Furthermore, these research papers were analyzed and categorized according to their research methodologies for each field of ECLO. The most preferred research method on SSCM emerged from quantitative modeling papers, where 100 out of 217 papers employed quantitative modeling and case study, followed by conceptual/theoretical model papers. Furthermore, the major research methodology implemented in the GSCM papers also quantitative modeling involving 176 out of 282 papers. Similarly, 120 out of 121 papers are found that apply quantitative modeling to CLSCM where hypothetical models constituted the majority of these studies. These results show that, SSCM concept is mainly discussed over its conceptual perspective, while GSCM and CLSCM are primarily studied by quantitative modeling methodologies.
The remainder of this section focuses on the classification of quantitative models, where the fuzzy-based models and decision making models are detailed.
4.2.1. Quantitative Models
In this paper, quantitative modeling techniques include operations research tools such as mathematical programming, decision analysis, heuristics, simulation, and others. Conversely, statistical models that include structural equation modeling, Delphi study, hierarchical linear modeling, regression analysis, and the Taguchi method were excluded from the quantitative model class.
provides a basic framework of the analysis and depicts each model with its corresponding sub-classes [3
In total, 425 out of 707 papers were labeled as quantitative modeling papers. These papers were then separated based on their designated class in accordance with the constructed framework (Table A2
in Appendix A
Using the framework provided in Figure 13
, mathematical models were categorized into single-objective models and multi-objective optimization (MOP) models. These models include linear programming (LP), nonlinear programming (NLP), mixed integer programming (MIP), mixed integer linear programming (MILP), goal programming (GP), robust programming (RP), stochastic programming (SP), dynamic programming (DP), possibilistic programming, queuing theory, fuzzy mathematical programming, and bi-objective programming [58
Typically, in quantitative modeling techniques, heuristics are combined with other methodologies. Within this research, it is also determined that heuristics present a viable solution methodology when used with additional methods. Furthermore, heuristics can be categorized under two main classes referred as exact heuristics and meta-heuristics. In addition, neural network (NN) models can also be included under the umbrella of heuristics [3
]. With this reasoning, this paper categorized heuristics under three classes as follows: exact heuristics, meta-heuristics, and NN.
It is also observed that limited number of papers discussed simulation models. In this regard, system dynamics is included in 3 papers [76
], whereas discrete event simulation models and Monte Carlo simulation models are utilized in 4 and 3 papers, respectively [79
Various decision analysis methods include multi-criteria decision making, fuzzy set theory, rough set theory, game theory, grey systems, and life cycle analysis. For the purpose of the content analysis, papers that addressed sustainability through the use of fuzzy logic were considered. When various papers on decision analysis were compared, it is found that several papers cross-referenced fuzzy set theory and rough set theory with artificial intelligence techniques [3
]. Additionally, various papers discuss fuzzy set theory as an individual methodology in decision making [85
], while the vast majority of papers integrates fuzzy approaches into multi-criteria decision making (MCDM) [88
]. This paper examines fuzzy sets as one of the decision analysis tools, and divided these approaches into the sub-techniques depicted in Figure 14
Here, out of the 156 quantitative modeling papers that employ fuzzy techniques only 9 of them focus on fuzzy sets as the only method. The remaining papers propose hybrid approaches that combine the theory with various modeling techniques such as MCDM, mathematical modeling, rough sets, grey systems, and game theory. In this regard, the classification of these papers depends upon on each technique. It is also found that the majority of papers using fuzzy approaches is primarily correlated with MCDM, followed by mathematical programming. More specifically, the fuzzy MCDM approach constitutes 96 out of the 147 fuzzy papers, while fuzzy mathematical programming is used in 53 out of the 147 fuzzy papers. The remaining papers discuss other fuzzy related methods such as fuzzy rough sets, fuzzy game theory, and fuzzy grey systems. In an effort to further understand the relationship between fuzzy sets and its utilization in ECLO fields and subfields, a relationship matrix is generated as illustrated in Figure 15
. Moreover, Table A3
) further explains the classifications of each ECLO topic and the applied industry in focus. A detailed analysis of fuzzy focused approaches is provided in Section 4.2.2
An analysis of the research gaps in fuzzy-based literature is provided in the following.
4.2.2. Fuzzy Set Theory
] defined fuzzy set theory as “a class of objects with a continuum of grades of membership in which where human-thinking plays an important role, whereby uncertainty in the classes of objects is taken into consideration by using the linguistic terms and membership functions”. Olugu and Wong [96
] defined fuzzy set theory as a knowledge and concept utilization process based on human reasoning. The literature presented a variety of fuzzy focused papers that introduced fuzzy logic as defining membership functions, incomplete preference relationships and linguistic preferences, applying fuzzy arithmetic, fuzzy entropy and fuzzy c-means clustering, utilizing intuitionistic fuzzy sets, fuzzy logic controllers, defining interpretive ranking processes, fuzzy inference systems, fuzzy rule-based systems, and fuzzy axiomatic design (Table A2
Furthermore, Amindoust et al. [85
] studied the sustainable supplier selection problem through a new ranking model based on fuzzy inference system. In their paper, the degree of importance of supplier selection criteria and sub-criteria were evaluated with regards to the decision makers’ opinion. Similarly, Ghadimi and Heavey [87
] investigated a sustainable supplier evaluation and selection model by using fuzzy inference systems. The approach that Humphreys et al. [86
] used to examine the supplier assessment process was to implement a user-centered hierarchical fuzzy membership functions with a focus on environmental criteria. In addition, Olugu and Wong [96
] presented a performance evaluation system for CLSCM by processing an expert fuzzy rule-based model using Visual Basic.Net. Govindan and Murugesan [101
] proposed a fuzzy extent analysis on a 3PL reverse logistics provider selection problem. Another approach was used by Kannan et al. [102
] was a fuzzy axiomatic design (FAD) approach in order to select the best green supplier in the system.
a,b depict the top ten journals and publication years of fuzzy approach ECLO papers, respectively. Here, the most preferred journal is the Journal of Cleaner Production (JCLP), where the amount of papers on fuzzy-related ECLO topics remain stable in numbers until 2005. Then, the ECLO articles increased in growth in the beginning of 2010 reaching their peak in 2016 due to the increased attention of environmental concerns.
Fuzzy Mathematical Programming
Fuzzy mathematical modeling papers include both single-objective and multi-objective models wherein fuzzy LP/NLP, fuzzy MIP/MILP/MINLP, fuzzy RP, fuzzy GP, fuzzy DP, fuzzy MOP/MOLP, fuzzy bi-objective programming, and fuzzy SP are applied. Possibilistic programming is also identified as a fuzzy mathematical programming approach where possibilistic distributions are implemented to define the model parameters [103
]. In the literature, the manuscripts which include fuzzy mathematical modeling constitute 53 out of the 156 fuzzy-based articles.
Among these, Shaw et al. [89
] combined a hybrid method using fuzzy AHP and fuzzy MOLP to solve the supplier selection problem in order to optimize a low-carbon logistics network. Similarly, Kannan et al. [104
] applied a combination of fuzzy AHP, fuzzy TOPSIS, and fuzzy MOLP models to identify the green supplier selection and order allocation problem. Furthermore, Subulan et al. [105
] studied a multi-echelon, multi-product, and multi-period tire closed-loop supply chain network design model using multi-objective interactive fuzzy goal programming method. When Vahdani et al. [75
] constructed a reliable CLSC network design model, they used a hybrid method of robust optimization and bi-objective fuzzy queuing mixed integer linear programming. The following year, Vahdani et al. [106
] studied the reliable CLSC network design model by implementing a bi-objective fuzzy possibilistic-queuing mixed integer linear programming model. In addition, Amin and Zhang [107
] proposed a multi-objective mixed integer linear programming and fuzzy set theory in order to optimize the CLSC and the supplier selection process.
Fuzzy Multi-Criteria Decision Making
Multi-criteria decision making (MCDM) is known as an operations research method that evaluates multiple alternatives in order to obtain meaningful results in a complex decision making environment. Various MCDM approaches include Analytic Hierarchy Process/Analytic Network Process (AHP/ANP), Decision Making Trial and Evaluation Laboratory (DEMATEL), Technique for Order Performance by Similarity to Ideal Solution (TOPSIS), Vlse Kriterijumska Optimizacija Kompromisno Resenje’ (VIKOR) (the English translation of VIKOR is Methodology of Multi-criteria Optimization and Compromise Solution), Data Envelopment Analysis (DEA), Preference Ranking Organization method for enrichment evaluations (PROMETHEE), Measuring Attractiveness by a Categorical Based Evaluation Technique (MACBETH), Multi-attribute Utility Theory (MAUT), and weighted product/sum model. It is found that the papers that discussed various MCDM approaches constitute 181 out of 425 quantitative modeling papers, where 132 out of the 181 papers integrated other modeling approaches such as fuzzy sets, mathematical programming, heuristics, and/or simulation. Table A4
) represents the classification of each MCDM approach with regards to ECLO topics and its subfields. As seen in Table A4
, the most utilized MCDM method is AHP/ANP, while the most studied subfield is supplier selection/evaluation.
In the reviewed literature, fuzzy sets are mostly integrated into the multi-criteria decision making approaches [23
]. Here, the fuzzy-involved MCDM methods contain fuzzy ANP/AHP, fuzzy DEMATEL, fuzzy TOPSIS, fuzzy VIKOR, fuzzy DEA, fuzzy PROMETHEE, and fuzzy MACBETH. The total number of papers that applied fuzzy MCDM techniques constitutes 97 out of 156 fuzzy papers. Therefore, fuzzy sets are mainly integrated into MCDM. Furthermore, upon additional analysis, 54 out of the 97 fuzzy MCDM papers preferred the fuzzy AHP/ANP approach, 22 out of the 97 papers used fuzzy TOPSIS approach, and 13 out of the 97 papers followed fuzzy DEMATEL approach. The remaining 13 papers contain other fuzzy MCDM methods such as fuzzy VIKOR (5), fuzzy DEA (5), fuzzy PROMETHEE (2) and fuzzy MACBETH (1). It should be noted that the inequalities between total number of papers on fuzzy MCDM is based on the hybrid studies that combines two or more MCDM approaches.
Two commonly used MCDM tools are AHP and ANP methods. In the literature, these methods preferred in evaluating qualitative data when compared to the other models such as mathematical programming [65
]. Saaty [108
] introduced AHP model that enables the decision maker to integrate and evaluate qualitative and quantitative criteria and sub-criteria by building a one-way hierarchical structure among decision levels. ANP [109
], which is an extension of AHP, contains a network structure that deals with more complex relationship by disregarding a strict hierarchical structure [91
]. In other words, AHP assumes independence between each criterion, whereas ANP introduces dependence and feedback among the criteria [90
]. In the quantitative literature, the majority of papers on AHP/ANP methods are combined with fuzzy sets since fuzzy set theory incorporates the uncertainty of human judgement into the decision-making process for more effective and realistic modelling [112
]. Moreover, fuzzy AHP/ANP models help detect environmental criteria weights where the majority of environmental criteria involves linguistic terms and imprecise judgements [111
As for the distribution of these manuscripts, 54 out of the 97 papers employ fuzzy AHP/ANP method. While AHP is the most preferred method in the ECLO subfields compared to ANP, both of the approaches are the leading methods of MCDM. Specifically, AHP is a widely used MCDM technique, where it is integrated as an additional methodology into the majority of other MCDM models. More specifically, it is observed that the majority of the papers on supplier selection/evaluation in the literature applied either solely AHP/ANP or AHP/ANP integrated hybrid approaches, which is an evident that it is the best suitable approach on solving supplier selection/evaluation models. Efendigil et al. [114
] investigated a holistic approach based on ANN and fuzzy AHP (FAHP) methods in order to select the most appropriate and desirable third-party reverse logistics service provider in consideration of various subjective requirements. Shaw et al. [89
] presented a hybrid approach based on FAHP and fuzzy MOLP in a supplier selection problem for a low-carbon supply chain. Similarly, Kannan et al. [104
] developed an integrated approach based on FAHP and MOP for a green supplier selection and order allocation model for a multiple sourcing problem.
Büyüközkan and Çifçi [88
] presented a novel approach based on fuzzy ANP under incomplete preference relations for an effective sustainable supplier selection problem. Their proposed solution was applied in a white good industry case study. Büyüközkan and Çifçi [90
] further published a case study on green supplier evaluation in the automobile industry by proposing a hybrid fuzzy MCDM approach integrating fuzzy DEMATEL, ANP, and TOPSIS. Additionally, Shakourloo et al. [115
] studied a supplier selection model in a closed-loop supply chain by combining fuzzy AHP and multi-objective mixed integer linear programming (MOMILP) methods, where they evaluated the proper third-party logistics provider.
The Technique for Order Performance by Similarity to Ideal Solution (TOPSIS) was first introduced by Hwang and Yoon [116
]. TOPSIS is a tool for selecting optimal solutions from a finite set of alternatives that have the shortest distance from the positive ideal solution and the farthest distance from the negative ideal solution. Here, the positive ideal solution refers to the best performance values and the negative ideal solution refers to the worst performance values [90
]. A hybrid fuzzy TOPSIS method evaluates the linguistic data based on the criteria weights for imprecision, subjectivity and vagueness in order for the data to be set into fuzzy numbers [90
The survey identified fuzzy TOPSIS as the second most utilized method within MCDM with 22 out of the 81 papers implementing the methodology. Another interesting finding was that, even though TOPSIS is also effective as a stand-alone methodology, it was mostly combined with AHP and/or ANP. In addition, similar to AHP/ANP, TOPSIS method was often applied to model supplier selection/evaluation problems. For instance, Fallahpour et al. [118
] developed a hybrid fuzzy AHP and TOPSIS model for sustainable supplier selection in order to improve the performance of sustainable textile supply chain. Similarly, Prakash and Barua [119
] studied a third-party reverse logistics provider selection by employing a hybrid fuzzy AHP and TOPSIS approach. Moreover, Awasthi et al. [92
] proposed a fuzzy TOPSIS approach to evaluate the environmental performance of suppliers. In an effort to solve for a green supplier selection problem within an electronics company, Kannan et al. [9
] created a framework on the criteria of green supply chain practices using fuzzy TOPSIS. Govindan et al. [120
] employed fuzzy TOPSIS to measure sustainability performance based on the triple bottom line approach that consists of four environmental criteria, four economic criteria, and four social criteria for supplier evaluation in a sustainable supply chain.
Decision making trial and evaluation laboratory (DEMATEL) was employed by Gabus and Fontela [121
] and Gabus and Fontela [122
] to build a causal diagram of interdependent complex factors through the formulated relationships between causes and effects in a comprehensive structural model [113
]. In addition to the DEMATEL approach, the hybrid fuzzy DEMATEL addresses the flexibility issue of the fuzziness in order to obtain accurate and reliable results [123
The proposed literature survey found that 13 out of the 97 papers utilized the fuzzy DEMATEL method. Compared to AHP/ANP and TOPSIS methods, DEMATEL is less favorable approach within MCDM techniques. Lin et al. [125
] used fuzzy DEMATEL for a green supply chain performance evaluation to improve environmental image and to present a competitive advantage in the automobile manufacturing industry. Lin [123
] implemented fuzzy DEMATEL to evaluate the criteria for green supply chain practices that included practices, performance, and external pressures. Wu et al. [124
] implemented fuzzy DEMATEL to identify the factors that affected an automobile company’s green supply chain performance. In order to assess supply chain risks and uncertainties towards sustainability in an electronics supply chain, Wu et al. [126
] employed a novel method based on big data analysis, fuzzy DEMATEL and grey DEMATEL techniques. With regards to this, they identified a set of attributes over fuzzy Delphi and grey Delphi techniques and transformed big data to a manageable scale in order to examine their impacts. At the last stage of the model, they implemented fuzzy DEMATEL and grey DEMATEL methods to evaluate the greater risk factors.
Other Fuzzy MCDM Approaches
VIKOR, first proposed by Opricovic [127
], is the method to sequence and select solutions from a set of alternatives with conflicting criteria. This method provides a more comprehensive evaluation in a fuzzy environment where the evaluation is based upon the means of linguistic terms [98
]. Total number of papers which solely apply VIKOR technique is pretty rare, and this method is majorly integrated with other MCDM techniques as a solution technique. Furthermore, similar to other MCDM methods, a greater number of VIKOR-employed MCDM papers encompasses supplier selection and performance evaluation models. In this regard, Banaeian et al. [129
] utilized a hybrid method based on fuzzy TOPSIS, VIKOR, and GRA in order to identify the best supplier within a green agri-food supply chain. Similarly, Prakash and Barua [130
] presented a combined model of fuzzy AHP and VIKOR to evaluate and select the best third party logistics provider for an Indian electronics manufacturer. Rostamzadeh et al. [53
] applied fuzzy VIKOR in a case study for a Malaysian laptop manufacturer to measure the uncertainty of green supply chain activities, to evaluate green supply chain practices, and to select the best green supply chain practitioner. In addition, Awasthi and Kannan [131
] proposed an integrated method based on fuzzy nominal group technique, (NGT)-VIKOR, in order to evaluate and select the best green supplier development programs. Here, fuzzy NGT assigned linguistic ratings to each alternative and criterion, and then fuzzy VIKOR generated the program rankings and introduced the best solutions. Similarly, Akman [98
] addressed the evaluation of green supplier development programs through the confirmatory factor analysis and fuzzy c-means-VIKOR in order to determine the environmental performance of suppliers in the automobile industry.
A nonparametric approach called Data Envelopment Analysis (DEA) utilizes linear programming to evaluate the relative efficiencies of a set of decision making units which convert multiple inputs to multiple outputs [17
]. Despite the fact that DEA models designate the relative efficiency evaluation of decision making units, solely DEA models can be weak to assess the effectiveness of input data in case of lack of input data noise or insufficient input information. Therefore, incorporating DEA and fuzzy sets would deal with the uncertain information in a proper manner. With regards to further analysis on this method, total number of papers that employ DEA technique sharply increased during the last 5-year period, which shows it is still an emerging modelling technique in the literature. Moreover, it is found that a significant amount of DEA based papers applies artificial neural networks as a solution approach. Azadi et al. [134
] utilized a comprehensive fuzzy DEA method to solve the problem of sustainable supplier selection in a resin production company based on environmental, economic, and social dimensions. Mirhedayatian et al. [133
] proposed a DEA model that incorporated dual factors, undesirable outputs, and fuzzy data in order to evaluate GSCM performance. Additionally, Fallahpour et al. [135
] developed a hybrid DEA-Genetic Programming (GP) approach to evaluate and select the best green supplier, where they referred to adaptive neuro-fuzzy inference system. Similarly, Zhou et al. [136
] utilized a type-2 multi-objective DEA model in order to evaluate the most appropriate suppliers within SSCM, where they select the best supplier in case of the balance of economic, environmental, and social dimensions.
Preference Ranking Organization Method for Enrichment Evaluation (PROMETHEE) by Brans et al. [137
] is a MCDM technique that allows decision makers to rank a finite number of alternatives in accordance with various criteria to indicate the relative importance of preferred function. It is a simple method compared to other MCDM techniques [17
], and no study found that uses PROMETHEE as a single approach. This method is not only rarely studied in the literature, but it is supported by an additional MCDM method, specifically AHP/ANP. Tuzkaya et al. [111
] utilized a novel fuzzy ANP and fuzzy DEMATEL approach in order to evaluate suppliers’ environmental performance in a Turkish white goods case study. Similarly, Tuzkaya [138
] developed a hybrid fuzzy AHP-DEMATEL method to evaluate and rank the environmental effects of five alternative transportation modes in a specific region among nine criteria.
Measuring Attractiveness by a Categorical Based Evaluation Technique (MACBETH) by Bana e Costa and Vansnick [139
] is a ranking method that helps quantify the relative attractiveness of each alternative. This method is similar to AHP even though MACBETH uses an interval scale, while AHP uses a ratio scale [17
]. It can be said that MACBETH is the least favorable MCDM technique in the literature. Dhouib [140
] examined an extension of the MACBETH technique in order to assess various reverse logistics options for used automobile tires.
Other Fuzzy-Integrated Approaches
Few studies involve fuzzy-integrated techniques such as fuzzy rough sets, fuzzy game theory, and fuzzy grey systems. Grey system, a theory derived from grey sets, is a system which copes with both known and unknown information. Grey systems involve five major approaches that include: grey prediction, grey relational analysis, grey decision, grey programming, and grey control [94
]. Total number of fuzzy ECLO articles that incorporate with grey theory increased in the last 5-year period. This hybrid method helps to improve insufficient information, so that overcome uncertainty. Bali et al. [94
] studied a green supplier selection problem for an automobile company based on an integrated method of intuitionistic fuzzy set (IFS) and grey relational analysis (GRA). Here, IFS was used to calculate the weights of criteria, where GRA obtained the most appropriate alternative supplier by ranking all alternatives. Moreover, Tseng and Chiu [143
] applied a hybrid fuzzy sets and GRA approach to evaluate a Taiwanese printed circuit board manufacturing company’s GSCM in order select the suitable green supplier among four suppliers. Wu et al. [144
] evaluated the performance of SSCM by combining interval-valued triangular fuzzy numbers with grey relational analysis.
Rough set theory is a mathematical methodology that measures the vagueness, impreciseness, and ambiguity of data. It can be considered as an alternative to fuzzy sets [145
]. Kusi-Sarpong et al. [146
] evaluated green supply chain practices in the mining industry through a combined rough sets and fuzzy TOPSIS approach. Bai et al. [147
] proposed a novel hybrid rough set theoretic and fuzzy clustering means technique for green supplier development to help the organizations manage a thorough and rigorous investment analysis.
Game theory, as defined by Myerson [148
], is the study of mathematical models of conflict and cooperation between intelligent rational decision-makers. The theory was first proved by Neumann and Morgenstern [149
]. As Zhao et al. [150
] also pointed out, game theory is an essential tool in SCM, and its applications in GSCM is still under development. Wei and Zhao [151
] combined game theory and fuzzy set theory to solve an optimal pricing decision problem in CLSCM. Additionally, Wei and Zhao [152
] implemented a fuzzy theory and game theoretic approach to further focus on the decisions of reverse channel choice in CLSCM. Yang and Xiao [153
] utilized Stackelberg scenario analysis with fuzzy degree parameters in order to assess GSCM towards governmental interventions.