To avoid this, the constraint (of a positive annual energy balance) in [16, 30] was handled by integrating it as the first criterion (out of three) in the NSGA-II selection steps (reproduction and replacement). The two remaining criteria are the Pareto Front rank of the solution, and its Crowding distance. In other works [14, 15], constraints were imposed to specify the thermal features of windows, e.g. if the triple glazing is preferred to the double glazing, set the thermal transmittance to x rather than to y, or to impose only one insulation material and only one cladding material on external walls or roof. These studies were developed on MOBO, that is equipped with an automatic constraint handling technique for most of the algorithms [25].

Fig. 7 Comparison between unconstrained (left) and constrained (right) optimization problem

Multicriteria-Oriented Optimization of Building Energy …

255

3.4 Parameters and Data Quality The adoption of accurate data is a crucial issue in Life Cycle Assessment studies, since the results may be sensitive to site-specific conditions [37]. Nevertheless, optimization studies are often employed to obtain generic indications on the problem analysed in the study, adopting simplified mathematical models to quickly obtain an indication and then verifying the solutions through more detailed simulations [38]. According to this philosophy, all the studies analysed in this review employed secondary data, i.e. average data from literature, also in order to generalize their results. LCA impacts were drawn from LCA international databases as Ecoinvent [16, 17], KBOB [10, 11], Ökobau [9, 12, 15], or from the Environmental Product Declarations, i.e. reports based on ISO 14025 [39] and EN 15804 [40] standards that evaluate the LCA impacts of specific products, developed by companies in order to improve their sustainability or to show their attention to the environment-related issues [41]. In the same way, costs data were collected from databases [15, 16] or market reports [13].

4 Results from the Annex 72 Case Studies 4.1 General Guidelines All the studies reviewed in this chapter can be included in a unique, generic workflow, according to Fig. 8 [12]. In detail, the reference geometrical model of the building is first created and used as an input, indicating the building materials and layer thickness for each envelope component or the type of building services. In a second step, auxiliary information as the local climate or the reference period is defined. Subsequently, the variables are specified and the optimization run is started, assessing the embodied and use phase terms related to each impact function separately and then aggregating these terms (usually). The fitness function of each building configuration is assessed until the optimization ends, i.e. when a convergence criterion is satisfied or when the maximum number of iterations is reached. The results are then investigated in order to identify the variables behaviour in the best solutions and the optimal interventions to be adopted for the building.

4.2 Building Envelope For a given heating/air conditioning system, very different interventions on the envelope were identified, up to the point of preferring no improvement (e.g. for heat pump heating powered by electricity from renewable sources) [9].

256

M. Francesco et al.

Fig. 8 General workflow for the optimization of buildings energy performance

Although it seems obvious, from the point of view of the associated impact, materials of natural origin (e.g. cellulose) are preferred to synthetic ones (e.g. EPS), while an economic optimization suggests the opposite, since natural materials are more expensive [11].

Multicriteria-Oriented Optimization of Building Energy …

257

4.3 Renewable Energy Systems Very few of the analysed studies assessed the installation of RES. Montana et al. [15] proposed to install a rooftop photovoltaic system and solar thermal collectors in a large residential building, with both technologies being disregarded during the optimization, also because of the large number of variables assessed. In detail, the optimization preferred to install a district heating system instead of the solar collectors, maybe for cost-related reasons. The result did not take into account thermodynamic considerations, such as the temperature level at which heat must be supplied, but only the energy that can be provided by the different sources and the impacts incorporated into the systems (solar collectors vs. heat exchanger for heat pump or district heating, assuming the network being already installed). On the opposite, results from [16] showed that the installation of PV should be promoted in addition to insulating the envelope (an optimal thickness being identified) as intervention optimizing both installation costs and GWP.

4.4 Climate and Occupancy Influence Most of the reviewed studies regarded continental or oceanic climate cities, confirming that these countries are more sensitive to the assessment of the energy and environmental impacts of buildings. In detail, one study was developed in Northern France [16], five in Germany [9–12], two in Denmark [13, 15], two in Southern Italy [5, 6] and two in Hungary [17]. Thus, only two studies were focussed on mild climate, with one of them being a comparison of the performance of the same building in Mediterranean and oceanic climate. The outcomes of this study highlighted that the same insulation materials are optimal in both climates, with higher thicknesses being preferable in cold climates. The existing variability in occupants’ behaviour also influences the identification of the best solutions. In [16, 30], the optimization results with three types of households (single person, a retired couple and a young working couple with a child) were compared. Results of this study showed the existence of preferred solutions (e.g. triple glazing on the North East facade), but also solutions that strongly depend on the household kind, especially on the equipment features (e.g. the number of PV modules or the installation of a grey water heat recovery system).

5 Conclusion This chapter described the research experience of IEA-EBC Programme Annex 72 members on the application of multi-objective optimization processes for selecting

258

M. Francesco et al.

the more suitable design or retrofit actions aimed at improving different aspects (energy, environmental, economic, etc.) of buildings in a life cycle perspective. Thirteen case studies were examined to identify and analyse different optimization approaches and to provide useful information to building designers and decisionmakers. As a generic summary of the present analysis, it is possible to highlight the difficulty in drawing generic guidelines on the methods for the optimization of life cycle performance of buildings. This is mainly due to the influence of approaches, software and algorithms, selected objective functions, variables, constraints and parameters, which are generally different for each selected study. However, a generic methodological framework can be identified, starting from the preliminary building model development and ending with the identification of the optimal solutions (e.g. regarding building envelop, use of renewable energy technologies, influence of climate, occupancy and time horizon). This framework, detailed case by case according to the peculiarities of the building under investigation, can help the stakeholders involved in the building design, construction and management in the selection of optimal interventions to be implemented.

References 1. Eurostat (2019) Complete energy balance 2019, edition 2019. https://ec.europa.eu/eurostat/ web/energy/data/energy-balances. Accessed 30 Sept 2019 2. International Energy Agency (IEA) (2018) Global alliance for buildings and construction (GlobalABC). 2018 global status report 3. European Parliament and Council (2010) Directive 2010/31/EU of 19 May 2010 on the energy performance of buildings (recast) 4. European Parliament and Council (2018) Directive (EU) 2018/844 of 30 May 2018 amending Directive 2010/31/EU on the energy performance of buildings and Directive 2012/27/EU on energy efficiency 5. International Organization for Standardization (ISO) (2006) ISO 14040:2006—environmental management. Life cycle assessment. Principles and framework 6. International Organization for Standardization (ISO) (2017) ISO 14044:2006 + Amd 1:2017— environmental management. Life cycle assessment. Requirements and guidelines 7. European Committee for Standardization (CEN) (2011) EN 15978:2011—sustainability of construction works. Assessment of environmental performance of buildings. Calculation method 8. International Energy Agency (IEA) (n.d.) IEA EBC Annex 72—assessing life cycle related environmental impacts caused by buildings. https://annex72.iea-ebc.org. Accessed 15 Sept 2020 9. Hollberg A, Ruth J (2014) A parametric life cycle assessment model for façade optimization. Build Simul Optim 10. Hollberg A, Ruth J (2013) Parametric performance evaluation and optimization based on life cycle demands. In: 8th energy forum on advanced building skins, Bressanone 11. Klüber N, Hollberg A, Ruth J (2014) Life cycle optimized application of renewable raw materials for retrofitting measures. In: World sustainable building 2014, Barcelona, pp 1–7 12. Hollberg A, Ruth J (2016) LCA in architectural design—a parametric approach. Int J Life Cycle Assess 21:943–960. https://doi.org/10.1007/s11367-016-1065-1

Multicriteria-Oriented Optimization of Building Energy …

259

13. Cellura M, Longo S, Montana F, Riva Sanseverino E (2019) Multi-objective building envelope optimization through a life cycle assessment approach. In: 2019 IEEE International Conference on Environment and Electrical Engineering and 2019 IEEE Industrial and Commercial Power Systems Europe (EEEIC/I&CPS Europe), Genoa. Institute of Electrical and Electronics Engineers (IEEE), pp 1–6 14. Montana F (2020) Multi-objective optimization of buildings and building clusters performance: a life cycle thinking approach. University of Palermo 15. Montana F, Kanafani K, Wittchen KB, Birgisdottir H, Longo S, Cellura M, Riva Sanseverino E (2020) Multi-objective optimization of building life cycle performance. A housing renovation case study in Northern Europe. Sustainability 12(18):7807. https://doi.org/10.3390/su1 2187807 16. Recht T, Schalbart P, Peuportier B (2016) Ecodesign of a ‘plus-energy’ house using stochastic occupancy model, life-cycle assessment and multi-objective optimisation. In: Hamza N, Underwood C (eds) Building simulation and optimization conference, Newcastle upon Tyne 17. Kiss B, Szalay Z (2020) Modular approach to multi-objective environmental optimization of buildings. Autom Constr 111. https://doi.org/10.1016/j.autcon.2019.103044 18. Hollberg A, Heidenreich C, Ruth J, Hartung R, Herzog S (2014) Using evolutionary optimization for low-impact solid constructions. In: World sustainable building 2014, Barcelona 19. Robert McNeel & Associates (n.d.) Rhinoceros web page. https://www.rhino3d.com/. Accessed 7 Apr 2020 20. Trimble Navigation (n.d.) SketchUp website. https://www.sketchup.com/. Accessed 3 Apr 2020 21. Izuba Énergies (n.d.) PLEIADES software website. https://www.izuba.fr/logiciels/. Accessed 28 Aug 2020 22. U.S. Department of Energy (n.d.) EnergyPlus web page. https://energyplus.net/. Accessed 28 Apr 2020 23. International Organization for Standardization (ISO) (2008) ISO 13790:2008—energy performance of buildings. Calculation of energy use for space heating and cooling 24. DIN (2011) DIN V 18599-2:2011—Energetische Bewertung von Gebäuden. Berechnung des Nutz-, End- und Primärenergiebedarfs für Heizung, Kühlung, Lüftung, Trinkwasser und Beleuchtung. Teil 2: Nutzenergiebedarf für Heizen und Kühlen von Gebäudezonen (in German) 25. Palonen M, Hamdy M, Hasan A (2013) MOBO a new software for multi-objective building performance optimization. In: 13th conference of international building performance simulation association, pp 2567–2574 26. MOBO web page (n.d.) https://ibpsa-nordic.org/tools.html. Accessed 27 Mar 2020 27. Gilles F, Bernard S, Ioannis A, Simon R (2017) Decision-making based on network visualization applied to building life cycle optimization. Sustain Cities Soc 35:565–573. https://doi.org/ 10.1016/j.scs.2017.09.006 28. Wortmann T (2019) Genetic evolution vs. function approximation: benchmarking algorithms for architectural design optimization. J Comput Des Eng 6:414–428. https://doi.org/10.1016/ j.jcde.2018.09.001 29. Longo S, Montana F, Riva SE (2019) A review on optimization and cost-optimal methodologies in low-energy buildings design and environmental considerations. Sustain Cities Soc 45:87–104 30. Recht T (2016) Study of the ecodesign of plus-energy houses (Étude de l’écoconception de maisons à énergie positive, in French). PSL Research University (prepared in MINES ParisTech) 31. Bahramian M, Yetilmezsoy K (2020) Life cycle assessment of the building industry: an overview of two decades of research (1995–2018). Energy Build 109917 32. Lasvaux S, Favre D, Périsset B, Bony J, Hildbrand C, Citherlet S (2015) Life cycle assessment of energy related building renovation: methodology and case study. Energy Procedia 78:3496– 3501. https://doi.org/10.1016/j.egypro.2016.10.132 33. De Meester B, Dewulf J, Verbeke S, Janssens A, Van Langenhove H (2009) Exergetic lifecycle assessment (ELCA) for resource consumption evaluation in the built environment. Build Environ 44:11–17. https://doi.org/10.1016/j.buildenv.2008.01.004

260

M. Francesco et al.

34. Reza B, Sadiq R, Hewage K (2014) Emergy-based life cycle assessment (Em-LCA) of multiunit and single-family residential buildings in Canada. Int J Sustain Built Environ 3:207–224. https://doi.org/10.1016/j.ijsbe.2014.09.001 35. European Commission (2012) Commission Delegated Regulation (EU) No 244/2012 of 16 January 2012 supplementing Directive 2010/31/EU of European Parliament and of the Council 36. Chehouri A, Younes R, Perron J, Ilinca A (2016) A constraint-handling technique for genetic algorithms using a violation factor. J Comput Sci 12:350–362 37. Cellura M, Longo S, Mistretta M (2011) Sensitivity analysis to quantify uncertainty in life cycle assessment: the case study of an Italian tile. Renew Sustain Energy Rev 15–9:4697–4705. https://doi.org/10.1016/j.rser.2011.07.082,ISSN1364-0321 38. Pereira S, Ferreira P, Vaz AIF (2015) A simplified optimization model to short-term electricity planning. Energy 93:2126–2135. https://doi.org/10.1016/j.energy.2015.10.040 39. International Organization for Standardization (ISO) (2006) ISO 14025:2006—environmental labels and declarations—type III environmental declarations—principles and procedures 40. European Committee for Standardization (CEN) (2012) EN 15804:2012—sustainability of construction works. Environmental product declarations. Core rules for the product category of construction products 41. The International EPD® System (n.d.) Environmental product declarations (EPD). https:// www.environdec.com/. Accessed 27 Mar 2020

Multi-objective Genetic Algorithm Optimization of HVAC Operation: Integrating Energy Consumption, Thermal Comfort, and Productivity Sokratis Papadopoulos and Elie Azar

Abstract An important share of the energy demand of buildings is attributed to the heating, ventilation, and air conditioning (HVAC) systems. Simple changes in the operational settings of these systems, such as adjusting the thermostat setpoint temperatures, can have a significant impact on building performance (e.g., energy consumption and costs). In parallel, changes in indoor environmental conditions can directly impact occupants’ comfort, wellbeing, and productivity. A review of the literature indicates that the stated metrics of building performance are often studied in isolation, failing to capture their cross-effects and potential implications for building operation strategies. This chapter presents a genetic algorithm (GA) multi-objective optimization (MOO) that captures the trade-offs between—and optimizes—three competing objectives of building performance: (i) energy consumption, (ii) thermal comfort, and (iii) productivity. Using building performance simulation (BPS), models of three archetype office buildings located in different climate zones are used to showcase and validate the framework’s capabilities. Optimal HVAC setpoint settings are found to reduce energy consumption by up to 25.8% while maintaining acceptable comfort and productivity levels of occupants. Additionally, the non-dominated solutions for buildings located in different weather zone vary statistically, motivating the need for climate-sensitive HVAC operation strategies and standards. Keywords HVAC setpoints · Multi-objective optimization · Genetic algorithm · Energy · Comfort · Productivity

S. Papadopoulos Department of Civil and Urban Engineering and Center for Urban Science and Progress, New York University, 1 Metrotech Center, Brooklyn, NY 11201, USA e-mail: [email protected] E. Azar (B) Department of Industrial and Systems Engineering, Khalifa University of Science and Technology, PO Box 127788, Abu Dhabi, UAE e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2021 J. Ren (ed.), Energy Systems Evaluation (Volume 2), Green Energy and Technology, https://doi.org/10.1007/978-3-030-67376-5_11

261

262

S. Papadopoulos and E. Azar

1 Introduction 1.1 Background The building sector is a crucial element of the infrastructure due to its impact on resources and people. On the one hand, it is considered as one of the most energyintensive sectors with significant implications on natural resources and carbon emissions [21]. On the other, and as people spend the majority of their lifetimes indoors [24], building indoor environmental conditions can significantly impact the comfort, wellbeing, and productivity of occupants [4]. Among the different building systems, the heating, ventilation, and air conditioning (HVAC) system plays a major role in building performance as it is responsible for controlling indoor environmental conditions and providing comfortable and healthy conditions for occupants. It is also estimated to contribute to more than 40% of building energy demand [27, 39]. Consequently, building retrofit strategies are commonly studied and adopted to improve the performance of HVAC systems, mainly consisting of two types: design-based and operation-based retrofits [6]. Design-based retrofits mostly aim to improve the performance of building systems (or components) by replacing them with more energy-efficient ones. Common targets of such retrofits include the building envelope (e.g., windows, walls, and roofs), HVAC components (e.g., fans and chillers), and electrical systems (e.g., lighting and appliances) [15]. In parallel to the research and development of design solutions, policy instruments, such as building standards, codes, and certifications, often promote their adoption on large scales. Such instruments often follow a prescriptive path to energy efficiency, where the properties or minimal efficiencies of a certain system are specified (e.g., a minimum acceptable U-value for exterior walls or the minimum accepted coefficient of performance for chillers). In some cases, performance-driven instruments are also adopted to rather evaluate buildings based on their actual measured performance benchmarked against other buildings or baselines [31]. Design-based retrofitting is well-known and documented in the literature; however, it also presents important drawbacks or limitations, three of which are presented next [35]. Firstly, design retrofits can be very capital intensive with diminishing returns on investment. Secondly, they often require modifications to current building systems, which is not always possible especially in old or historic facilities. Thirdly, despite their significant energy saving potential, numerous studies indicate that their benefits can be negated by inefficient operation patterns. The abovementioned shortcomings motivate the need for operation-based retrofits, which are discussed next. It is important to note that the two approaches are not mutually exclusive but rather complementary. Operation-based retrofits consist of actions or strategies taken by people (i.e., facility managers or building occupants) to ensure efficient operation of various building systems (e.g., lighting, equipment, and HVAC) [6]. Simple strategies include minimizing the use of equipment and lighting when spaces are unoccupied or

Multi-objective Genetic Algorithm Optimization of HVAC …

263

adjusting the HVAC thermostat temperate setpoints to avoid excessive cooling and heating loads. In recent years, a particular emphasis is found on the latter as numerous studies and initiatives confirm that thermostat setpoint adjustments can lead to significant energy reductions at little to no cost (e.g., [3, 5, 18, 20, 26]. For instance, Al Amoodi and Azar [3] found that a ±2 °C variation in cooling setpoints can lead to energy reductions exceeding 16% for office buildings and 25% for classroom buildings. Ghahramani et al. [18] observed that selecting daily optimal setpoints in office buildings in the range of 22.5 ± 3 °C, instead of a fixed baseline of 22.5 °C, can result in average savings exceeding 16%. In parallel to their implications on energy consumption, setpoint settings directly determine the indoor air temperature of buildings, a key determinant of occupants’ thermal comfort levels. The most common and widely adopted model of thermal comfort is Fanger’s model, which expresses human thermal sensation based on a steady-state heat balance principle [16]. The inputs of this model include environmental factors (i.e., air and radiant temperature, and airspeed) as well as personal factors (i.e., clothing level and metabolic rate). The main outcome of the models are the predicted mean vote (PMV) and the predicted percentage of dissatisfied (PPD) metrics [22]. Alternatives to Fanger’s model exist and are worth noting, such as adaptive comfort models that consider acceptable indoor temperatures as a function of outdoor temperatures [11, 30]. Finally, some efforts aimed to link and quantify the relationship between indoor temperature and productivity. Since it is challenging to directly quantify productivity in office buildings, researchers often rely on in-situ measurements of work performance, such as absenteeism, reported productivity levels, or performance at basic cognitive ability tests (e.g., reading and typing). Two studies successfully quantified the impact of temperature on worker’s performance. The first study is by Seppanen and Fisk [37], who observed an inverted-U relationship between indoor temperature and cognitive performance. Similarly, Kosonen and Tan [25] investigated productivity loss in thinking and typing tasks of occupants as a function of PPD, finding a logarithmic relationship between the two metrics. This study is particularly interesting in the context of the current work as it connects comfort to productivity.

1.2 Problem Statement A sustainable building performance requires a holistic approach that optimizes diverse metrics of building performance, such as energy, comfort, and productivity [38]. In practice, studies on operation-based retrofits present the following shortcomings that motivate the need for this research. Starting with the scope of the studies, most cover a single metric of performance, leaving the impact of other metrics unknown. In particular, productivity-related metrics are often overlooked despite a growing interest in the literature to connect indoor conditions to occupants’ performance. Finally, studies that consider more than one performance metric rarely integrate them in their analysis (e.g., in an optimization scheme) to account for

264

S. Papadopoulos and E. Azar

possible synergetic or conflicting effects. The analysis is often limited to a discrete and predefined set of potential solutions to test (e.g., sensitivity analysis of parameters over a pre-defined range to determine the optimal setting), as opposed to a less supervised design that can potentially lead to unexpected solutions to the problem.

1.3 Objectives and Chapter Organization The goal of this study is to present a scalable multi-objective optimization (MOO) framework to guide building operation strategies while integrating key building and occupant performance metrics. The independent (decision) variables investigated are the thermostat cooling and heating setpoint temperature settings for occupied hours (e.g., working hours on weekdays) and unoccupied hours (e.g., after-hours, weekends, and holidays). The main outcome consists of Pareto optimal solutions that are identified based on three objectives: (i) energy consumption, (ii) thermal comfort, and (iii) productivity. The framework can aid researchers and decision makers (e.g., facility managers) in understanding the trade-offs between competing objectives of building performance and devise optimal occupant-centric operation strategies. The methods presented in this chapter are demonstrated via a case study of three archetype medium office buildings located in three different US climate zones. It is worth noting that the chapter builds on two previous works by the authors. In Papadopoulos and Azar [33], an initial proof-of-concept was presented covering comfort and productivity, however, with an application limited to a single building and without exploring practical applications to guide decision making. In Papadopoulos et al. [32], multiple buildings across different climate zones were considered, but only energy and comfort were included in the optimization scheme. The current chapter (i) combines and expands both works, (ii) integrates three metrics of performance, (iii) tests the framework on three buildings in different climate zones, and (iv) thoroughly discusses implications and applications. The rest of the chapter is organized as follows. In Sect. 2, the proposed framework is presented, starting with a general overview of the methods and BPS models, followed by the coupling engine and the genetic algorithm (GA) MOO scheme. Section 3 presents the results of the case study on the three selected archetype office buildings. Section 4 discusses the implications of the work, limitations, and possible future directions.

Multi-objective Genetic Algorithm Optimization of HVAC …

265

2 Methodology 2.1 Overview The general methodology is presented in Fig. 1 and further discussed in the following sections. At the center of the figure, a coupling engine is shown to automate the optimization scheme and connect the BPS capabilities of the framework to its MOO GA capabilities (Sect. 2.3). Starting with the former, BPS models are first developed to emulate the energy performance of the studied building(s) (Sect. 2.2). In parallel, the lower part of Fig. 1 illustrates the MOO GA, which evaluates the objective functions given a stop criterion (Sect. 2.4). Once met, a Pareto front of the nondominated solutions is generated, and the simulation ends. The coupling engine is also used to tune the BPS models’ parameters based on the decision variables used in the optimization.

2.2 BPS Modeling BPS is a bottom-up engineering modeling technique that is often used to emulate the performance of buildings [9]. The main inputs of BPS models include building characteristics (e.g., civil, mechanical, and electrical), operation settings and schedules, and outdoor climate conditions. The EnergyPlus software is chosen in the current

Building Performance Simulation

.idf

Run

.csv

Coupling Engine

Objective Functions

No

Multi-objective Optimization Genetic Algorithm

Stopping Criterion Met?

Fig. 1 Proposed framework [32]

Yes

Pareto Front

266

S. Papadopoulos and E. Azar

work as the BPS engine, given its robustness, open-source nature, and interoperability with other software. BPS models of three archetype medium-sized office US buildings are used in this study. The models were obtained from the US DOE’s database of commercial prototype buildings [10] and cover three climatic zones: 2B (Phoenix), 4A (Baltimore), and 6A (Minneapolis). Temperature wise, the selected buildings cover hot (Phoenix), cold (Minneapolis), and mild (Baltimore) weather zones. In terms of humidity, Minneapolis and Baltimore fall into the moist zone, whereas Phoenix is a rather dryer location. The reasoning behind covering a spectrum of weather conditions is twofold. First, it highlights the generalization capabilities of the proposed framework, which can be applied to any building as long as an energy model (e.g., BPS) of the building is available. Second, it helps investigate whether the optimal HVAC operation changes over the building’s location. Building standards (e.g., ASHRAE 90.1) do not recommend different HVAC setpoints for commercial buildings located in different climates. This study will examine whether different buildings require different setpoints for optimal overall performance. The main building characteristics used to develop the BPS models are summarized in Table 1. These include information pertaining to the geometry of the buildings, construction material, systems, and schedules. Additional information can be found in Deru et al. [14]. Table 1 Description of the three studied archetype buildings [14] Parameter

Building #1

Building #2

Building #3

Climate zone

2B (Phoenix, AZ)

4A (Baltimore)

6A (Minneapolis)

Type and size

Office (medium-sized)

Construction year category

After 1980

Number of floors

3

Total floor area

4982 m2

Window-to-wall ratio (WWR)

0.33

Wall U-value (W/m2 -K)

1.36

0.51

0.37

Roof U-value (W/m2 -K)

0.26

0.33

0.26

Glazing U-value (W/m2 -K)

6.93

3.35

2.95

Glazing solar heat gain coefficient (SHGC)

0.25

0.36

0.39

Lighting density

16.89 W/m2

Equipment density

10.76 W/m2

HVAC—cooling system

Packaged unit (continued)

Multi-objective Genetic Algorithm Optimization of HVAC …

267

Table 1 (continued) Parameter

Building #1

HVAC—heating system

Furnace

Building #2

HVAC—ventilation

Variable air volume (VAV)—multi-zone

Building schedule

Weekdays: 6 a.m.–10 p.m. Saturdays: 9 a.m.–5 p.m. Sundays and holidays: closed

Building #3

2.3 Simulation Coupling The aim of the coupling scheme is to communicate with and control both the BPS models and the optimization algorithm. This allows automating the running of the models and the execution of the MOO GA scheme. More specifically, the coupling engine developed in MATLAB loads the text-based input file from EnergyPlus (.idf format). Then, it edits the .idf files by assigning specific values to the decision variables of interest as part of the optimization process (detailed in the upcoming sections). The coupling engine then launches the EnergyPlus simulations and saves the output files (.csv format) containing the energy estimates of the models. Next, the objective functions are evaluated, and the same process keeps repeating until the stopping criteria are met. Upon convergence, a set of non-dominated solutions is generated.

2.4 Multi-objective Optimization 2.4.1

Overview

The objective functions in the current work are evaluated using simulation, which makes the MOO nonlinear, non-differentiable, and discontinuous. Heuristic approaches (e.g., genetic algorithms) are known to handle such characteristics of the optimization well and are often used in similar simulation-based studies [29]. The specific algorithm used is a controlled (elitist) GA, which is an NGSA-II variant [12]. A GA emulates the evolution of living organisms by simulating natural selection processes [19]. As such, GA exploits the search space to solve optimization problems in an intelligent manner by following the principles of C. Darwin’s “survival of the fittest” theory. More specifically, each variable is encoded as a set of decision variables (analogous to ‘genes’), and the population individuals as solutions (analogous to ‘chromosomes’). In each generation (i.e., iteration), only the individuals considered to be the fittest survive the evolution process, eventually leading to a convergence where only the fittest remain. These form the optimal solutions to the problem.

268

S. Papadopoulos and E. Azar

The constraints of the decision variables used in the current work are well defined; they correspond to the lower and upper bounds of the ranges over which they were varied. Consequently, the initial population was generated using a random process to cover a broad search space and maximize diversity between the individuals [36]. Following the initialization of the population, the MOO GA algorithm begins according to the step-by-step process shown in Fig. 2. Table 2 lists the main genetic operators used in the algorithm, their description, and the values assigned to them. Additional details on the concepts of the NSGA-II algorithm can be found in Deb et al. [13], which was also explained and applied in Papadopoulos et al. [32].

Fig. 2 Pseudo code of the GA MOO process [32]

Table 2 Description and values used for the genetic operators Variable

Description

Value [28]

Elite count

Individuals in a specific iteration with the best fitness values 2 and surviving to the next generation

Pcrossover

Probability that parents crossover, forming new solutions (children)

0.8

Pmutation

Probability that parents’ characteristics change, producing mutated solutions (children)

0.01

Stopping criterion #1 Reaching a specific number of generations

50

Stopping criterion #2 Average relative change in fitness falling below a threshold

1e−6

Multi-objective Genetic Algorithm Optimization of HVAC …

2.4.2

269

Decision Variables and Objective Functions

The decision variables of the MOO algorithm are set to the four HVAC parameters of interest considered in this work, which are the cooling and heating setpoint temperatures for occupied and unoccupied periods. All variables are continuous and are defined by lower and upper bounds guided by building standards that specify “acceptable” indoor temperature ranges [1, 2], which are listed next: – – – –

Cooling setpoint temperature during ‘occupied’ periods (x occ_cool ), 22–27 °C Heating setpoint temperature during ‘occupied’ periods (x occ_heat ), 17–22 °C Cooling setpoint temperature during ‘unoccupied’ periods (x unocc_cool ), 27–30 °C Heating setpoint temperature during ‘unoccupied’ periods (x unocc_heat ), 14–17 °C.

Three objectives are evaluated by the MOO algorithm, namely energy consumption, thermal comfort, and productivity. Starting with the first, ‘energy consumption’ in this study represents the amount of energy in MWh consumed by the HVAC system over a period of one year to cover the cooling and heating loads of the building. While the EnergyPlus model estimates energy consumption with higher temporal granularity (e.g., daily or hourly), an aggregate yearly value is considered to simplify the MOO problem. As for the second objective, thermal comfort is assessed using the PPD metric, which represents the percentage of people dissatisfied with the thermal conditions of their building space [16, 22]. This metric is widely used in the field, and its possible range of values (0–100%) is easy to use in an optimization scheme. Productivity is the third objective function considered and is based on the work of Kosonen and Tan [25]. In their study, the authors experimentally observed the loss of productivity in thinking and typing tasks of occupants in air-conditioned spaces under different environmental conditions and while measuring the PPD among the study subjects. The data presented in their work were used in the current study to fit a curve and model productivity loss, as shown in Eq. 1: ProdLoss (%) = 13.366 ∗ log(PPD) − 14.895

(1)

In this formula, both the performance in typing and thinking tasks are accounted for with equal weights. The authors acknowledge that different buildings might exhibit different relationships between PPD and productivity loss. Furthermore, the formula was not validated by a large-scale experiment to allow generalizing its results. Nonetheless, it is still deemed adequate to illustrate the capabilities of the proposed general optimization framework. To summarize, the three objective functions are as follows: – Annual cooling and heating energy demand in MWh (EnergyConsumption) – Percentage of people dissatisfied (PPD) – Percentage loss of productivity in thinking and typing tasks (ProdLoss).

270

2.4.3

S. Papadopoulos and E. Azar

Implementation

The last step of the methodology consists of solving the MOO problem, which is done using the MOO GA solver of MATLAB. Three minimization functions defined, one for each of the objective functions defined in the previous section: min Z 1 (X ) for ‘EnergyConsumption’, min Z 2 (X ) for ‘PPD’, and min Z 3 (X ) for the decision variables to test   ‘ProdLoss’, where X is a vector containing xocc_ cool , xocc_ heat , xunocc_ cool , xunocc_ heat . These are bounded by their respective lower and upper values defined in the previous section, namely 22–27 °C for x occ_cool , 17–22 °C for x occ_heat , 27–30 °C for x unocc_cool , and 14–17 °C for x unocc_heat . The solver then outputs the set of non-dominated solutions for each of the three objectives.

3 Results The results and discussion of the MOO are presented next. The following sections cover the non-dominated solutions obtained (Sect. 3.1), an analysis of constrained solutions to guide decision making (Sect. 3.2), and a comparison between the studied weather zones (Sect. 3.3).

3.1 Non-dominated Solutions Non-dominated solutions are observed for the three studied buildings, given the conflicting nature of the studied objectives. The 3D Pareto fronts (sets of nondominated solutions) are shown in Fig. 3; the X-, Y-, and Z-axes represent the three objectives of the problem. Two main observations can be made. The first observation is the evident positive correlation between productivity loss and PPD, whereas they are both negatively correlated with energy consumption. In other words, solutions targeting to improve the buildings’ energy performance tend to have a negative impact on occupants’ thermal comfort and productivity levels. Second, the correlation between PPD and productivity loss is nonlinear. The finding can be attributed to the logarithmic relationship between the two metrics, as illustrated earlier in Eq. 1. In order to enhance the interpretability of Fig. 3, several solutions are labeled and discussed next. Each point is accompanied with an explanatory table, including the values of the three objectives: energy consumption (EC), thermal comfort (TC), productivity loss (PL), and the values of the decision variables (setpoints): cooling occupied (CO), heating occupied (HO), cooling unoccupied (CU), and heating unoccupied (HU). Specifically, point A in each graph corresponds to the metrics’ scores of the base case buildings, according to the operation setpoints defined by Deru et al. [14]. For instance, in the Baltimore building, the base case setpoints obtained from Deru et al. [14] for occupied hours are 24 °C for CO and 21 °C for HO. For the unoccupied

Multi-objective Genetic Algorithm Optimization of HVAC … a. Phoenix

EC: 247.09 MWh TC: 15.81 % PL: 22 %

EC: 338.61 MWh TC: 9.39 % PL: 15.04 %

B

CO: 27 oC HO: 17.8 oC CU: 29.9 oC HU: 14 oC

271

CO: 25 oC HO: 20.9 oC CU: 28.4 oC HU: 16.2 oC EC: 397.68 MWh TC: 8.31 % PL: 13.41 % EC: 386.22 MWh TC: 9.43 % PL: 15.10 %

12.3% energy savings

o

CO: 24 C HO: 21 oC CU: 26.7 oC HU: 15.6 oC

CO: 24.3 oC HO: 22 oC CU: 28 oC HU: 16.7 oC

D

A (Base Case)

C

b. Baltimore

B

EC: 216.72 MWh TC: 17.75 % PL: 23.55 %

EC: 247.66 MWh TC: 13.73 % PL: 20.11 %

CO: 26.8 oC HO: 17.9 oC CU: 27.9 oC HU: 14.3 oC

CO: 26.4 oC HO: 20.1 oC CU: 28.3 oC HU: 14.3 oC

25.8 % energy savings

EC: 333.74 MWh TC: 13.82 % PL: 20.21 % CO: 24 oC HO: 21 oC CU: 26.7 oC HU: 15.6 oC

D EC: 311.52 MWh TC: 10.16 % PL: 16.09 %

A

CO: 25.9 oC HO: 22 oC CU: 28.5 oC HU: 15.6 oC

(Base Case)

C

c. Minneapolis EC: 335.58 MWh TC: 27.25 % PL: 29.28 %

B

CO: 27 oC HO: 17 oC CU: 28.9 oC HU: 14.3 oC

EC: 460.54 MWh TC: 18.14 % PL: 23.84 % CO: 24 oC HO: 21 oC CU: 26.7 oC HU: 15.6 oC

EC: 396.04 MWh TC: 17.07 % PL: 23.02 %

14% energy savings

CO: 26.8 oC HO: 20.6 oC CU: 28.7 oC HU: 14.1 oC

D EC: 442.17 MWh TC: 13.25 % PL: 19.64 %

A (Base Case)

C

CO: 26.6 oC HO: 22 oC CU: 28.8 oC HU: 14.7 oC

Fig. 3 Pareto fronts for different building locations: a Phoenix, b Baltimore, c Minneapolis

272

S. Papadopoulos and E. Azar

hours, the setpoints follow more extreme values since no occupants are present during these periods, with 26.7 °C for cooling and 15.6 °C for heating. Under this particular HVAC operation scheme, the building consumes 333.74 MWh annually to cover its HVAC loads, the annual PPD level is 13.83%, and the productivity loss could reach 20.21%. Points B, C, and D, on the other hand, are specific solutions in the illustrated Pareto fronts. In turn, point B represents the solution where energy consumption is minimized, while point C corresponds to solutions where productivity loss and PPD are minimized. Solution D is one of the most interesting points on the Pareto front. In each graph, it is the non-dominated solution that improves the building’s energy performance from the base case scenario, without compromising occupants’ thermal comfort and productivity levels. Starting with the Baltimore results, point D solution corresponds to setpoints of 26.4 °C (CO), 20.1 °C (HO), 28.3 °C (CU), and 14.3 °C (HU). Through this optimal configuration of the HVAC system’s setpoints, energy savings up to 25.8% can be realized for this particular building. For the Phoenix and Minneapolis buildings, where cooling and heating loads prevail, lower energy consumption reduction can be achieved (i.e., 12.3% for Phoenix and 14% for Minneapolis). The observed savings are still considered significant, especially that they are solely based on changes in HVAC setpoints, potentially at no upfront costs (unlike technical retrofits). Thus, a major benefit of applying MOO to the HVAC system’s operations is to guide decision making and help identify HVAC operation strategies. In this case, point D, for example, illustrates optimal HVAC settings that reduce energy consumption without compromising occupants’ thermal comfort and productivity levels.

3.2 Constrained Solutions for Decision Making Another benefit of MOO is that problem-specific constraints can be applied to the non-dominated solution space to help select, given particular constraints, an optimal HVAC strategy. For instance, thresholds could be set by decision makers (e.g., building owner or facility manager) on the values of minimum or maximum desired energy consumption, thermal comfort, and productivity values. Thresholds can be guided by sources such as building standards (ASHRAE), green labeling codes (e.g., LEED, BREEAM), cost considerations (i.e., operation costs based on employees’ productivity levels), or other limits set by decision makers. An example of constrained decision making on MOO solutions is shown in Fig. 4. The threshold values used in the example are artificial and are used only to showcase the capabilities of the approach. These include an upper limit of 20% in productivity loss, 15% in PPD (as an average of 10 and 20% of occupant discomfort levels suggested in Brager and de Dear [7]), and 270 MWh/year in energy consumption (approximately 20% reduction in HVAC system’s energy consumption from base case values). As shown in Fig. 4, the set thresholds can be visually shown using a cuboid, where only two non-dominated solutions satisfy the above-mentioned constraint. When comparing both solutions to the base case building, we observe

Multi-objective Genetic Algorithm Optimization of HVAC …

Baltimore

273 EC: 253.02 MWh TC: 13.04 % PL: 19.43 %

EC: 333.74 MWh TC: 13.82 % PL: 20.21 %

Constrained Non-dominated Solutions

CO: 26.6 oC HO: 20.5 oC CU: 28 oC HU: 14.3 oC

CO: 24 oC HO: 21 oC CU: 26.7 oC HU: 15.6 oC

Base Case

EC: 261.85 MWh TC: 12.61 % PL: 18.98 % CO: 26.3 oC HO: 20.6 oC CU: 28 oC HU: 14.6 oC

Constrained Search Space

Fig. 4 Pareto front and constrained space (Baltimore building)

that energy savings up to 24% can be achieved, while slightly improving the thermal sensation and productivity of occupants. Looking deeper at the solutions’ corresponding setpoint values, we observe that during occupied periods, cooling setpoints can be increased by 2–2.5 °C without impairing thermal comfort and productivity. While lower in magnitude, a similar decrease can be observed in the heating setpoint with a drop of 0.5 °C. During unoccupied periods, building performance can be enhanced by setting the cooling temperature to 28 °C and heating temperature to 14.5 °C.

3.3 Comparison by Weather Zones From the analysis of the office buildings in different climate zones (Fig. 3), it was easy to observe that different setpoint configurations comprise each Pareto front. Such finding raises the question of whether adopting the same HVAC operation standards across different weather zones is valid or not. To get a deeper insight on this aspect, the values of the decision variables for the non-dominated solutions were plotted for the three building locations, as shown in Fig. 5. On a first look, except for the heating occupied setpoints, the solution ranges for the considered buildings appear to be different both in terms of average values and variances. For instance, for the cooling occupied setpoints of the Minneapolis building (upper left quadrant of Fig. 5), the solutions are clustered around a high setpoint of approximately 26.7 °C. This is expected given the cold climate of Minneapolis that requires low cooling loads, hence the limited negative impact of this parameter on the other studied performance metrics (i.e., thermal comfort and productivity). On the other hand, the Phoenix building shows a wider variability, given the direct impact of the same

274

S. Papadopoulos and E. Azar

Fig. 5 Box plots of non-dominated solutions

parameter on the problem’s objectives. In other words, changing occupied cooling setpoints in Arizona highly impacts energy use, thermal comfort, and productivity simultaneously. To statistically verify what is visually observed in Fig. 5, a one-way analysis of variance (ANOVA) was performed on all four setpoints. The purpose of oneway ANOVA is to statistically determine whether there are significant differences between the means of two or more independent groups (i.e., weather locations or cities). Assuming a population with j different groups and µ1,..., j being the mean of each category, one-way ANOVA tests the null hypothesis that all group means are equal (H0 : µ1 = µ2 = . . . = µ j ), compared to the alternative hypothesis that at least one group mean is different than the others (H1 : at least one mean is different). Table 3 presents the results of one-way ANOVA for the problem’s decision variables. As expected, in all variables but the occupied heating setpoint, the p-values are less than 0.05 (95% confidence level), suggesting that H0 is rejected and showing that the mean setpoint value is different in each location. The current practice in commercial building benchmarking [14] suggests different HVAC operation strategies among different building typologies, but not across different climate zones. The insights gained from this work could help policy makers propose optimal climate-dependent ranges of thermostat setpoints for buildings located in different climate zones. Table 3 One-way ANOVA for decision variables (setpoints)

Variable

F-statistic

p-value

Occupied cooling setpoint

14.01

1.41e−05

Occupied heating setpoint

0.86

0.4291

Unoccupied cooling setpoint

14.78

8.65e−06

Unoccupied heating setpoint

32.64

7.47e−10

Multi-objective Genetic Algorithm Optimization of HVAC …

275

4 Conclusions, Limitations, and Future Work This chapter presented a simulation-based MOO GA framework to identify optimal HVAC operation strategies. The MOO formulation accounts for three building performance metrics (energy consumption, thermal comfort, and productivity), providing a detailed understanding of the trade-offs between the competing objectives. The decision variables used in this chapter include a spectrum of cooling and heating HVAC setpoints. The setpoints covered both occupied and unoccupied building period, resulting in a wide solution search space. After a thorough discussion of the proposed optimization methodology, the framework was applied to three prototype office buildings located in Phoenix, AZ, Baltimore, MD, and Minneapolis, MN. The analysis of the obtained Pareto fronts validated the high potential of humanbased retrofits as energy conservation measures. Energy savings of 25.8% (Baltimore), 12.3% (Phoenix), and 14% (Minneapolis) can be achieved through optimal HVAC operation, without compromising occupants’ thermal comfort and productivity levels. Moreover, the diversity of the Pareto front can be used by decision makers who can set constraints on the objectives in the solution space and choose among the non-dominated solutions that satisfy their conditions. Finally, a statistical analysis using one-way ANOVA proves that the current practice of using the same setpoint ranges for buildings in different climates is not optimal. Important improvements in building energy consumption, thermal comfort, and productivity can be achieved through weather-dependent HVAC strategies, such as the ones demonstrated in this work. An important shortcoming of the current work is the rather simplistic representation and modeling of productivity, which was guided by an experimental study from the literature [25]. In reality, peoples’ work performance is a highly complex phenomenon to study as it can be affected by numerous other factors beyond environmental conditions. Therefore, simplifying it to a single formula (see Eq. 1) might not be capturing the full picture. Moreover, the results of Kosonen and Tan [25] might not be generalizable in different testing conditions. As online learning advances and Internet of Things (IoT) penetrate commercial buildings, researchers are accessing new data streams that allow to draw a more nuanced picture on occupants’ comfort and productivity levels [8, 23, 34]. Nonetheless, and despite its limitations, the abovementioned study helped demonstrate the capabilities of the proposed GA MOO framework, which was the main goal of this chapter. As part of future work, the authors plan to apply the framework on data collected from a large number of buildings to validate the findings of the current study. Important efforts are being deployed in the literature to develop large datasets of indoor environmental conditions and how occupants perceive and react to them. One such effort is the “ASHRAE Global Thermal Comfort Database II” [17], an open-source database gathering and combining data from a large number of studies from around the world. Future expansions of the current work could also include a higher temporal granularity of analysis (e.g., hourly), which can help devise more targeted operation strategies. Additionally, further occupant-related energy conservation actions could be

276

S. Papadopoulos and E. Azar

integrated into the optimization process. For instance, window opening frequency or reducing lighting and equipment usage could be studied along with HVAC operation, resulting in a more holistic and occupant-centric retrofit optimization framework.

References 1. ASHRAE (2013a) ASHRAE 90.1-2013, energy standard for buildings except low rise residential buildings, vol 2013 2. ASHRAE (2013b) ASHRAE 55: thermal environmental conditions for human occupancy. In: American Society of Heating, Refrigerating and Air-Conditioning Engineers (ed.). American Society of Heating, Refrigerating and Air-Conditioning Engineers 3. Al Amoodi A, Azar E (2018) Impact of human actions on building energy performance: a case study in the United Arab Emirates (UAE). Sustainability (Switzerland) 10(5). https://doi.org/ 10.3390/su10051404 4. Andargie MS, Azar E (2019) An applied framework to evaluate the impact of indoor office environmental factors on occupants’ comfort and working conditions. Sustain Cities Soc 46:101447 5. Azar E, Menassa CC (2014) A comprehensive framework to quantify energy savings potential from improved operations of commercial building stocks. Energy Policy 67:459–472. https:// doi.org/10.1016/j.enpol.2013.12.031 6. Barbosa JD, Azar E (2018) Modeling and implementing human-based energy retrofits in a green building in desert climate. Energy Build 173:71–80 7. Brager GS, de Dear R (2001) Climate, comfort & natural ventilation: a new adaptive comfort standard for ASHRAE Standard 55. University of California. https://escholarship.org/uc/item/ 2048t8nn 8. Cosma AC, Simha R (2019) Machine learning method for real-time non-invasive prediction of individual thermal preference in transient conditions. Build Environ 148:372–383 9. Crawley DB, Hand JW, Kummert M, Griffith BT (2008) Contrasting the capabilities of building energy performance simulation programs. Build Environ 43(4):661–673 10. DOE (2018) Commercial prototype building models. United States Department of Energy (DOE). https://www.energycodes.gov/development/commercial/prototype_models 11. de Dear R, Brager GS (1998) Developing an adaptive model of thermal comfort and preference. ASHRAE Trans 104(1):145–167 12. Deb K (2001) Multi-objective optimization using evolutionary algorithms: an introduction. Wiley 13. Deb K, Pratap A, Agarwal S, Meyarivan T (2002) A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE Trans Evol Comput 6(2):182–197 14. Deru M, Field K, Studer D, Benne K, Griffith B, Torcellini P, Liu B, Halverson M, Winiarski D, Rosenberg M, National Renewable Energy Laboratory (NREL) (2011) U.S. Department of Energy commercial reference building models of the national building stock (issue February 2011). National Renewable Energy Laboratory (NREL). https://www.nrel.gov/docs/fy11osti/ 46861.pdf 15. Diakaki C, Grigoroudis E, Kolokotsa D (2008) Towards a multi-objective optimization approach for improving energy efficiency in buildings. Energy Build 40(9):1747–1754. https:// doi.org/10.1016/j.enbuild.2008.03.002 16. Fanger PO (1970) Thermal comfort. Analysis and applications in environmental engineering. Danish Technical Press 17. Földváry Liˇcina V, Cheung T, Zhang H, de Dear R, Parkinson T, Arens E, Chun C, Schiavon S, Luo M, Brager G, Li P, Kaam S, Adebamowo MA, Andamon MM, Babich F, Bouden C, Bukovianska H, Candido C, Cao B, Zhou X (2018) Development of the ASHRAE global

Multi-objective Genetic Algorithm Optimization of HVAC …

18.

19. 20.

21. 22.

23. 24.

25. 26. 27. 28. 29.

30. 31.

32.

33.

34. 35.

36. 37.

277

thermal comfort database II. Build Environ 142:502–512. https://doi.org/10.1016/j.buildenv. 2018.06.022 Ghahramani A, Zhang K, Dutta K, Yang Z, Becerik-Gerber B (2016) Energy savings from temperature setpoints and deadband: quantifying the influence of building and system properties on savings. Appl Energy 165:930–942 Goldberg DE, Holland JH (1988) Genetic algorithms and machine learning. In: Machine learning. Kluwer Academic Publishers, pp 95–99. https://doi.org/10.1023/A:1022602019183 Hoyt T, Arens E, Zhang H (2015) Extending air temperature setpoints: simulated energy savings and design considerations for new and retrofit buildings. Build Environ 88:89–96. https://doi. org/10.1016/j.buildenv.2014.09.010 IEA (2019) Energy efficiency 2019. International Energy Agency (IEA). https://webstore.iea. org/download/direct/2891 ISO (2005) ISO 7730: ergonomics of the thermal environment—analytical determination and interpretation of thermal comfort using calculation of the PMV and PPD indices and local thermal comfort criteria. International Standard Organization Jung W, Jazizadeh F (2019) Human-in-the-loop HVAC operations: a quantitative review on occupancy, comfort, and energy-efficiency dimensions. Appl Energy 239:1471–1508 Klepeis NE, Nelson WC, Ott WR, Robinson JP, Tsang AM, Switzer P, Behar JV, Hern SC, Engelmann WH (2001) The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J Expo Anal Environ Epidemiol 11(3):231– 252. https://doi.org/10.1038/sj.jea.7500165 Kosonen R, Tan F (2004) Assessment of productivity loss in air-conditioned buildings using PMV index. Energy Build 36(10):987–993. https://doi.org/10.1016/j.enbuild.2004.06.021 Lakeridou M, Ucci M, Marmot A, Ridley I (2012) The potential of increasing cooling set-points in air-conditioned offices in the UK. Appl Energy 94:338–348 Lausten J (2008) Energy efficiency requirements in building codes, energy efficiency policies for new buildings. IEA information paper. International Energy Agency (IEA) MathWorks (2011) Global optimization toolbox. User’s guide R2011b. The MathWorks Inc. Nguyen AT, Reiter S, Rigo P (2014) A review on simulation-based optimization methods applied to building performance analysis. In: Applied energy, vol 113. Elsevier Ltd, pp 1043– 1058. https://doi.org/10.1016/j.apenergy.2013.08.061 Nicol JF, Humphreys MA (2002) Adaptive thermal comfort and sustainable thermal standards for buildings. Energy Build 34(6):563–572. https://doi.org/10.1016/S0378-7788(02)00006-3 O’Brien W, Tahmasebi F, Andersen RK, Azar E, Barthelmes V, Belafi ZD, Berger C, Chen D, De Simone M, d’Oca S, Hong T, Jin Q, Khovalyg D, Lamberts R, Novakovic V, Park JY, Plagmann M, Rajus VS, Vellei M, Zhou J (2020) An international review of occupant-related aspects of building energy codes and standards. Build Environ 179:106906. https://doi.org/10. 1016/j.buildenv.2020.106906 Papadopoulos S, Kontokosta CE, Vlachokostas A, Azar E (2019) Rethinking HVAC temperature setpoints in commercial buildings: the potential for zero-cost energy savings and comfort improvement in different climates. Build Environ 155(April):350–359. https://doi.org/10.1016/ j.buildenv.2019.03.062 Papadopoulos S, Azar E (2016) Optimizing HVAC operation in commercial buildings: a genetic algorithm multi-objective optimization framework. In: Proceedings—winter simulation conference. https://doi.org/10.1109/WSC.2016.7822220 Peng Y, Nagy Z, Schlüter A (2019) Temperature-preference learning with neural networks for occupant-centric building indoor climate controls. Build Environ 154:296–308 Pisello AL, Asdrubali F (2014) Human-based energy retrofits in residential buildings: a costeffective alternative to traditional physical strategies. Appl Energy 133:224–235. https://doi. org/10.1016/j.apenergy.2014.07.049 Reeves CR, Rowe JE (2002) Genetic algorithms—principles and perspectives. Kluwer Academic Publishers. https://doi.org/10.1007/b101880 Seppanen O, Fisk W (2006) Some quantitative relations between indoor environmental quality and work performance or health. HVAC&R Res 12(4):957–973. https://doi.org/10.1080/107 89669.2006.10391446

278

S. Papadopoulos and E. Azar

38. Thatcher A, Milner K (2016) Is a green building really better for building occupants? A longitudinal evaluation. Build Environ 108:194–206. https://doi.org/10.1016/j.buildenv.2016. 08.036 39. UNEP (2007) Buildings can play a key role in combating climate change. United Nations Environment Programme (UNEP). www.sciencedaily.com/releases/2007/04/070407150947. htm