Sustainable Supply Chain Network Design

Omega 54 (2015) 11–32
Contents lists available at ScienceDirect
Omega
journal homepage: www.elsevier.com/locate/omega
Review
Sustainable supply chain network design:

An optimization-oriented review$
Majid Eskandarpour a, Pierre Dejax a, Joe Miemczyk b,c, Olivier Péton a,n
a
Ecole des Mines de Nantes, IRCCyN UMR CNRS 6597 (Institut de Recherche en Communications et Cybernétique de Nantes), 4 rue Alfred Kastler, F-44307
Nantes, France
b
Audencia Nantes School of Management, 44312 Nantes, France
c
Aix Marseille Université, CRET-LOG EA 881, 13625, Aix en Provence, France
art ic l e i nf o a b s t r a c t
Article history: Supply chain network design (SCND) models and methods have been the subject of several recent
Received 23 April 2014 literature review surveys, but none of them explicitly includes sustainable development as a main
Accepted 14 January 2015 characteristic of the problem considered. The aim of this review is to bridge this gap. The paper analyzes
Available online 24 January 2015
87 papers in the field of supply chain network design, covering mathematical models that include
Keywords: economic factors as well as environmental and/or social dimensions. The review is organized along four
Supply chain management research questions asking (i) which environmental and social objectives are included, (ii) how are they
Network design integrated into the models, (iii) which methods and tool are used and finally (iv) which industrial
Location applications and contexts are covered in these models. The review finds that there are a number of
Sustainable development
limitations to the current research in sustainable SCND. The narrow scope of environmental and social
combinatorial optimization
measures in current models should go beyond limited greenhouse gas indicators to broader life-cycle
approaches including new social metrics. The more effective inclusion of uncertainty and risk in models
with improved multi-objective approaches is also needed. There are also significant gaps in the sectors
used to test models limiting more general applicability. The paper concludes with promising new
avenues of research to more effectively include sustainability into SCND models.
& 2015 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2. Review methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1. Delimitations and search for literature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2. Position in the literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3. Distribution across the time period and main journals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4. The 3 dimensions of sustainable development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3. Environmental supply chain network design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1. LCA based models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.1. Scope definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1.2. Life-cycle impact assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.1.3. LCIA methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.4. Impact categories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2. Partial assessment of environmental factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2.1. Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2.2. Performance measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
☆
This manuscript was processed by Associate Editor Campbell.
n
Corresponding author. Tel.: þ 33 251 858 313; fax: þ 33 251 858 349.
E-mail addresses: majid.eskandarpour@mines-nantes.fr (M. Eskandarpour),
pierre.dejax@mines-nantes.fr (P. Dejax), jmiemczyk@audencia.com (J. Miemczyk),
olivier.peton@mines-nantes.fr (O. Péton).
http://dx.doi.org/10.1016/j.omega.2015.01.006
0305-0483/& 2015 Elsevier Ltd. All rights reserved.
12 M. Eskandarpour et al. / Omega 54 (2015) 11–32
3.3. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4. Social supply chain network design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5. Modeling approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.1. Models with a single objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.1.1. Deterministic models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.1.2. Stochastic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2. Multi-objective models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2.1. Deterministic models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2.2. Stochastic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.3. Conclusions on modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6. Solution methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.1. Solution methods for models with a single objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.2. Solution methods for multi-objective models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.2.1. Weighted sum of objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.2.2. Epsilon-constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.2.3. Metaheuristics for multi-objective models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
6.2.4. Multi-criteria decision analysis and interactive methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
6.2.5. Other methods and hybrid approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
6.3. Modeling tools and solvers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
6.4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
7. Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
7.1. Intersectorial analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
7.2. Conclusion on applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
8. General conclusions and research directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
8.1. Summary of findings and contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
8.2. Research avenues. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Further references . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1. Introduction
of the facility location problem, which has become one of the standard
problems in the operations research community. Recent years have
Supply chain management has become a strategic issue for any
company looking to meet targets in terms of economic competitive- been characterized by a rapid enrichment of these mathematical
ness, time and quality of service especially in an economic environ- model solutions. Rich models now handle multiple levels in the
ment characterized by the globalization of trade and the acceleration logistics network, multiple periods, products, technologies, transporta-
of industrial cycles. The trade press is replete with examples of logistics tion modes and types of facilities. They integrate capacity constraints,
network configuration, re-configuration, re-organization, mergers, out- tactical decisions and complex product flows. Thus, Supply Chain
sourcing, and so on. These developments have been influenced by Network Design (SCND) can be considered as the meeting point of the
successive trends in the economy and society resulting from compu- academic facility location problem and the real-life SCND problem.
terization, increased complexity of trade flows, increased competition As already mentioned, among the major trends in SCM, the
and certainly not least, sustainable development. Thus the strategic principles of sustainable development have spread across the scientific
design and planning of logistics networks is a topic that is becoming literature. Current research mainly consists of assessing SCM policies
more important for businesses and researchers alike. Supply chain according to a triple bottom line including economic aspects, environ-
network design is at the intersection of disciplines such as manage- mental performance and social responsibility. Sustainable SCM has
ment, strategy, logistics, operations research and as such, there is a been the subject of numerous survey papers in both qualitative and
significant challenge to researchers to consolidate and synthesize the quantitative disciplines. A number of review papers have been
research in this field, which leads to the focus of this paper. published in recent years, which relate to major trends in Supply
Supply chain management (SCM) spans all movements and storage Chain Management and investigate and suggest research opportu-
of raw materials, work-in-process inventory, and finished goods from nities. Importantly, research in sustainable SCND has hardly been
the point-of-origin to the point-of-consumption [153]. It encompasses reviewed at all. However, the integration of sustainability into SCND
three decision levels: strategic, tactical and operational. In particular, at may change the locations of production facilities and inventories (see
the strategic level, supply chain design comprises the decisions for example You and Wang [81, Figs. 12–14]) and therefore have a
regarding the number and location of production and storage facilities, significant impact on the environment and society. The goal of this
the amount of capacity at each facility, the conciliation of market paper is to bridge this gap.
demand, and decisions on supplier selection from a total cost More precisely, our objective is to review SCND problems that
perspective [107]. From an operations research point of view, Supply include a clear assessment of at least two of the three dimensions of
Chain Network Design (SCND) is the discipline used to determine the sustainable development: economic aspects, environmental perfor-
optimal location and size of facilities and the flow through the facilities mance and social responsibility. We review papers containing math-
[93]. As recalled in Zanjirani Farahani et al. [165], “there are many ematical models (linear and nonlinear programs with integer or
models in the SCND literature. Different decisions are made in the SCND mixed-integer variables) with binary decision variables modeling the
and perhaps the most critical one is locating the facilities in different tiers selection of candidate facilities.
of the supply chain”. Our research questions can be briefly stated as follows: (i) Which
Fifty years ago, the seminal paper by Hakimi [121] generalized the environmental and social criteria are considered in sustainable SCND
original Weber problem [159] from a single facility location problem to research? (ii) How are they integrated into mathematical models? (iii)
a multiple facility location problem. This publication marked a revival Which optimization methods and tools are used? (iv) Which real-life
M. Eskandarpour et al. / Omega 54 (2015) 11–32 13
applications of sustainable SCND are described in the scientific lite- start of the time period was chosen such that the Brundtland Report
rature? of the World Commission on Environment and Development [103]
Section 2 describes the methodology adopted for the collection of served as a starting point, in a similar way to Seuring and Müller [152]
research papers and compares our work with existing reviews on and Chen et al. [106].
related topics. SCND problems with environmental and social aspects We searched papers published in international peer-reviewed
are investigated in Sections 3 and 4, respectively. In Section 3, we journals from the main electronic bibliographical sources (Scopus,
give a special focus on LCA-based methods and review the scope of Web of Science) using keywords such as sustainable development,
the environmental assessment, the environmental criteria used and green, environmental or social along with classic keywords such as
the metrics chosen to evaluate these criteria. Section 5 reviews the supply chain, network design or facility location in the titles or the
mathematical models. We used 3 main classification dimensions: topics covered. We use back-tracking to find earlier relevant
mono-objective vs. multi-objective models, linear vs non-linear, sources, and forward-tracking in Web of Science to find literature
deterministic vs stochastic. The solution methods are described in that are referring to the central sources. We also looked for recent
Section 6, which lists the use of solvers, other exact methods and surveys in related domains in order to find additional sources
heuristic or metaheuristic approaches. We devote Section 7 to the including a few conference papers.
description of case studies and real-life applications of sustainable From the collected material, we filtered the papers according to
SCND. The references are classified according to the type of economic the following rules: (i) the papers must be written in English
activity and the nature of the data. Finally, in Section 8 we conclude language, (ii) they include decision variables modeling the location
and suggest a number of future research directions. or selection of candidate facilities, (iii) the measure of environ-
mental or social impact is explicit either in the objective function
or in the constraints of the model.
2. Review methodology From the second rule, we excluded a large number of articles
dealing with the routing of product flows in an already defined
2.1. Delimitations and search for literature network. This is the case, for example, in the paper by Ramos et al.
[146], in which the authors present depot selection as an extension of
A comprehensive search of related research from 1990 to 2014 their work. The third rule enabled us to filter many papers in the field
was applied to produce a synthesis of peer-reviewed literature. The of reverse logistics and management of undesirable facilities. Reverse
Table 1
Existing reviews in related areas. RL is the reverse logistics and CL is the closed-loop.
Article Facility location Sustainability Scope or special focus
Supply chain management

Srivastava [156] ○ Green SCM, RL, CL
Awudu and Zhang [94] ○ Biofuel SCM, uncertainty
Dekker et al. [111] ○ Green logistics
Soysal et al. [155] ○ Quantitative models, food logistics
Nikolopoulou and Ierapetritou [139] ○ Chemistry
Boukherroub et al. [101] ○ Multi-criteria models
Brandenburg et al. [102] ○ OR models and methods
Masounik et al. [133] ○ RL,CL
Barbosa-Póvoa [96] ○ ○ Chemical process
Yue et al. [161] ○ ○ Biomass-for-bioenergy
Arioglu Salmona et al. [91]
Sarkis et al. [149] Green SCM
Ashby et al. [92]
Miemezyk et al. [137] Purchasing
Seman et al. [150] Green SCM
Zailani et al. [164] Malaysia
Beske et al. [99] Dynamic capabilities, food industry
Seuring [151] Forward supply chain
Yusuf et al. [162] UK oil and gas supply chains
Ashby et al. [92]
Seuring and Müller[152]
Gupta and Palsule-Desai [119]
Johnsen et al. [127]
Supply chain network design
Terouhid et al. [158] Socially responsible location

Chen et al. [106] Manufacturing
Devika et al. [20] ○ Forward, RL,CL
Zanjirani Farahani et al. [165] ○ Competitive SCND
Beamon [97]
Owen and Daskin [141]
Daskin et al. [108]
M. and G. [132]
Sahin and Süral [148]
Akçalı et al. [89] RL,CL
Melo et al. [135]
Aras et al. [90] RL,CL
Pati et al. [142] RL,CL, single objective
Hassini et al. [122] ○ Metrics
De Meyer et al. [109] ○ Biomass-to-bioenergy
logistics and closed-loop supply chain have become a major area of Boukherroub et al. [101] focus on multi-criteria decision mak-
supply chain management. Several surveys have been published in ing models for supply chain design. They point 42 papers with
the last fifteen years (see for example the surveys by Fleischmann environmental or social concern, and 43 papers with facility
et al. [113], Dekker et al. [110], Bostel et al. [100], Pokharel and Mutha location decisions, 12 of them having both characteristics. The
[145] or the special issues [116,117]). Clearly, the goal of reverse or broad review by Dekker et al. [111] contains one section on facility
closed-loop supply chain is closely related to that of sustainable location (7 shared papers).
supply chain management. However, as explained in Srivastava [157] Terouhid et al. [158] and Chen et al. [106] propose a framework
(Fig. 4), the main optimization often relies on a single economic for classifying the sustainability characteristics. They study the
objective. Environmental and social dimensions are generally not factors affecting location decisions, but these reviews do not review
explicitly assessed, but the resolution of these problems evidently the quantitative models and methods. Devika et al. [20] is a research
contributes to designing sustainable supply chain networks. paper including a section with a review of the literature.
Undesirable facilities are those facilities that have adverse We conclude that none of these reviews addresses the subject of
effects on people or the environment. They generate some form OR models and methods for sustainable supply chain network design.
of pollution, nuisance, potential health hazard, or danger to nearby
residents; they also may harm nearby ecosystems [134]. Thus, the
modeling of SCND problems that include undesirable facilities 2.3. Distribution across the time period and main journals
often implicitly include environmental or social aspects.
On that basis, 87 papers were identified. In the following, they Fig. 1 displays the yearly distribution of the reference papers. A
are denoted as reference papers and listed in a separate category in remarkable fact is that almost 90% of these papers have been
the reference list in the end of this review. published since 2008, making it clear that sustainable SCND has
been receiving growing attention.
The reference papers can be found in 41 distinct journals, only
2.2. Position in the literature
17 of them having published more than 1 paper. Fig. 2 shows the
distribution of the reference papers in these 17 journals, which
As many review papers have been written in neighboring
represent 72% of the reference papers. The high number of papers
domains, we needed to check whether the scope of the present
in Computers and Chemical Engineering and Industrial and Engineer-
paper was not already covered by the existing literature. Table 1
ing Chemistry Research underlines the importance of sustainability
summarizes the reviews published in related areas. The symbol
in columns 2 and 3 means that the corresponding paper considers
facility location as a main topic. The symbols ○ and mean that
facility location is one topic among others or is not studied in the
paper. The symbols have the same meaning in further tables.
We can classify the review papers in two categories. The first
category gathers papers dealing with Supply Chain Management in
general. In these papers, facility location is either not studied or is
only one feature among many others. For example, Brandenburg et al.
[102] mention network design as one out of 13 application areas.
They mention 13 papers in this area, all except one being published
between 2010 and 2013. Seuring [151] indicates that more than 300
articles have been published in the last 15 years on the topic of green
or sustainable (forward) supply chains, only 36 articles of which apply
quantitative models. Note that the review by Barbosa Póvoa [96]
concerns supply chain management, but with a strong emphasis on
supply chain network design. The second category regroups review
papers on SCND. Only 5 of them deal with sustainability.
Table 2 details the content of the reviews which could poten-
tially cover Sections 4–6 of our work: LCA based approaches
(column 4), optimization models (column 5) and optimization
methods (column 6). The last column reports the number of
references also mentioned in the present review.
Several reviews are dedicated to one activity: chemical and
process industries [96139], biomass-to-energy [109161]. Fig. 1. Time distribution of reference papers.
Table 2
Existing reviews in related areas.
Article Facility location Sustainability models LCA Optimization methods Optimization methods # of shared references
Nikolopoulou and Ierapetritou [139] ○ ○ ○ 12

Boukherroub et al. [101] ○ 12
Dekker et al. [111] ○ ○ 7
De Meyer et al. [109] ○ 5
Barbosa Póvoa [96] ○ ○ ○ ○ 7
Yue et al. [161] ○ ○ ○ ○ ○ 12
Terouhid et al. [158] 1
Chen et al. [106] 1
Devika et al. [20] ○ 10
Zanjirani Farahani et al. [165] ○ 9
Fig. 2. Distribution of reference papers by journal.
in chemical and process industry. Many papers are published in

journals which focus on sustainability or on one field of applica-
tion. For example, Resources, Conservation and Recycling and Waste
Management fall into this category. On the other hand, the papers
published in Industrial Engineering and Operational Research
journals are spread out in a large variety of journals.
2.4. The 3 dimensions of sustainable development
The reference papers do not all address the 3 dimensions of

sustainable development: economic aspects, environmental per-
formance and social responsibility. Fig. 3 shows their distribution
with respect to these dimensions. This distribution is consistent
with that already observed in other reviews, such as Chen et al. Fig. 3. Distribution of reference papers with respect to the 3 sustainability
[106]. The paucity of papers including social aspects has been dimensions.
already observed by many preceding reviews, and this is even
more striking in quantitative models.
supply chain from extraction of raw material to disposal of end
products. Thus, in Section 3.1, we focus on papers which assess
3. Environmental supply chain network design supply chain environmental impact through an LCA approach. Some
reference papers adopt a full LCA approach and others only calculate
Network design mathematical models traditionally aimed at one or a few LCA indicators which are further integrated into
minimizing cost or maximizing profit, with very little considera- optimization models. These two approaches will be discussed in
tion of environmental objectives and constraints. The increasing Section 3.1.2. Section 3.2 concerns papers that do not adopt an LCA
importance of environmental issues has prompted decision- approach. They rather propose partial assessment of environmental
makers to incorporate environmental factors fully into the deci- factors, focused on one or several dominating aspects of the
sion process [124], giving birth to Environmental Supply Chain application considered, for example emissions caused by transporta-
Network Design (ESCND). In other words, ESCND generalizes SCND tion or facilities.
by incorporating environmental factors, which may concern facil-
ities, transportation modes, processes, product design, technolo-
gical choices, etc. As shown for example in the case study in You 3.1. LCA based models
and Wang [81], the optimal solutions of pure economic, environ-
mental or intermediate models differ a lot. LCA assesses environmental impacts associated with all stages
This raises several questions that should be clarified when of a product life-cycle from raw material extraction to final
designing supply chains. Which environmental factors should be disposal or recycling [126]. It compiles and evaluates inputs,
considered? How can they be quantified? How can they be outputs and potential environmental impacts of a product system
integrated into mathematical models and optimization methods? throughout its manufacturing process (see the reviews by Azapa-
Table 4 in Brandenburg et al. [102] or Table 1 in Seuring [151] gic [95] and Pieragostini et al. [143]), its life-cycle and all related
shows many possible ways to model environmental decision mak- supply chain decisions.
ing: Life Cycle Assessment (LCA), reasoning maps, Analytic Hierarchy Fig. 4 represents the four main steps of LCA as defined by the
Process (AHP), Analytic Network Process (ANP), Data Envelopment ISO 14040 and 14044 standards [126].
Analysis (DEA), equilibrium models, simulation, etc. However, LCA is
the most commonly used technique and it is particularly convenient 1. Goal and scope definition sets out main objectives of the study,
to integrate its output in optimization models. Moreover, this defines functional units considered and boundaries of the
technique is a general framework for a holistic assessment of a system.
2. Inventory analysis is inventory of all flows from and to nature assessment of environmental impact through the whole supply chain
for a product system. All emissions (in air, water and soil), from raw material to materials processing, manufacture, distribution,
extractions and land use are listed and quantified. use, repair and maintenance, disposal and recycling. This category
3. Impact assessment measures environmental impact of all emis- regroups 12 papers. In the context of fuel supply chains, cradle-to-
sions listed in the preceding step. grave is called well-to-wheel (WTW). For example, Elia et al. [24]
4. Results interpretation consists in analyzing and interpreting provide an analysis for hybrid coal, biomass, and natural gas to liquid
results of each of the three preceding steps. The outcome of (CBGTL) plants. The supply chain described includes both cultivation of
the interpretation phase is a set of conclusions and recommen- biomass and coal and natural gas mining, followed by industrial and
dations for the study. logistics operations. In the context of biomass supply chains, cradle-to-
grave is called field-to-wheel (FTW). This is applied to cellulosic
We found 39 papers that integrate principles of LCA into their ethanol [82], sugar cane to ethanol [55] or to a general “biomass-to-
supply chain network design models. Among the four LCA steps, liquid” supply chain [81].
we review the goal and scope definition and the impact assessment The cradle-to-gate scope concerns all steps from extraction to the
steps. The inventory analysis is an important intermediate step but factory gate (23 papers). This scope is frequent for B2B companies
it is directly related to supply chain decisions. The mathematical having multiple customers. In fuel supply chains, this LCA scope is
models resulting from the preceding steps are considered by called well-to-tank in order to distinguish the GHG emitted during
several authors as a part of the interpretation step. fuel production from those emitted by the vehicle operations. It is
called field-to-tank in biomass supply chains.
Gate-to-gate (3 papers) generally concerns companies in inter-
3.1.1. Scope definition mediate echelons of a supply chain, which manufacture or transform
To determine boundaries of the supply chain is the first critical and deliver goods to their customers without extracting raw materials
decision in LCA. or playing any role in disposal of end-of-life products. This scope is
Fig. 5 illustrates the various LCA scopes found in the reference also used in transformation of end-of-life products which are re-used
papers. The cradle-to-grave scope assumes a comprehensive in the same or another supply chain. For example, Dehghanian and
Mansour [19] study a recovery network for scrap tires which can be
used as a substitute for fuel in cement plants.
Gate-to-Grave (1 paper) focuses on the last steps of a supply
chain, from factory gate to product disposal. This scope is con-
venient in the study of waste supply chains or reverse logistics
activities.
3.1.2. Life-cycle impact assessment

The goal of life-cycle impact assessment (LCIA) is to express the
complex output of inventory analysis into a few environmental areas
of interest. Mid-point oriented LCIA methods cover various impact
categories such as greenhouse effect (or climate change), natural
resource depletion, stratospheric ozone depletion, acidification, eutro-
phication, human toxicity and aquatic toxicity. Damage-oriented
Fig. 4. Conceptual framework of LCA [126]. methods (or endpoint methods) aggregate mid-point categories into
Fig. 5. LCA scopes.

fewer categories of damage: damage to human health, ecosystem and needs expertise in environmental management. Several
health or damage to resources. There exist several LCIA methodolo- authors do not lead an exhaustive LCIA approach and only borrow
gies, which include different midpoint and endpoint categories. one or a few impact categories which are directly integrated into
In the mathematical models described in the reference papers, their mathematical models. The papers that adopt this approach
the environmental assessment can be based either on midpoint or are listed in Table 4.
endpoint categories. Models can include exhaustive LCA or only a Climate change is often quantified by the Global Warming
small subset of pertinent impact categories. We call the latter Potential (GWP) indicator [125]. It is captured by inventorying
approach an LCA-based approach. CO2 , CH4 , N2 O emissions and regrouping them in a single indicator
expressed as CO2 -equivalent emissions during a predefined per-
iod, typically 100 years. GWP is mainly used with cradle-to-grave,
3.1.3. LCIA methods
well-to-wheel, field-to-wheel and well-to-tank scopes.
Table 3 lists papers based on endpoint methodologies. Three
It is often used as a single indicator of environmental impact or
methods are described in reference papers: Eco-Indicator 99 (EI-
is completed with some application dependent indicators: Ber-
99), Impact 2002 þ and ReCiPe.
nardi et al. [9] consider GWP and water footprint, which indicates
Eco-indicator 99 [114] gathers 11 impact categories into three
the amount of freshwater consumed or polluted during the whole
damage categories (human health, ecosystem quality and
production process of a commodity.
resources). The overall environmental impact is finally measured
Mele et al. [55] measure environmental performance with
as a single metric. EI-99 is chosen in 15 papers, mainly with a
biochemical oxygen demand (BOD20), because of its importance
cradle-to-gate scope.
as an indicator of pollution of watercourses.
Depending of the industrial activity, some impact categories
There can be two reasons for resorting to partial LCIA approach
can be omitted. For example, in the context of chemical supply
instead of exhaustive LCIA: simplifying calculation or focusing on
chain, Hugo and Pistikopoulos [40] use the 10 most relevant
impacts which are most relevant for the application considered.
impact indicators.
Guillén-Gosálbez et al. [37] explore the environmental benefits of
IMPACT 2002þ [128] has 14 midpoint indicators and 4 cate-
adopting a hydrogen economy, in terms of overall contribution to
gories of damage: human health, quality of ecosystems, climate
climate change. For this reason, instead of calculating the EI-99
change and resource depletion. It is used in 3 papers with a cradle-
itself, they focus on only one of its impact categories: damage to
to-gate scope. In these papers, an overall environmental objective
human health caused by climate change. Pinto-Varela et al. [64]
is the sum of all endpoint damages for each facility in the
calculate a partial EI-99 by only considering damage to human
supply chain.
health caused by electricity and diesel consumption. Other authors
CML92 is used in Bloemhof-Ruwaard et al. [10] with seven impact
choose to consider individual impact indicator to complement one
categories. ReCiPe [115] has 18 midpoint categories combined into
LCIA score. Kostin et al. [45] consider three impact categories from
3 endpoint damage categories (human health, ecosystems, resource
the EI-99 (damage to human health, damage to eco-system quality,
surplus cost). This method also results in one single score.
damage to resource), the EI-99 itself, and the GWP. Mele et al. [56]
consider the EI-99 and the GWP.
3.1.4. Impact categories
The score provided by Eco-Indicator 99 or ReCipe can be easily 3.2. Partial assessment of environmental factors
incorporated into optimization models as an environmental objec-
tive function. However, although they use this approach, Pishvaee For various reasons, implementing a methodology such as LCA
and Razmi [65] claim that LCA process is costly, time consuming is not always possible. Awareness of environmental concerns in
Table 3
LCIA methods.
Method Articles
Eco-Indicator 99 (EI- Pishvaee and Razmi [65], Hugo and Pistikopoulos [40], Guillén-Gosálbez et al. [36], Guillén-Gosálbez and Grossmann [34], Duque et al. [22],
99) Guillén-Gosálbez and Grossmann [35], Mele et al. [56], Abdallah et al. [1], Pozo et al. [69], Dehghanian and Mansour [19], AChaabane et al. [15],
Kostin et al. [45], Santibañez-Aguilar et al. [76], Yue et al. [83]
IMPACT 2002 þ Bojarski et al. [11], Pérez-Fortes et al. [63], Muñoz et al. [60]
CML92 Bloemhof-Ruwaard et al. [10]
ReCiPe Mota et al. [59]
Table 4
Impact categories and indicators.
Impact Articles
Climate Change Hugo et al. [41], Quariguasi Frota Neto et al. [70], Zamboni et al. [85], Elia et al. [24], Giarola et al. [30], Mele et al. [56], You an Wang [81],
Zamboni et al. [86], Akgul et al. [3], Chaabane et al. [15], Elia et al. [25], Giarola et al. [25], Giarola et al. [31], Giarola et al. [32], Kostin et al.
[45], Pishvaee et al. [67], You et al. [82], Bernardi et al. [9], Ruiz-Femenia et al. [72], Elia et al. [26], Yue et al. [84]
Biochemical Oxygen Demand Mele et al. [55]

(BOD20)
Damage to human health Guillén-Gosálbez et al. [37], Pinto-Varela et al. [64], Kostin et al. [45],
Water footprint Bernardi et al. [9]

companies is generally gradual, so that assessing only a subset of Facilities: Since facility location is a central decision in SCND
environmental factors can be viewed as an intermediate step models, integrating environmental impact of facilities into
towards full integration. Partial assessment of environmental mathematical models seems to be natural. This impact is
factors also makes sense when obtaining environmental data considered in 28 of the papers in Table 5, but surprisingly
and modeling the whole supply chain is too difficult. This section enough only 6 of them measure the GHG emissions due to
has a structure similar to that of the preceding section. We first facilities. The most classic metric to assess the environmental
review the scope chosen for integrating environmental concerns, impact of facilities is energy consumption, which can depend
i.e. which activity in the supply chain is concerned. Then, we list on sizing decisions and technological choices. The models by
the performance measures used in each paper. Amin and Zhang [5], Caruso et al. [13], Costi et al. [17], Galante
et al. [29], Lam et al. [48], Papapostolou et al. [61], Pishvaee
et al. [66] and Wang et al. [79] include the choice between
3.2.1. Scope competing technologies as decision variables. Pishvaee et al.
The easiest way to partially assess environmental factors has [66] integrate the average amount of waste generated with
been to enrich traditional SCND models with one or a number of each technology in their environmental and social objective
environmental objectives, constraints or parameters. This keeps function. Other measures include the number of obnoxious
the focus on logistics operations in the supply chain, while facilities installed [28] (which is influenced by technological
integrating new concerns into the decision process. For example, choices) or the risk placed on the nearby population [4].
knowing that transport and industrial facilities account for 22% Transport: One of the easiest ways to incorporate environmen-
and 20% of global CO2 emissions respectively [140], several SCND tal criteria into pure economic models is to calculate emissions
models integrate CO2 emissions due to transport or facilities. of GHG and particulates due to transport. Tools for calculating
Table 5 lists three categories in which environmental criteria and converting emissions into a single CO2 equivalent mea-
are most often incorporated: facilities, transport and product surement can be provided by national or regional organiza-
related criteria. Next paragraphs detail the content of this table. tions, such as the Mobile6 software used by Elhedhli and
Merrick [23] for heavy duty diesel vehicles.
Table 5 Some models integrate selection of transportation modes into
Scope used for partial assessment of environmental impact. strategic network design decisions. In these models, transpor-
tation modes generally compete on cost, environmental impact
Article Facilities Transport Product and capacity. The choice between transportation modes can
Caruso et al. [13] also be determined by loading/unloading conditions, frequency,
Berger et al. [8] minimum lot-size, etc.
Pati et al. [62] Since SCND models generally consider aggregated data, opera-
Corsano et al. [16] tional characteristics such as vehicle speed and daily variations
Erkut et al. [27]
are mostly ignored. We did not find any reference considering
Lira-Barragán et al. [49]
Pishvaee et al. [66] more global assessment of transportation, such as impact of
Eskandarpour et al. [28] vehicles on road network.
Costi et al. [17] Process and product design: Decisions concerning product flows
Minciardi et al. [57] and design can also be fully integrated into environmental
Alçada-Almeida et al. [4]
Saffar et al. [75] SCND. Krikke et al. [47] propose a Mixed-Integer Linear
Pourmohammadi et al. [68] Program (MILP) whose decision variables concern both net-
Galante et al. [29] work design and product design. They analyze interactions
Elhedhli and Merrick [23] between both types of variables and conclude that logistics
Mallidis et al. [53]
network structure has most impact on costs, whereas product
Bouzembrak et al. [12]
Sadrnia et al. [73] design has most impact on energy and waste. Abdallah et al. [2]
Xifeng et al. [80] observe that price of raw material increases as the product
Zhang et al. [87] becomes greener. Thus, supplier selection has contradictory
Saffar et al. [74] impact on cost and environmental dimension [131]. Amin and
Harraz and Galal [38]
Zhang [5] assess impact of choosing environmentally friendly
Ramudhin et al. [71] materials in the production process.
Chaabane et al. [14]
Harris et al. [39]
Liu et al. [51] 3.2.2. Performance measures
Tuzkaya et al. [77]
Wang et al. [79] According to Krikke et al. [47] or Harraz and Galal [38], given
Jamshidi et al. [42] LCA complexity, it becomes regular practice to use more pragmatic
Kannan et al. [43] indicators such as residual waste and energy used.
Kanzian et al. [44] Ahi and Searcy [88] identified 2555 unique metrics to measure
Lam et al. [48]
Govindan et al. [33] performance in green and sustainable supply chains. Due to lack of a
Devika et al. [20] generic assessment methodology, a wide range of ad hoc perfor-
Marufuzzaman et al. [54] mance measures have been developed to assess environmental
Mohammadi et al. [58] performance of a supply chain, so that identifying the most appro-
Papapostolou et al. [61] priate performance measures is still a challenging issue [139].
Amin and Zhang [5]
Table 6 details the metrics found in the reference papers for
Krikke et al. [47] assessing the environmental impact. Columns 2–6 report various
Krikke (2011) [46] families of performance measures: GHG emissions (column 2),
Abdallah et al. [2] amount of waste generated (column 3), energy consumption
Diabat et al. [21]
Baud-Lavigne et al. [6] (column 4), amount of material recycled (column 5) and others
measures (column 6).
Table 6 pollution. In Harris et al. [39], the quantity of energy used is a

Metrics used for partial assessment of environmental impact. mean to estimate GHG emissions.
Waste generated, energy use, material recovery and other mea-
Article GHG Waste Energy Material Others
emissions use recovery sures:
Environmental performance can be measured by many possible
Ramudhin et al. [71] criteria which generally arise from the economic sector con-
Chaabane et al. [14] cerned. The quantity of waste generated is mentioned in 16
Krikke [46]
Wang et al. [79]
papers, the use of energy is included in 8 models and the
Elhedhli and Merrick material recovery in 7 models. For example, Pati et al. [62]
[23] measure the value of wastepaper recovered by a paper recy-
Jamshidi et al. [42] cling system. Amin and Zhang [5] measure the use of renew-
Kannan et al. [43]
able and recycled energy, such as solar power. Finally, 7 papers
Mallidis et al. [53]
Abdallah et al. [2] use criteria that cannot be classified in the preceding cate-
Diabat et al. [21] gories. Other metrics include use of water [135061], noise
Govindan et al. [33] pollution [1258] or an overall estimation of long-term impact
Kanzian et al. [44] and cleanup cost of oil-spill caused by vessels [78]. Pourmo-
Sadrnia et al. [73]
Xifeng et al. [80]
hammadi et al. [68] measure a virgin material opportunity costs
Zhang et al. [87] which is the extra expense that a firm is willing to pay when it
Baud-Lavigne et al. [6] refuses to substitute the virgin material market by an accep-
Marufuzzaman et al. table recycled material. Other metrics are sometimes not
[54]
detailed, such as the land specific technical requirement in
Saffar et al. [74]
Saffar et al. [75] Tuzkaya et al. [77].
Caruso et al. [13]
Berger et al. [8] 3.3. Conclusion
Lira-Barragán et al.
[49]
Pishvaee et al. [66] The integration of environmental criteria in SCND is a natural
Eskandarpour et al. idea for activities with a high impact. The 83 papers dealing with
[28] environmental SCND share almost equally between LCA (39) and
Galante et al. [29] non-LCA (44) approaches. The most popular LCIA methods are
Pati et al. [62]
Lira-Barragán et al.
Eco-Indicator 99 and Impact 2002 þ. To our knowledge, ReCipe
[50] has been used only in Mota et al. [59]. Since it is a more recent
Verma et al. [78] method and it provides a single score, more authors are likely to
Alçada-Almeida et al. use it in forthcoming years. As far as impact indicators are
[4]
concerned, GWP is particularly designed for very long-horizon
Liu et al. [51]
Lam et al. [48] activities such as process industries or fuel/energy supply chains.
Devika et al. [20] Other indicators are used when they are relevant for their
Harris et al. [39] respective domain of application. Non-LCA approaches measure
Tuzkaya et al. [77] environmental performance on tangible domains (facilities, trans-
Krikke et al. [47]
Harraz and Galal [38]
port, product design) and measures (GHG emission, waste pro-
Amin and Zhang [5] duced, energy used etc.). This goes along with a collection of
Corsano et al. [16] various ad hoc measures depending on the application considered.
Papapostolou et al. Finally, let us point out 2 papers about sustainable extensions
[61]
to special facility location problems: the hub location problem [58]
Costi et al. [17]
Erkut et al. [27] and the location routing problem [33]. Incorporating environmen-
Minciardi et al. [57] tal criteria into these problems seems to be a novel research issue.
Bouzembrak et al.
[12]
Mohammadi et al.
4. Social supply chain network design
[58]
Pourmohammadi
et al. [68] Social sustainability has been examined to a far lesser degree
than environmental or green supply chain management [152].
Furthermore the definition of social sustainability itself is still
under development [98]. Social sustainability in supply chains
Carbon footprint: The most popular metric for measuring addresses issues of social justice and human rights with studies
environmental impact is the carbon footprint, which is the focusing on practices such as supplier human rights actions, labor
total amount of GHG emitted by a company or a supply chain conditions, codes of practices and social auditing, supplier com-
(25 papers). All GHG emissions may be considered, but for pliance with child labor laws, and the delivery of social equity
practical reasons, baseline indicators with only CO2 , CH4 and through sourcing from diverse suppliers in terms of gender, size,
N2 O are also used [160]. For example, Jamshidi et al. [42] ethnicity and avoidance of conflicts of interest. Including social
consider two objective functions: one minimizes the total costs aspects in network design decisions allows to better evaluate the
within the supply chain and the other one minimizes danger- impact of a supply chain on its stakeholders: employees, custo-
ous gases produced, such as NO2 , CO and Volatile Organic mers and local communities. This also helps obtaining consistency
Compounds. between qualitative and quantitative decisions. We analyze 13
GHG emissions are not always calculated explicitly. In Pourmo- references papers having an assessment of social impact with the
hammadi et al. [68] the amount of GHG emitted is considered framework proposed in Chardine-Baumann and Botta-Genoulaz
through an input-output approach to estimate the cost of air [105]. Table 7 presents this classification. Columns 2–4 correspond
Table 7 indicator. Since the social impact is often qualitative by nature, it is

Models with the social dimension. difficult to build a single metric to measure it. Multi-Criteria Decision
Making (MCDM) can be a suitable tool to overcome this problem.
Article Work Societal Customer
conditions commitment issues Dehghanian and Mansour [19] aggregate their four social criteria into
a single indicator with an Analytical Hierarchy Process (AHP) [147].
Harraz and Galal [38] AHP is also used in Datta [18]. Hence, selecting the most appropriate
Péerez-Fortes et al. [63] criteria and incorporating them into mathematical models are still
Pishvaee et al. [66]
You et al. [82] challenging issues.
Devika et al. [20]
Mota et al. [59]
Santibañez-Aguilar et al. [76] 5. Modeling approaches
Yue et al. [84]
Caruso et al. [13]
Datta [18]n In this section, we review the main characteristics of mathe-
Bouzembrak et al. [12] matical models for sustainable supply chain design problems. Such
Beheshtifar and problems have resulted in a large variety of models. This can be
Alimoahmmadi [7]n
explained not only by the variety of industrial contexts (single or
Malczewski and Ogrycżak
[52]n multiple period, single or multiple products, structure of the
Tuzkaya et al. [77] logistics network), but also by modeling issues: single or multiple
Dehghanian and Mansour objectives, deterministic or uncertain data.
[19] The main decision variables in SCND models are binary variables
concerning the location of facilities, sizing decisions, the selection of
suitable technology levels and the selection of transportation modes
between facilities. Since product flows along the supply chain are
to 3 of the 5 fields proposed in Chardine-Baumann and Botta- generally modeled by continuous constraint, the SCND models are
Genoulaz [105]. It is worth noting than no paper addresses the two often mixed-integer formulations, which can be linear or nonlinear.
last fields: human rights (child and forced labor, freedom of Some stochastic models are also found that enable the consideration
association, discrimination) and business practice (fight against of uncertainties such as the demand level.
corruption, fair-trading, promotion of corporate social responsi- The section is organized into two parts. In Section 5.1, we
bility in the sphere of influence). Reference papers followed by a n review the models with a single objective function. This objective
do not include the environmental dimension. can be either economic or environmental, but is never social only.
In the field work conditions, employment is the main social Multi-objective models are then described in Section 5.2. Both
indicator used in the literature. The number of jobs created is sections are divided into two subsections, describing deterministic
considered by most authors, with slight variations. Mota et al. [59] and stochastic models, respectively.
define a social benefit indicator which prefers job creation in the
less developed regions. Devika et al. [20] distinguish the fixed jobs
5.1. Models with a single objective
opportunities (which are independent of the level of activity) from
the variables jobs (which increase with the level of activity).
The easiest way to incorporate environmental issues into
Dehghanian and Mansour [19] aim at creating jobs in the widest
classic SCND models is to express the objective function as a
range of communities. They maximize the number of facilities
weighted average of all objective functions. This requires applying
installed, which corresponds to the idea of fixed jobs in Devika
conversion factors to convert non-homogeneous measures into a
et al. [20]. The damage to workers and security measures, such as
single one.
the exposure to chemical elements is considered in Dehghanian
For example, when the whole environmental impact can be
and Mansour [19] and Pishvaee et al.[66].
expressed through a quantity of GHG emissions, it is possible to
The Societal commitment field regroups all decisions contribut-
convert the environmental impact into its monetary equivalent by
ing to improve a population's health, education, culture [18]. It
using conversion factors. Then the monetized environmental
includes local development policies [19], equity in access to
damage can be aggregated with the economic objective into a
healthcare [7], the impact of the supply chain on real estate [12],
single objective.
but also the political opposition [1377].
The main characteristics of single objective deterministic and
The field customer issues regroup all impacts individually
stochastic models are summarized in Table 8.
affecting each customers. Malczewski and Ogrycżak [52] consider
the environmental pollution at hospital sites as a social criterion
since its impact relates directly to patients and users. In Dehgha- 5.1.1. Deterministic models
nian and Mansour [19], the customer issue concerns the risk of As mentioned above, some authors consider the economic
using recycled material. objective as the main one and represent the environmental dimen-
As pointed by several previous reviews, including social concerns sion by constraints in their models. These constraints may express a
into sustainable SCND models raises many modeling and assessment maximal authorized level of GHG emissions. For example, the
difficulties. First of all, social responsibility is a fully multi-disciplinary objective function in Elia et al. [24] is to minimize the cost of facility
and multi-stakeholder issue [66]. As a consequence, social perfor- investment, feedstock purchase and transportation. The authors
mance is generally hard to model with pertinent quantitative introduce an environmental constraint by imposing an overall
indicators. For example, Chaabane et al. [15] state that tangible GHG emission target level for each hybrid coal, biomass and natural
indicators such as noise and pollution can play the role of indicators liquid gas plant. Papapostolou et al. [61] consider a pure economic
of both environmental and social performance. However, they do not objective function. Environmental constraints limiting the land use
integrate them in their MILP since they do not identify good measures and water consumption are included in their linear model.
of social sustainability. Moreover, social and environmental impacts Other authors mix economic and environmental criteria into
sometimes strongly interact. [66] aggregate three social impacts and the objective function. In Elhedhli and Merrick [23], the objective
one environmental impact (amount of waste generated) into a single function includes two terms related to pollution cost and three
Table 8 planning horizon. Other factors such as transportation costs or the

Models with single objective. amount of waste or emissions generated or returned products may
also be considered as uncertain parameters. Moreover, the data
Article Dimensions Multi- Multi- Multi-
product period mode available at the moment strategic decisions are made are generally
aggregated and lose accuracy as the time horizon recedes. A
Linear models survey on the inclusion of stochastic components in facility
Krikke [46] Eco–Env location models is proposed by Snyder [154].
Datta [18] Eco–Soc
Elhedhli and Merrick [23] Eco–Env
However we found only two references of single objective
Kannan et al. [43] Eco–Env stochastic models for sustainable SCND problems. Giarola et al.
Bloemhof-Ruwaard et al. Eco–Env [31] propose a MILP for the design of a bio-ethanol supply chain, in
[10] which the costs of carbon and biomass are considered as uncertain
Liu et al. [51] Eco–Env
parameters. To overcome this uncertainty, a two-stage stochastic
Papapostolou et al. [61] Eco–Env
Abdallah et al. [1] Eco–Env programming approach is used. Verma et al. [78] present a two-
Abdallah et al. [2] Eco–Env stage stochastic programming approach which tackles both the
Amin and Zhang [5] Eco–Env location and stockpile of equipment at emergency response
Diabat et al. [21] Eco–Env facilities that deal with potential oil-spill emergencies on the
Mallidis et al. [53] Eco–Env
Bouzembrak et al. [12] Eco–Env–
south cost of Newfoundland in Canada. Their model includes
Soc two variants corresponding to linear and non-linear formulation
Elia et al. [24] Eco–Env of equipment acquisition cost.
Elia et al. [25] Eco–Env
Elia et al. [26] Eco–Env
5.2. Multi-objective models
Stochastic linear models
Verma et al. [78] Eco–Env 5.2.1. Deterministic models
Giarola et al. [31] Eco–Env
The deterministic multi-objective models for sustainable SCND
Non-linear models are summarized in Table 9 (linear models) and Table 10 (non-
Lira-Barragán et al. [50] Eco–Env
linear models)
Verma et al. [78] Eco–Env
Costi et al. [17] Eco–Env In practice, most sustainable SCND models are bi-objective
Corsano et al. [16] Eco–Env linear models. Many authors see the economic objective as the
Lira-Barragán et al. [49] Eco–Env traditional objective function, whereas the environmental or social
objectives are considered as extensions of the traditional single
objective models. A frequent modeling approach is to consider one
economic objective and one environmental objective such as
terms related to the cost of logistics operations. In Abdallah et al. minimizing GHG emissions.
[1] and Kannan et al. [43] the objective function is the sum of Amin and Zhang [5] extend their mono-objective model by
various logistics costs and an additional term associated with CO2 considering an additional environmental objective. In the area of
emissions above the amount allocated by the government. domestic waste treatment, Berger et al. [8] propose a comprehen-
Lira-Barragán et al. [49] minimize the total annual cost of a new sive multi-periodic MILP model for the strategic design and
industrial plant which impacts the water quality throughout a tactical planning of an integrated regional solid waste manage-
surrounding watershed. The objective function includes the waste- ment planning. The model considers several types of treatment
water treatment costs whereas the water quality appears as a technologies and sites for treatment and land-fill as well as the
constraint. Mallidis et al. [53] propose a model with several possibility of recycling waste on the markets. Several environ-
objective functions related to cost, and the emission of CO2 or mental parameters and indicators may be used.
particulate matters (fine dust). The model is solved with each Chaabane et al. [15] propose a bi-objective model for the design
objective being considered one by one. of an aluminum supply chain. A carbon credit component is
Note that Krikke [46] proposes a linear variant of mixed integer included in the economic objective, whereas the second objective
programming: binary facility location variables are pre-fixed, is to minimize the GHG emissions. The model also considers tactical
resulting in one linear program for each scenario. issues such as inventory control decisions. Akgul et al. [3] propose a
In some of the reference papers, the technical context leads to multi-period, multi-product MILP model for the optimization of a
the formulation of nonlinear models. Costi et al. [17] propose an biofuel supply chain regarding cost and environmental issues. All
MINLP model for the location of treatment facilities for solid waste stages of the biofuel life-cycle, such as cultivation, transportation
management. The objective function concerns the economic cost and production, are integrated into the proposed model. Quariguasi
and environmental issues are modeled as constraints. The binary Frota Neto et al. [70] propose a bi-objective model to assess the flow
decision variables concern the existence of facilities. Continuous of materials, the amount of production at each plant and to select
variables model the material flows between facilities. Non- the most suitable end-of-use alternatives, such as refurbishing and
linearity comes from multiplications between continuous vari- recycling. Guillén-Gosálbez et al. [37] develop a bi-objective MILP
ables. Corsano et al. [16] consider ethanol plant design and ethanol model for a hydrogen supply chain design. The influence of the
supply chain design simultaneously. Non-linearity arises from hydrogen network operation on climate change is investigated as an
some non-convex constraints in the ethanol plant design model. environmental issue. In chance-constrained programming, the mod-
els embed the probability of satisfying constraints subject to
uncertain data. The model considers capacity expansion (see also
5.1.2. Stochastic models Hugo et al. [41]).
By definition, sustainable SCND models aim at impacting the A two-echelon multiple-vehicle location-routing problem with
structure of the logistics network of a company in the long term. It time windows for optimization of sustainable supply chain net-
is therefore realistic to expect to face uncertainties in the analysis work of perishable food is studied by Govindan et al. [33]. They
of the problem. This is particularly true for the consideration of the propose a deterministic model involving an economic goal for the
uncertainties on the level of customer demands within a strategic minimization of all fixed and variable costs and an environmental
Table 9 consists of locations and technologies for transfer stations, mate-

Deterministic multi-objective linear models. rial recovery facilities, incinerators and sanitary landfills, as well as
the waste flow between these locations.
Article Dimensions multi- multi- multi-
product period mode Only a few bi-objective models are non-linear. In Beheshtifar
and Alimoahmmadi [7], one of the objective is to minimize the
Malczewski and Ogrycÿzak Eco–Soc standard deviation of distances from the place of demand points to
[52] the open facilities. Due to economies of scale, Zhang et al. [87]
Caruso et al. [13] Eco–Env–
Soc
include non-linear CO2 emissions due to transportation. In the last
Erkut et al. [27] Eco–Env five references in the table, the models can be linearized. Yue et al.
Minciardi et al. [57] Eco–Env [83] and Yeu et al. [84] linearize their model with the Charnes-
Alçada-Almeida et al. [4] Eco–Env Cooper transformation and Glover's linearization. The authors
Dehghanian and Mansour Eco–Env–
compare the performance of the linear and non-linear formula-
[19] Soc
Galante et al. [29] Eco–Env tions of their models.
Tuzkaya et al. [77] Eco–Env–
Soc
Pozo et al. [69] Eco–Env 5.2.2. Stochastic models
Xifeng et al. [80] Eco–Env Like many supply chain management problems, SCND pro-
Govindan et al. [33] Eco–Env blems are subject to uncertainty. Uncertainty can have many
Devika et al. [20] Eco–Env–
Soc
different sources, like the level of demand or the proportion of
Krikke et al. [47] Eco–Env returned products in closed-loop supply chains. Uncertainty can
Pati et al. [62] Eco–Env also affect the outputs and depend on the performance of the
Quariguasi Frota Neto et al. Eco–Env process. Such an example is the level of GHG emissions. As pointed
[70]
out by Guillén-Gosálbez and Grossmann [34], many uncertainties
Harraz and Galal [38] Eco–Env–
Soc exist in the life-cycle inventory but many LCA methods assume
Amin and Zhang [5] Eco–Env nominal values for the input data. These authors mention however
Lam et al. [48] Eco–Env that the Eco-indicator 99 methodology is affected by three main
Baud-Lavigne et al. [6] Eco–Env sources of uncertainty: the operational or data uncertainty, but
Hugo et al. [41] Eco–Env
Jamshidi et al. [42] Eco–Env
also the fundamental or model uncertainties, and the uncertainty
Kanzian et al. [44] Eco–Env on the completeness on the model. If well taken into account,
Sadrnia et al. [73] Eco–Env uncertainty will impact the design of a supply chain. The number
Hugo and Pistikopoulos Eco–Env
[40]
Pourmohammadi et al. [68] Eco–Env Table 10
Mele et al. [55] Eco–Env Deterministic multi-objective non-linear models.
Bojarski et al. [11] Eco–Env
Pinto-Varela et al. [64] Eco–Env Article Dimensions Multi- Multi- Multi-
Zamboni et al. [86] Eco–Env product period mode
Giarola et al. [32] Eco–Env
Pérez-Fortes et al. [63] Eco–Env– Beheshtifar and Eco–Soc
Soc Alimoahmmadi [7]
Zamboni et al. [85] Eco–Env Guillén-Gosálbez et al. [36] Eco–Env
Ramudhin et al. [71] Eco–Env Muñoz et al. [60] Eco–Env
Chaabane et al. [14] Eco–Env Zhang et al. [87] Eco–Env
Mota et al. [59] Eco–Env–
Soc Eskandarpour et al. [28] Eco–Env
Marufuzzaman et al. [54] Eco–Env Wang et al. [79] Eco–Env
Berger et al. [8] Eco–Env Yue et al. [84] Eco–Env–
Duque et al. [22] Eco–Env Soc
Guillén-Gosálbez et al. [37] Eco–Env Liu et al. [51] Eco–Env
Giarola et al. [30] Eco–Env Yue et al. [83] Eco–Env
Mele et al. [56] Eco–Env
You and Wang [81] Eco–Env
Akgul et al. [3] Eco–Env Table 11
Chaabane et al. [15] Eco–Env Stochastic multi-objective models.
Kostin et al. [45] Eco–Env
You et al. [82] Eco–Env– Article Dimensions Multi- Multi- Multi-
Soc product period mode
Bernardi et al. [9] Eco–Env
Santibañez-Aguilar et al. Eco–Env– Linear models
[76] Soc Pishvaee et al. [66] Eco–Env–
Soc
Ruiz-Femenia et al. [72] Eco–Env
goal for the global minimization of environmental impacts of Saffar et al. [74] Eco–Env
Saffar et al. [75] Eco–Env
opening manufacturing and distribution facilities and for the Pishvaee and Razmi [65] Eco–Env
emissions due to shipments between facilities. Pishvaee et al. [67] Eco–Env
Very few models have more than three objective functions. Amin and Zhang [5] Eco–Env
Erkut et al. [27] develop a multi-criteria facility location model for Non-linear models
the municipal solid wastes management at the regional level in Guillén-Gosálbez and Eco–Env
North Greece. Their MILP model includes 5 objective functions : Grossmann [34]
1 relative to minimum total cost of facilities implementation and Guillén-Gosálbez and Eco–Env
Grossmann [35]
flows, and 4 related to the environmental impacts (GHC effects, Mohammadi et al. [58] Eco–Env
landfilling, energy and materials recovery. A solution to the model
and size of production and transport facilities clearly depends on problems are NP-hard [144], since they generalize facility location
the mean values of input data, but also of their possible variation. problems. However, instances of average size are still tractable by
Uncertainty will also affect the evaluation of a supply chain in mathematical solvers. Thus a large variety of solution methods are
terms of costs, GHG emissions, etc. used. This section is divided into three subsections. Section 6.1
The stochastic multi-objective models encountered in our reviews the methods used for solving single-objective models.
review are summarized in Table 11. This includes multi-objective models for which the objective
In Pishvaee et al. [66], a first objective function minimizes a function is a weighted sum of the objectives. Section 6.2 is devoted
sum of logistics costs and a second objective function aggregates to methods for multi-objective models: ε-constraint, metaheur-
the four social and environmental impacts already presented in istics, multi-criteria decision analysis (MCDA), and other methods.
section 4. Amin and Zhang [5] extend their deterministic model by Finally, Section 6.3 describes the use of modeling tools and solvers
considering uncertain demand and amount of returned products. in all reference papers.
They use a scenario-based stochastic programming approach.
Ruiz-Femenia et al. [72] study the effect of demand uncertainty 6.1. Solution methods for models with a single objective
on the economic and environmental performance of supply chains.
Their model seeks to maximize the expected profit and minimize Heuristics and metaheuristics are widely applied in the SCND
the probability for environmental factors to exceeding a given limit. literature, but still rarely employed in sustainable SCND. Elhedhli
Guillén-Gosálbez and Grossmann [34] provide a MINLP model to and Merrick [23] use Lagrangean relaxation to decompose their
maximize the net present value and minimize the environmental three-echelon model into a capacitated facility location problem
impact for chemical supply chains, with uncertainty about the with single sourcing and a concave knapsack problem that can be
amount of emissions released and the feedstock requirement. In solved easily. The Lagrangean relaxation is completed with a
Guillén-Gosálbez and Grossmann [35], the value of damage factors Lagrangean heuristic which finds a near-optimal solution for a
is considered an uncertain parameter so a chance-constraint model set of instances with up to 10 suppliers, 20 plants and 150
is applied to handle them. customers. Tuzkaya et al. [77] use the weighted sum to integrate
Mohammadi et al. [58] propose a novel variant of the hub the two objective functions of their bi-objective model. Then they
location model called the sustainable hub location problem (SHLP) resort to a genetic algorithm to solve single objective models.
in which two new environmental-based cost functions accounting
for air and noise pollution of vehicles are incorporated and related
6.2. Solution methods for multi-objective models
to fuel consumption. The cost of emission at the hubs is also
considered. To cope with uncertain data incorporated in the
The multi-objective methods for solving sustainable SCND
model, a mixed possibilistic-stochastic programming approach is
models are summarized in Table 12.
proposed to construct the crisp counterpart, resulting in a mixed
integer nonlinear programming (MINLP) optimization model
according to the nonlinear form of the objective functions. 6.2.1. Weighted sum of objectives
Fuzzy set theory [163] provides an efficient tool to capture the An intuitive approach to handle multi-objective models is to
imprecision of data. It is employed when there are not enough weight each criterion and to minimize the weighted sum of all
historical data to estimate probability distribution functions of criteria. The main advantage of this approach is to model and solve
uncertain parameters. This approach is chosen in Pishvaee and multi-objective problems with single-objective approaches. Unfor-
Razmi [65], Pishvaee et al. [66,67]. tunately, this modeling may not represent the decision-maker's
Pinto-Varela et al. [64] model two case studies in a Portuguese interest and may modify the Pareto structure of the problem [69].
industry with multiple products and periods. Their approach It can be used only when the Pareto set is convex. Such an
includes a fuzzy-like modeling to indicate the trade-off between approach is chosen in Bernardi et al. [9] where the three conflict-
the economic and environmental objectives considered. Like in ing objectives are the economic one, the impact on global warm-
Guillén-Gosálbez and Grossmann [34], the stochastic model is ing, and the impact on water resources.
converted into a deterministic one to facilitate its solution. Pinto-Varela et al. [64] use a symmetric fuzzy linear program-
ming (SFLP) for a bi-objective model. The model maximizes a
single variable 0 r λ r 1 representing the degree to which each
5.3. Conclusions on modeling objective must be satisfied.
In summary, a large variety of modeling techniques have been

used in order to address sustainable SCND problems among which 6.2.2. Epsilon-constraint
most used techniques are MIP for linear or non-linear problems. The ε-constraint method consists in prioritizing a primary
Non-linearity often arises from the modeling of non-linear indus- objective while expressing other objectives as constraints. Fixing
trial processes. various values of constraint enables the Pareto front to be
Some models consider a single objective aggregating the economic approximated. This method is well adapted to the extension of a
and environmental or sometimes social factors. However most of the single-objective economic approach to bi-objective models inte-
models explicitly consider two or three different objectives functions grating environmental or social criteria. Indeed, by considering the
(or sometimes more), which is natural to cope with the different economic model as the primary objective, this approach enables
dimensions of sustainable development. Since the social impact can decision makers to measure the financial impact of environmental
be difficult to quantify, it is sometimes not addressed explicitly into a or social constraints.
mathematical model, but rather in a preliminary step of scenario The model in Pérez-Fortes et al. [63] includes economic, social
definition or in a post-optimization evaluation of the solutions. and environmental criteria. Since the social metric is discrete, only
the environmental criterion is represented in the ε-constraint and
the authors represent one Pareto front for each possible value of
6. Solution methods the social metric.
Guillén-Gosálbez and Grossmann [34,35] propose bi-criteria
The goal of this section is to review the solution methods and MINLPs. In Guillén-Gosálbez and Grossmann [34], the environ-
the tools employed for solving sustainable SCND models. SCND mental criterion is transferred to the ε-constraint. The MINLP
Table 12
Solution methods for multi-objective models.
Type of method Articles
Weighted sum of Caruso et al. [13], Krikke et al. [47], Bojarski et al. [11], Galante et al. [29], Amin and Zhang [5], Bernardi et al. [9], Kanzian et al. [44],
objectives Marufuzzaman et al. [54]
ε-constraint Guillén-Gosálbez et al. [36], Guillén-Gosálbez and Grossmann [34], Mele et al. [55], Duque et al. [22], Guillén-Gosálbez and Grossmann [35],
Guillén-Gosálbez and Grossmann [37], Chaabane et al. [14], Mele et al. [56], You and Wang [81], Akgul et al. [3], Kostin et al. [45], Pérez-Fortes
et al. [63], You et al. [82], Pishvaee and Razmi [65], Pishvaee et al. [66], Pozo et al. [69], Amin and Zhang [5], Ruiz-Femenia et al. [72], Xifeng et al.
[80], Yue et al. [83], Baud-Lavigne et al. [6], Marufuzzaman et al. [54], Mota et al. [59], Santiba ñez-Aguilar et al. [76], Yue et al. [84]
Goal Programming Alçada-Almeida et al. [4], Galante et al. [29], Pati et al. [62], Ramudhin et al. [71], Chaabane et al. [14], Harraz and Galal [38]
Interactive fuzzy Malczewski and Ogrycÿzak [52], Pinto-Varela et al. [64], Pishvaee et al. [67]
approach
Metaheuristics GA: Dehghanian and Mansour [19], Tuzkaya et al. [77], Zhang et al. [87], MA: Jamshidi et al. [42], VNS: Eskandarpour et al. [28], Devika et al. [20],
PSO: Govindan et al. [33], SAþICA: Mohammadi et al. [58]
NSGA 2þ TOPSIS: Beheshtifar and Alimoahmmadi [7]
NSGA 2þ Fuzzy: Saffar et al. [74] NSGA 2þ ε-constraint: Saffar et al. [75]
Others Hugo and Pistikopoulos [40], Erkut et al. [27], Minciardi et al. [57], Quariguasi Frota Neto et al. [70], Zamboni et al. [85], Galante et al. [29], Wang
et al. [79], Datta [18], Sadrnia et al. [73]
model is decomposed into two levels: a master convex MINLP is Eskandarpour et al. [28] use a parallel Variable Neighborhood
solved to provide a vector of integer variables. In the second level, Search (VNS) to solve a multi-objective reverse supply chain design
a continuous nonlinear problem is solved to obtain a lower bound. problem for a post-sales service. The effectiveness of parallelization
The approach in Guillén-Gosálbez et al. [37] is similar: an upper is proved by a comparison with the results of a generic VNS.
level problem and a lower level problem are solved repeatedly. The closed-loop MILP model proposed by Devika et al. [20] is
Integer and logic cuts are added until the bounds converge. The solved through an hybrid approach combining three novel hybrid
model in Guillén-Gosálbez and Grossmann [35] is non-convex metaheuristics based on adapted imperialist competitive algorithms
with a specific structure. The net present value is transferred to the and variable neighborhood search. The 2-echelon location routing
ε-constraint. The resulting single-objective model is solved with a model proposed by Govindan et al. [33] is solved using a hybrid
spatial branch-and-bound that exploits the specific structure of metaheuristic algorithm combining the adapted multi-objective
the model. particle swarm optimization (MOPSO) and the adapted multi-
Pozo et al. [69] solve their multi-objective optimization problem objective variable neighborhood search algorithm (AMOVNS).
with an ε-constraint approach. They then use Principal Component In order to solve their sustainable hub location problem,
Analysis (PCA) to reduce the dimensionality of the model with the Mohammadi et al. [58] model their MINLP with GAMS and solve
objective of preserving its Pareto structure. Finally, the ε-constraint it with the BARON software. However computing times are huge
approach is used again on the reduced model. In Kostin et al. [45], for the instances with 15 nodes. Due to this limitation, they
the ε-constraint is followed by the rigorous MILP dimensionality developed a simulated annealing and an Imperialist Competitive
reduction approach based on the δ-error definition [118]. Algorithm (ICA) to find good solutions.
In their multi-objective uncapacitated facility location problem,
Xifeng et al. [80] consider the minimization of CO2 emissions as
the main objective. The economic and the service objectives are 6.2.4. Multi-criteria decision analysis and interactive methods
reformulated as constraints. The single-objective problem is Multi-Criteria Decision Analysis is able to handle a larger
solved with a greedy-drop heuristic. number of environmental and social criteria. Interactive methods
are generally preferred when the number of objective functions
increases and when the decision makers wish to be involved in the
construction of a solution.
6.2.3. Metaheuristics for multi-objective models The hospital location problem described by Malczewski and
Dehghanian and Mansour [19], Tuzkaya et al. [77], and Zhang Ogrycżak [52] is solved as an illustration of an interactive approach
et al. [87] propose Genetic Algorithms (GA) to solve their models. proposed by the authors : DINA (Dynamic Interactive Network
Tuzkaya et al. [77] propose a two-stage methodology for the Analysis System). This method is specialized for the solution of
strategic design of a reverse logistics network. The weights of facility location or transport problem and facilities user-system
each criterion are calculated with an Analytic Network Process interactions for the determination of Pareto optimal solutions.
(ANP) procedure, and then the candidate locations are evaluated As an alternative to Analytic Hierarchy Process [147], Datta [18]
with a fuzzy TOPSIS (Technique for Order of Preference by develop a multi-criteria decision making process based on Reason-
Similarity to Ideal Solution). In a second stage, the facility location ing Maps [138] to solve a rural development problem.
problem is solved by means of a genetic algorithm. In Zhang et al. Pishvaee et al. [67] propose an interactive fuzzy solution
[87], the upper level searches for the optimal terminal network approach based upon a credibility measure. At each iteration, a
configurations by using a genetic algorithm, while the lower level crisp bi-objective MILP is converted into a single objective model
performs multi-commodity flow assignment over a multimodal according to a dedicated aggregation function. The model is then
network. Jamshidi et al. [42] develop a Memetic Algorithm (MA) to solved by LINGO 8.0. The decision maker can then alter the main
solve a multi-objective supply chain problem with cost and parameters of the model if the proposed solution is not satisfactory.
environmental issues. The Taguchi method is used to reduce the Alçada-Almeida et al. [4] describe an Interactive Decision
computational time in the crossover step. Support System (IDSS) based on goal programming and
integrating techniques from the fields of atmospheric dispersion However, they are not always used to solve the whole optimization
modeling, facility location and geographical information systems. model. Dehghanian and Mansour [19] use Lindo to solve single
The goals are the ideal solution value for each of the five objective models considering each objective separately in order to
objectives. find the ideal point. Mallidis et al. [53] minimize the economic
objective, or the GHG emissions, or the particulate matter. Pourmo-
hammadi et al. [68] use Cplex to solve an LP subproblem once the
6.2.5. Other methods and hybrid approaches
facilities have been set by a genetic algorithm.
Galante et al. [29] analyze the solution space by means of goal
The other solvers are generally non-linear programming solvers,
programming, weighted sum and fuzzy multi-objective programming
which include DICOPT [363483164950], SBB [8360], BARON Yue
techniques. First, the value of the objectives is determined via goal
et al. [838458] CONOPT Mohammadi et al. [3635] and SNOPT [34].
programming. Next, a Pareto-optimal solution between these solu-
tions is obtained by means of weighted sum and fuzzy multi-objective
6.4. Conclusion
programming methods. Goal programming is also used in Alçada-
Almeida et al. [4], Pati et al. [62] and Ramudhin et al. [71]. Quariguasi
In conclusion to this section, many generic or specific solution
Frota Neto et al. [70] evaluate Pareto efficiency using Data Envelop-
techniques have been used to solve the complex and usually large
ment Analysis (DEA). The model aims to minimize the necessary
size SCND models analyzed in this review. Many problems are
reduction in cost and environmental impact to eliminate efficiency.
solved using modeling tools such as GAMS, Lingo or AMPL and
Hugo and Pistikopoulos [40] and Zamboni et al. [85] reformulate their
linear or non-linear programming solvers. Single objective models
multi-objective model as a multi-parametric MILP which is solved by
are often modeled as MIPs and solved with standard solvers. To
the algorithm described in Dua and Pistikopoulos [112]. Wang et al.
the opposite, a large variety of techniques have been proposed for
[79] use the normalized normal constraint method [136] and the
solving multi-objective models, including MIP techniques again,
subproblems are solved with IBM Ilog Cplex 9.0.
but also metaheuristic approaches and hybrid exact/metaheuristic
methods. Interactive and scenario analysis methods involving the
6.3. Modeling tools and solvers decision maker's expertise are often called for. In the future, we
can still expect further use of standard solvers to handle real-life
Faced with high complexity of the supply chains, modeling the problems, but solvers will probably not be able to solve all rich
chain network design problems is often an issue it itself. Modeling problems such as sustainable location routing problems. Moreover,
languages are often used in combination with an MIP solver. we observe a contradictory situation: most papers report huge
Table 13 details the use of modeling tools and solvers in the calculation effort in seeking optimal solutions to problems that
reference papers. We distinguish the LCA-based models (column contain much uncertainty or aggregated data. Obtaining good
2) for the non-LCA-based models (column 3) in order to exhibit quality robust solutions within limited computation time would
the differences between the two branches. probably enable better interaction with the decision makers. There
The table shows that almost all LCA-based approaches use a is a real need for developing efficient solution technique methods
modeling tools combined with a solver (GAMS/Cplex or Lingo/ for large complex problems involving uncertainty, as well as the
Lindo are the most popular combinations). This suggests the main development of robust multi-criteria heuristic methods.
difficulty in these problems is the modeling of the processes and
their environmental burden. In contrast, usual optimization meth-
ods can solve the model to optimality, although sometimes with a
very long calculation time. On the contrary, non-LCA models are 7. Applications
generally more simple to express and do not always require using
modeling tools. Most published papers on sustainable SCND are based upon
The solvers can be used to solve either single-objective or multi- specific applications or an industrial context. Indeed, the study of
objective models with the weighted sum or ε-constraint techniques. sustainable development problems emerged from real-life
Table 13
Use of modeling tools and solvers.
Modeling Reference papers

/solver
tool LCA-based models Other models
GAMS/ Guillén-Gosálbez et al. [36], Bojarski et al. [11], Guillén-Gosálbez and Galante et al. [29], Liu et al. [51], Marufuzzaman et al. [54]
CPLEX Grossmann [34], Mele et al. [55], Zamboni et al. [85], Duque et al. [22], Guillén-
Gosálbez et al. [37], Giarola et al. [30], Mele et al. [56], Pinto-Varela et al. [64],
You and Wang [81], Zamboni et al. [86], Abdallah et al. [1], Akgul et al. [3],
Giarola et al. [31], Giarola et al. [32], Kostin et al. [45], Pérez-Fortes et al. [63],
Pozo et al. [69], You et al. [82], Bernardi et al. [9], Ruiz-Femenia et al. [72], Yue
et al. [83], Mota et al. [59], Santibañez-Aguilar et al. [76], Yue et al. [84]
GAMS/ Guillén-Gosálbez et al. [36], Guillén-Gosálbez and Grossmann [34], Guillén- Papapostolou et al. [61], Corsano et al. [16], Lira-Barragán et al. [49], Lira-
others Gosálbez and Grossmann [35], Muñoz et al. [60], Yue et al. [83], Yue et al. [84] Barragán et al. [50], Mohammadi et al. [58]
Lingo/ Dehghanian and Mansour [19], Chaabane et al. [15], Pishvaee and Razmi [65], Costi et al. [17], Minciardi et al. [57], Pati et al. [62], Harraz and Galal [38],
Lindo Pishvaee and Razmi [66], Pishvaee et al. [67] Kannan et al. [43], Mallidis et al. [53], Lam et al. [48]
AMPL/ Berger et al. [8]
Cplex
None/ Elia et al. [24] Krikke et al. [47], Erkut et al. [27], Pourmohammadi et al. [68], Ramudhin
Cplex et al. [71], Chaabane et al. [14], Wang et al. [79], Elhedhli and Merrick [23],
Elia et al. [25], Amin and Zhang [5], Bouzembrak et al. [12], Diabat et al. [21],
Verma et al. [78], Baud-Lavigne et al. [6], Elia et al. [26]
None/ Krikke [46]
Excel
concerns and the modeling of environmental or social factors actually the public sector paper considers the economic and social
generally requires the description of a specific context and factors only, but concerns a very specific study.
depends of a particular case. Few papers propose generic models Consistently with the analysis of Section 5, we could not find
not based upon a specific application or sector, but that can apply any correlation between the type of models used and the sector of
to different contexts and address fundamental questions for the application. We believe that the use of a linear or non-linear
supply chain design. In classical approaches of SCND or reverse formulation with a deterministic or stochastic context is more
logistics, we indeed observe a much larger proportion of generic linked to the technical specificity of the problem studied than to
models compared to sector specific rich models. Analyzing, mod- the economic sector. Indeed advanced modeling calls for the
eling and solving supply chain design problems integrating envir- inclusion of a multi-objective and a stochastic approach rather
onmental or social factors is much more complex and makes it than a deterministic one, whatever the considered sector. More-
difficult to design generic models without a specific case in mind. over, we did not find any correlation between these approaches
The goals of this section are to classify and discuss the and the fact that the models are generic or applied to a given
published works according to their application area or economic sector. The choice of approach depends more on the complexity of
sector, types of problems and type of experiments. In doing so, we the problem and size of experimental data.
want to identify what are the leading sectors of application on
which research on sustainable SCND has been focused, what are
the reasons for that and to investigate possible differences 7.2. Conclusion on applications
between sectors and the reasons for that. We wish to address
these questions in view of the analysis conducted in the previous In summary to this section, we have observed that the research
sections, and investigate if sectorial approaches differ in environ- on quantitative optimization models for sustainable SCND pro-
mental and social factors considered and their assessment meth- blems covers a wide variety of areas and specific applications,
ods, analyze the types of models and solution techniques used and while only a few works only are devoted to the study of generic
what is the influence of including environmental and social sustainable SCND problems. Process industry sectors such as
aspects in the network design in these sectors. energy and chemical processes as well as waste management
It is also of interest to discuss the kind of experiments that have concentrate more than half of the works, while the rest is
been conducted order to validate or apply the models and solution concentrated on the analysis of consumer and industrial goods
techniques developed for a given problem. Likewise for other problem, the public sector and generic problems. The specificity of
supply chain design or optimization problems in general, we found supply chains in different areas, especially for the assessment of
two different experimental approaches in the reviewed papers : environmental factors makes it very difficult to develop generic
papers based upon empirical data, that are based on real data models that would remain realistic enough. But this should be a
arising from one or several companies, and papers pertaining to an goal for the future.
industrial context, that are inspired from a realistic context. To Sustainable SCND problems for biomass-to-bioenergy, chemical
some extent, this latter category may address problems in a more processes or waste management are already well studied due to
generic way than the former one. the importance of environmental factors (mainly energy consump-
In addition to generic papers, the papers which we have tion and pollution), but should be further investigated. To the
reviewed belong to six main application sectors. Fig. 6 indicates contrary, industrial and consumer goods sectors are well studied
the classification of these papers according to these sectors and in some areas such as tires, steel and aluminum, paper, glass and
the type of experiments conducted (either from empirical data or containers, as well as medical items, and white goods. But
from an industrial context). surprisingly enough, areas like manufacturing in general, aero-
As can be seen, most papers in Fig. 6 use empirical data and are nautics and the automotive industry, transportation services, retail
based on real applications. Also one can observe that a few economic and food distribution have hardly been studied, although they are
sectors related to the process industries (biomass-to-bioenergy, che- well present in the research on SCND in general. Extending
mical processes) or waste management concentrate about half of the research on SCND by explicitly incorporating the environmental
research. This is probably due to the great impact of these activities in and social dimensions, should be a fruitful area of research. Finally,
environmental factors, both regarding energy consumption and pollu- applications regarding the public sector have been limited so far
tion generation. These industries are probably those with the greatest and provide a great potential for the consideration of both the
maturity on these topics, while sectors related to the production and environmental and social dimensions of sustainable development.
distribution of industrial and consumer goods are still mainly focused
on the economic factors.
8. General conclusions and research directions
7.1. Intersectorial analysis
8.1. Summary of findings and contributions
We can see than generic models rarely resort to LCA-based
assessment but mostly for partial assessment based on the GHG The broad field of supply chain management has become an
emissions. W e explain this because using LCA requires a very essential domain with the globalization and the constant search
detailed analysis of product and activity which is difficult for a for competitiveness. Simultaneously, the growing consideration
generic approach. Besides, evaluating GHG emissions is a fairly for sustainable development has led private and public actors to
straightforward method and results in formulae that can be easily integrate the three pillars of sustainability within their manage-
incorporated into a mathematical model. Indeed we observe that ment. At the strategic level, the design or re-engineering of supply
LCA is used for a majority of papers in the bio-energy and chemical chain networks is a key issue, centered around questions of
processes sectors, but also for consumer and industrial goods locating and sizing facilities and defining material flows trough
sectors. This is understandable because the concerned works are the network. Optimization techniques have always been a key tool
very specific, which allows using LCA. for addressing these problems. The consideration of sustainable
Regarding the explicit inclusion of the social dimension into the development factors within the network design problem has
models, we did not identify reasons explaining that the social indeed been the subject of many works since the publication of
factors are considered or not for a given sector of application and the Bruntland Report. These were our motivations for proposing
Fig. 6. Review of industrial applications.
this review of the literature focused on optimization models and reverse logistics or facility locations have been the subject of many
techniques on supply chain network design problems integrating previous reviews.
sustainable development factors, for which no previous review Within our literature survey, we have addressed the four ques-
had been published. The overall justification of research in this tions stated in the Introduction, (i) Which environmental and social
area can be summarized by the observation that the consideration criteria are considered in sustainable SCND research? (ii) How are
of sustainable development factors may have a significant impact they integrated into mathematical models? (iii) Which optimization
on the design and configuration of the supply chain, as illustrated methods and tools are used? (iv) Which real-life applications of
by the case study in You and Wang [81]. sustainable SCND are described in the scientific literature?
Amidst many works on closely related areas to sustainable SCND We summarize our findings below and point out a number of
problems, we decided to limited our analysis to works relying on research directions for the future in the following sub section. The
mathematical optimization models, and integrating explicitly at least global contribution of our work has been to identify, to our best
two of the three dimensions of sustainable development in the knowledge, classify and analyze all the published literature within
objective function(s) or the constraints. We therefore excluded papers the scope of survey and determine key factors of these works as
focused on only one of the dimensions or on closely related areas well as identify future directions. We have indeed identified 87
such as reverse logistics or undesirable facility location when they papers published in 41 international peer-reviewed journals,
only addressed sustainable development implicitly. Besides, fields like among which 10 addressed simultaneously the three dimensions
of sustainable development, 74 the economic and environmental weighted sum of objectives or the ε-constraint approach with
factors and only 3 were focused on both the economic and social minimization of an economic criterion and an environmental
dimensions, while no work integrates the environmental and criterion expressed as a constraint. However a significant number
social factors only. We have identified that a majority of the works of works calls for available multi-objective solution techniques
were focused on specific areas of applications, while only some of such as goal programming or metaheuristics. We identified a
the published papers addressed generic sustainable SCND models. significant lack of studies on truly multi-objective approaches
The major contribution of our work has been to analyze and with adequate consideration of uncertainties and risks (see Heck-
compare the research works and determine their key character- mann et al. [123] for a review on supply chain risk).
istics: methodologies used for environmental assessment, factors In terms of applications, besides the proposition of generic
retained for integrating social dimension, mathematical modeling models, a strong emphasis is made on process industries (biofuel,
approaches and solution methods developed, as well as the chemical processes) and on waste management problems. Such
applications developed in different sectors and types of experi- works account to about half of the published works devoted to
ment conducted with these models. specific applications. We can argue that the upstream part of a
As mentioned above, we identified that a large majority of the supply chain is often where greatest environmental impact arises
works focus on the economic and environmental factors. In con- and so this focus makes sense. However these applications reflect
trast, social aspects of sustainable development are rarely consid- highly integrated, often automated processes, whereas supply
ered in quantitative studies in comparison with environmental chains in the industrial or consumer goods areas are often
issues and even less research addresses all three dimensions decentralized and involve more uncertainty due to human factors.
together. Furthermore, there are a limited number of sub-factors It is noteworthy to remark that many sectors (automotive industry,
of the three main dimensions considered in published studies. The distribution of consumer products and transport) have not yet or
consideration of environmental factors is often limited to GHG little been considered.
emissions or energy consumption, or the consideration of social
factors is often limited to evaluation of jobs created or respect of 8.2. Research avenues
working legislation. Many possible factors proposed by specialized
works such as Chardine-Baumann and Botta-Genoulaz [105] are far These findings lead us to suggest some directions for future
from being considered. Regarding environmental factors, several theoretical and applied research works to fill the gaps found in the
performance measures have been considered to tackle environ- literature of sustainable SCND. Some of the possible future direc-
mental impacts, especially for proposing analytic measures for GHG tions are however direct consequences of the analysis of the
emissions or their global cost impact. These are the principal factors preceding paragraphs and will not be repeated here.
used in quantitative models. Simultaneously, we observed that LCA Regarding the environmental dimension, it is worthy to con-
is the dominant approach to incorporate environmental issues in sider GHG emissions relative to nodes (facilities) and arcs (trans-
SCND, but all impact categories are not considered in general. port links) of the supply chain network together with other
In contrast, the lack of published research addressing social performance measures such as waste generation or energy con-
factors together with other dimensions appears to be due to the sumption. In other words, optimizing only one criteria does not
difficulty of modeling such factors. Social factors are sometimes allow the minimization of overall environmental impact. Classical
considered indirectly within the evaluation of economic and process-based LCA is the most frequently used method to assess
environmental factors. Hence, research that is able to find a the environmental impacts. But employing this approach is some-
balance between supply chain costs and the broad spectrum of times difficult for practical reasons. Besides, LCA pays a greater
impact categories remains largely an uncharted territory to date. attention to the early stages of the life-cycle of a new product
Still the models reviewed are from integrating the characteristics development which is often before the supply chain network has
of the ISO 26000 norm. been designed. Therefore, developing novel approaches combining
Regarding modeling techniques, research concentrates on the Input-Output LCA (such as material flow analysis) and process-
development of deterministic MILP models solved with standard based approaches may better consider environmental damage
modeling tools and solvers. This is due to the ability of these throughout the entire product life-cycle. Carbon credit exchange
modeling techniques to integrate environmental or social aspects schemes (despite their current limitations) could be also be more
in complex industrial process for each particular sector. Perfor- widely considered at the strategic decision level together with
mance of state-of-the-art solvers allows solving real-life instances efforts to reduce the GHG emissions within the supply chain.
even though very long computing time are sometimes reported. Social aspects should be given more attention in future
Developing advanced heuristic solution techniques for solving research to achieve a sustainable SCND. However, developing
large-sized problems efficiently seems yet to be difficult because methodologies for quantifying the social aspects is a challenging
of the complexity of these types of problems. Indeed few works task. Their consideration at the stage of scenarios definition before
use heuristic or metaheuristic approaches. optimization may remain an effective alternative within a decision
Although uncertainty is often an intrinsic characteristic of the making process. A real challenge is probably to define the scope
studied problems, most authors still use deterministic models. One for the social impact to consider. Contrary to environmental
main reason is that large stochastic models would be intractable studies (and more especially LCA based approaches), this question
whereas deterministic models can be solved by state-of-the-art is never discussed in the papers we found addressing the social
solvers. Because of the characteristics of the addressed problems, dimension. This results in very disparate metrics, with a relative
some of the works consider non-linear models and call for specific dominance of metrics concerning employment and health
solution techniques or non-linear solvers. impacts. The generalization of LCA to the social dimension is
Sustainable SCND problems are multi-objective by nature and known as social LCA (or S-LCA). Its goal is to deliver decision-
the models that we have studied consider at least two dimensions making support related to the social impacts of products or
in the objective function or constraints and sometimes several systems (see the reviews by Jorgensen [129,130]). S-LCA was not
sub-factors. However about one third of the proposed models are used in our reference papers. This is a serious track to better
limited to a single aggregated objective, while two thirds explicitly integrate social dimension into quantitative models. Recent devel-
consider several objective functions. In terms of solution techni- opments have led the definition of the ISO 26000 norm on social
ques, however, a large majority of papers are limited to the use of a responsibility. However, due to its recent publication, there is still
no research on the impact of this standard on supply chain seek optimal solutions to models that include much uncertainty. On
practices [104,120]. This seems another fruitful research avenue. the other hand, many operational researchers sometimes use very
As already mentioned, sustainable development problems are sophisticated algorithms to solve problems with poor environmental or
clearly multi-objective problems. They cannot be expressed with a economic modeling. Solving rich environmental SCND models to
single dimension unless all factors are reduced to their cost optimality within acceptable time is still beyond the capabilities of
equivalent. Alternatively, a model focused on economic optimization current mathematical solvers. Thus, collaboration between researchers
has to consider explicit environmental or social factors as con- from various communities would result in realistic modeling of the
straints. Still very few published models handle the economic, real-life and efficient near optimal solution methods to solve the
environmental and social dimensions simultaneously. This calls for optimization problems.
the development of efficient multi-objective models and dimension-
ality reduction techniques that adequately address the different References
dimensions of sustainable development. Uncertainty and risk should
also be better considered in sustainable SCND. In real problems,
1 Abdallah T, Farhat A, Diabat A, Kennedy S. Green supply chains with carbon
uncertainty is present in many estimated factors: demand level, trading and environmental sourcing: formulation and life cycle assessment.
impact assessment, costs, social impacts, etc. The consideration of Applied Mathematical Modelling 2012;36(9):4271–85.
realistic management features such as supplier selection and risk 2 Abdallah T, Diabat A, Rigter J. Investigating the option of installing small scale
PVs on facility rooftops in a green supply chain. International Journal of
management have been frequently considered in supply chain and Production Economics 2013;146(2):465–77.
procurement research, but quantitative sustainable SCND models 3 Akgul O, Shah N, Papageorgiou LG. An optimisation framework for a hybrid first/
incorporating these features are still scarce. second generation bioethanol supply chain. Computers and Chemical Engineer-
ing 2012;42(0):101–14.
Strategic decisions such as network design have a significant 4 Alçada-Almeida L, Coutinho-Rodrigues J, Current J. A multiobjective modeling
influence on tactical and operational constraints and decisions. How- approach to locating incinerators. Socio-Economic Planning Sciences 2009;43
ever, the coordination of the different levels has been almost ignored (2):111–20.
5 Amin SH, Zhang G. A multi-objective facility location model for closed-loop
in the sustainable SCND literature. More attention should be given to supply chain network under uncertain demand and return. Applied Mathema-
integrated strategic and tactical models. Tactical decisions may have tical Modelling 2013;37(6):4165–76.
significant impacts on costs and impacts for example changes in 6 Baud-Lavigne B, Agard B, Penz B. Environmental constraints in joint product and
supply chain design optimization. Computers and Industrial Engineering
delivery frequencies improve customer satisfaction but impact on
2014;76(0):16–22.
vehicle fill rates and therefore efficiency of transport. 7 Beheshtifar S, Alimoahmmadi A. A multiobjective optimization approach for
Up to now, the literature concentrates on specific rich models location-allocation of clinics. International Transactions in Operational Research
2014. http://dx.doi.org/10.1111/itor.12088.
focused on a particular real-life application. For general industrial
8 Berger C, Savard G, Wizere A. EUGENE: an optimization model for integrated
companies, there is a need to develop generic models for sustainable regional solid waste management planning. International Journal of Environ-
SCND, such as in classical works on SCND. Generic models should ment and Pollution 1999;12(23):280–307.
include features such as multiple commodities, bill of materials, multi- 9 Bernardi A, Giarola S, Bezzo F. Spatially explicit multiobjective optimization for
the strategic design of first and second generation biorefineries including
layer supply chains and multiple periods. Assumptions such as carbon and water footprints. Industrial and Engineering Chemistry Research
capacity expansion and technology levels also deserve future research. 2013;52(22):7170–80.
Environmental impact should be measured at all steps in the supply 10 Bloemhof-Ruwaard J, Van Wassenhove L, Gabel H, Weaver P. An environmental
life cycle optimization model for the European pulp and paper industry. Omega
chain. For example GHG emissions should be considered at nodes 1996;24(6):615–29.
(production or storage facilities) and on the arcs (transport activities 11 Bojarski A, Laínez J, Espuña A, Puigjaner L. Incorporating environmental impacts
and the modes used). Studies that consider social dimension use a and regulations in a holistic supply chains modeling: an LCA approach.
Computers and Chemical Engineering 2009;33(10):1747–59.
large variety of assessment metrics and are all based on empirical case 12 Bouzembrak Y, Allaoui H, Goncalves G, Bouchriha H. A multi-modal supply
studies. This shows that we are not close to having generic models chain network design for recycling waterway sediments. International Journal
including the three dimensions of sustainable development. of Environment and Pollution 2013;51(1):15–31.
13 Caruso C, Colorni A, Paruccini M. The regional urban solid waste management
When it comes to solution techniques, standard (but powerful) system: a modelling approach. European Journal of Operational Research
solvers have been the most widely used tools to solve the resulting 1993;70(1):16–30.
models in sustainable SCND. However, the size and particularly the 14 Chaabane A, Ramudhin A, Paquet M. Designing supply chains with sustain-
ability considerations. Production Planning and Control 2011;22(8):727–41.
number of binary variables in practical SCND problems raises
http://dx.doi.org/10.1080/09537287.2010.543554.
difficulties for solving them in a reasonable amount of time. This 15 Chaabane A, Ramudhin A, Paquet M. Design of sustainable supply chains under
issue is even more crucial for adequately solving non-linear, the emission trading scheme. International Journal of Production Economics
stochastic or multi-objective models. The capability of solvers 2012;135(1):37–49.
16 Corsano G, Vecchietti A, Montagna J. Optimal design for sustainable bioethanol
practically restricts the scope of most studies. Therefore, develop- supply chain considering detailed plant performance model. Computers and
ing efficient exact or heuristic solution methods is a real need for Chemical Engineering 2011;35(8, SI):1384–98.
the future, especially for solving extension of SCND problems (e.g. 17 Costi P, Minciardi R, Robba M, Rovatti M, Sacile R. An environmentally
sustainable decision model for urban solid waste management. Waste Manage-
location-routing problems). ment 2004;24(3):277–95.
Government legislation and customers' awareness are among main 18 Datta S. Multi-criteria multi-facility location in Niwai block, Rajasthan. IIMB
reasons that prompt companies or organizations to pay an increased Management Review 2012;24(1):16–27.
19 Dehghanian F, Mansour S. Designing sustainable recovery network of end-of-
attention to environmental and social impacts of their activities. Many life products using genetic algorithm. Resources, Conservation and Recycling
major companies concentrate on their core business and outsource a 2009;53(10):559–70.
large of their production or distribution activities to subcontractors, 20 Devika K, Jafarian A, Nourbakhsh V. Designing a sustainable closed-loop supply
chain network based on triple bottom line approach: a comparison of
distributors, third party logistics providers. Thus, sustainable develop- metaheuristics hybridization techniques. European Journal of Operational
ment goals can indeed be truly achieved only by considering the Research 2014;235(3):594–615.
supply chain as a complex system with collaborating stakeholders 21 Diabat A, Abdallah T, Al-Refaie A, Svetinovic D, Govindan K. Strategic closed-
loop facility location problem with carbon market trading. IEEE Transactions on
(government, consumers and multiple companies) which address the
Engineering Management 2013;60(2):398–408.
life-cycle perspectives together. 22 Duque J, Barbosa-Povoa A, Novais A. Design and planning of sustainable
Finally, many challenging problems and solution methods have industrial networks: application to a recovery network of residual products.
been published separately by authors within the management, indus- Industrial and Engineering Chemistry Research 2010;49(9):4230–48.
23 Elhedhli S, Merrick R. Green supply chain network design to reduce carbon
trial engineering or operations research literature. For example, we emissions. Transportation Research Part D: Transport and Environment 2012;17
observed that all LCA-based approaches use standard solvers which (5):370–9.
24 Elia J, Baliban R, Xiao X, Floudas C. Optimal energy supply network determina- 52 Malczewski J, Ogrycżak W. An interactive approach to the central facility
tion and life cycle analysis for hybrid coal, biomass, and natural gas to liquid location problem: locating pediatric hospitals in Warsaw. Geographical Analysis
(CBGTL) plants using carbon-based hydrogen production. Computers and 1990;22(3):244–58.
Chemical Engineering 2011;35(8):1399–430. 53 Mallidis I, Dekker R, Vlachos D. The impact of greening on supply chain design
25 Elia JA, Baliban RC, Floudas CA. Nationwide energy supply chain analysis for and cost: a case for a developing region. Journal of Transport Geography
hybrid feedstock processes with significant CO2 emissions reduction. AIChE 2012;22:118–28.
Journal 2012;58(7):2142–54. http://dx.doi.org/10.1002/aic.13842 ISSN 1547- 54 Marufuzzaman M, Ekşioğlu SD, Hernandez R. Environmentally friendly supply
5905 URL: 〈http://dx.doi.org/10.1002/aic.13842〉. chain planning and design for biodiesel production via wastewater sludge.
26 Elia JA, Baliban RC, Floudas CA. Nationwide, regional, and statewide energy Transportation Science 2014;48(4):555–74.
supply chain optimization for natural gas to liquid transportation fuel (GTL) 55 Mele FD, GuillénGosálbez G, Jiménez L. Optimal planning of supply chains for
systems. Industrial and Engineering Chemistry Research 2014;53(13):5366–97. bioethanol and sugar production with economic and environmental concerns.
27 Erkut E, Karagiannidis A, Perkoulidis G, Tjandra SA. A multicriteria facility In: Jeżowski J, Thullie J, editors. Computer Aided Chemical Engineering, 26.
location model for municipal solid waste management in north greece. Cracow, Poland: Elsevier; 2009. p. 997–1002.
European Journal of Operational Research 2008;187(3):1402–21. 56 Mele F, Kostin A, Guillén-Gosálbez G, Jiménez L. Multiobjective model for more
28 Eskandarpour M, Zegordi S, Nikbakhsh E. A parallel variable neighborhood sustainable fuel supply chains a case study of the sugar cane industry in
search for the multi-objective sustainable post-sales network design problem. argentina. Industrial and Engineering Chemistry Research 2011;50(9):4939–58.
International Journal of Production Economics 2013;145(1):117–31. 57 Minciardi R, Paolucci M, Robba M, Sacile R. Multi-objective optimization of solid
29 Galante G, Aiello G, Enea M, Panascia E. A multi-objective approach to solid waste flows: environmentally sustainable strategies for municipalities. Waste
waste management. Waste Management 2010;30(8–9):1720–8. Management 2008;28(11):2202–12.
30 Giarola S, Zamboni A, Bezzo F. Spatially explicit multi-objective optimisation for 58 Mohammadi M, Torabi S, Tavakkoli-Moghaddam R. Sustainable hub location
design and planning of hybrid first and second generation biorefineries. under mixed uncertainty. Transportation Research Part E: Logistics and Trans-
Computers and Chemical Engineering 2011;35(9):1782–97. portation Review 2014;62(0):89–115.
31 Giarola S, Shah N, Bezzo F. A comprehensive approach to the design of ethanol 59 Mota B, Gomes A, Carvalho MI, Barbosa-Povoa AP. Towards supply chain sustain-
supply chains including carbon trading effects. Bioresource Technology ability: economic, environmental and social design and planning. Journal of Cleaner
2012;107:175–85. Production 2014, in press. http://dx.doi.org/10.1016/j.jclepro.2014.07.052.
32 Giarola S, Zamboni A, Bezzo F. Environmentally conscious capacity planning and 60 Muñoz E, Capón-García E, Laínez J, Espuña A, Puigjaner L. Considering
technology selection for bioethanol supply chains. Renewable Energy environmental assessment in an ontological framework for enterprise sustain-
2012;43:61–72. ability. Journal of Cleaner Production 2013;47:149–64.
33 Govindan K, Jafarian A, Khodaverdi R, Devika K. Two-echelon multiple-vehicle 61 Papapostolou C, Kondili E, Kaldellis J. Development and implementation of an
location-routing problem with time windows for optimization of sustainable optimisation model for biofuels supply chain. Energy 2011;36(10):6019–26.
supply chain network of perishable food. International Journal of Production 62 Pati R, Vrat P, Kumar P. A goal programming model for paper recycling system.
Economics 2014;152:9–28. Omega 2008;36(3):405–17.
34 Guillén-Gosálbez G, Grossmann I. Optimal design and planning of sustainable 63 Pérez-Fortes M, Laínez-Aguirre J, Arranz-Piera P, Velo E, Puigjaner L. Design of
chemical supply chains under uncertainty. AICHE Journal 2009;55(1):99–121. regional and sustainable bio-based networks for electricity generation using a
35 Guillén-Gosálbez G, Grossmann I. A global optimization strategy for the
multi-objective milp approach. Energy 2012;44(1):79–95.
environmentally conscious design of chemical supply chains under uncertainty 64 Pinto-Varela T, Barbosa-Póvoa A, Novais A. Bi-objective optimization approach
in the damage assessment model. Computers and Chemical Engineering
to the design and planning of supply chains: economic versus environmental
2010;34(1):42–58.
performances. Computers Chemical Engineering 2011;35(8):1454–68.
36 Guillén-Gosálbez G, Caballero J, Jiménez L. Application of life cycle assessment
65 Pishvaee M, Razmi J. Environmental supply chain network design using multi-
to the structural optimization of process flowsheets. Industrial Engineering
objective fuzzy mathematical programming. Applied Mathematical Modelling
Chemistry Research 2008;47(3):777–89.
2012;36(8):3433–46.
37 Guillén-Gosálbez G, Mele F, Grossmann I. A bi-criterion optimization approach
66 Pishvaee M, Razmi J, Torabi S. Robust possibilistic programming for socially
for the design and planning of hydrogen supply chains for vehicle use. AICHE
responsible supply chain network design: a new approach. Fuzzy Sets and
Journal 2010;56(3):650–67.
Systems 2012;206(0):1–20.
38 Harraz NA, Galal NM. Design of sustainable end-of-life vehicle recovery network
67 Pishvaee M, Torabi S, Razmi J. Credibility-based fuzzy mathematical program-
in Egypt. Ain Shams Engineering Journal 2011;2(3–4):211–9.
ming model for green logistics design under uncertainty. Computers and
39 Harris I, Naim M, Palmer A, Potter A, Mumford C. Assessing the impact of cost
Industrial Engineering 2012;62(2):624–32.
optimization based on infrastructure modelling on CO2 emissions. International
68 Pourmohammadi H, Rahimi M, Dessouky M. Sustainable reverse logistics for
Journal of Production Economics 2011;131(1):313–21.
distribution of industrial waste/byproducts: a joint optimization of operation
40 Hugo A, Pistikopoulos E. Environmentally conscious long-range planning and design
and environmental costs. Supply Chain Forum: An International Journal 2008;9
of supply chain networks. Journal of Cleaner Production 2005;13(15):1471–91.
41 Hugo A, Rutter P, Pistikopoulos S, Amorelli A, Zoia G. Hydrogen infrastructure (1):2–17.
69 Pozo C, Ruiz-Femenia R, Caballero J, Guillén-Gosálbez G, Jimenez L. On the use
strategic planning using multi-objective optimization. International Journal of
Hydrogen Energy 2005;30(15):1523–34. of principal component analysis for reducing the number of environmental
42 Jamshidi R, Fatemi Ghomi S, Karimi B. Multi-objective green supply chain objectives in multi-objective optimization: application to the design of chemical
optimization with a new hybrid memetic algorithm using the Taguchi method. supply chains. Chemical Engineering Science 2012;69(1):146–58.
Scientia Iranica 2012;19(6):1876–86. 70 Quariguasi Frota Neto J, Bloemhof-Ruwaard J, van Nunen J, van Heck E.
43 Kannan D, Diabat A, Alrefaei M, Govindan K, Yong G. A carbon footprint based Designing and evaluating sustainable logistics networks. International Journal
reverse logistics network design model. Resources, Conservation and Recycling of Production Economics 2008;111(2):195–208.
2012;67(0):75–9. 71 Ramudhin A, Chaabane A, Paquet M. Carbon market sensitive sustainable
44 Kanzian C, Kühmaier M, Zazgornik J, Stampfer K. Design of forest energy supply supply chain network design. International Journal of Management Science
networks using multi-objective optimization. Biomass and Bioenergy 2013;58 and Engineering Management 2010;5(1):30–8.
(0):294–302. 72 Ruiz-Femenia R, Guillén-Gosálbez G, Jiménez L, Caballero J. Multi-objective
45 Kostin A, Guillén-Gosálbez G, Mele F, Jiménez L. Identifying key life cycle optimization of environmentally conscious chemical supply chains under
assessment metrics in the multiobjective design of bioethanol supply chains demand uncertainty. Chemical Engineering Science 2013;95:1–11.
using a rigorous mixed-integer linear programming approach. Industrial and 73 Sadrnia A, Ismail N, Zulkifli N, Ariffin MKA, Nezamabadi-pour H, Mirabi H. A
Engineering Chemistry Research 2012;51(14):5282–91. multiobjective optimization model in automotive supply chain networks.
46 Krikke H. Impact of closed-loop network configurations on carbon footprints: a Mathematical Problems in Engineering 2013:10 (Article ID 823876).
case study in copiers. Resources, Conservation and Recycling 2011;55 74 Saffar MM, Shakouri HG, Razmi J. A new bi-objective mixed integer linear
(12):1196–205. programming for designing a supply chain considering CO2 emission. Uncertain
47 Krikke H, Bloemhof-Ruwaard J, Van Wassenhove LN. Concurrent product and Supply Chain Management 2014;2(4):275–92.
closed-loop supply chain design with an application to refrigerators. Interna- 75 Saffar MM, Shakouri HG, Razmi J. A new multi objective optimization model for
tional Journal of Production Research 2003;41(16):3689–719. designing a green supply chain network under uncertainty. International
48 Lam H, Wendy Ng W, Ng R, Ng E, Abdul Aziz MK, Ng D. Green strategy for Journal of Industrial Engineering Computations 2015;6(1):15–32.
sustainable waste-to-energy supply chain. Energy 2013;57(0):4–16. 76 Santibanez-Aguilar JE, González-Campos JB, Ponce-Ortega JM, Serna-González
49 Lira-Barragán LF, Ponce-Ortega JM, Serna-González M, El-Halwagi MM. An M, El-Halwagi MM. Optimal planning and site selection for distributed multi-
MINLP model for the optimal location of a new industrial plant with simulta- product biorefineries involving economic environmental and social objectives.
neous consideration of economic and environmental criteria. Industrial Engi- Journal of Cleaner Production 2014;65(0):270–94.
neering Chemistry Research 2011;50(2):953–64. 77 Tuzkaya G, Gülsün B, Önsel S. A methodology for the strategic design of reverse
50 Lira-Barragán LF, Ponce-Ortega JM, Nápoles-Rivera F, Serna-González M, El- logistics networks and its application in the Turkish white goods industry.
Halwagi MM. Incorporating property-based water networks and surrounding International Journal of Production Research 2011;49(15):4543–71.
watersheds in site selection of industrial facilities. Industrial and Engineering 78 Verma M, Gendreau M, Laporte G. Optimal location and capability of oil-spill
Chemistry Research 2013;52(1):91–107. response facilities for the south coast of Newfoundland. Omega 2013;41
51 Liu P, Whitaker A, Pistikopoulos EN, Li Z. A mixed-integer programming (5):856–67.
approach to strategic planning of chemical centres: a case study in the UK. 79 Wang F, Lai X, Shi N. A multi-objective optimization for green supply chain
Computers and Chemical Engineering 2011;35(8):1359–73. network design. Decision Support Systems 2011;51(2):262–9.
80 Xifeng T, Ji Z, Peng X. A multi-objective optimization model for sustainable 107 Chopra S, Meindl P. Supply chain management. 2nd ed.Pearson Prentice-Hall;
logistics facility location. Transportation Research Part D: Transport and 2004.
Environment 2013;22:45–8. 108 Daskin MS, Snyder LV, Berger R. Facility location in supply chain design. In:
81 You F, Wang B. Life cycle optimization of biomass-to-liquid supply chains with Langevin A, Riopel D, editors. Logistics systems. US: Springer; 2005. p. 39–65.
distributed centralized processing networks. Industrial and Engineering Chem- 109 DeMeyer A, Cattrysse D, Rasinmäki J, Orshoven JV. Methods to optimise the
istry Research 2011;50:10102–27. design and management of biomass-for-bioenergy supply chains: a review.
82 You F, Tao L, Graziano D, Snyder S. Optimal design of sustainable cellulosic Renewable and Sustainable Energy Reviews 2014;31(0):657–70.
biofuel supply chains: multiobjective optimization coupled with life cycle 110 Dekker R, Fleischmann M, Interfuth K, van Wassenhove L, editors. Reverse
assessment and input-output analysis. AIChE Journal 2012;58(4):1157–80. logistics: quantitative models for closed loops supply chains. Berlin and
83 Yue D, Kim MA, You F. Design of sustainable product systems and supply chains with Heidelberg, Germany: Springer-Verlag; 2004.
life cycle optimization based on functional unit: general modeling framework, mixed- 111 Dekker R, Bloemhof J, Mallidis I. Operations Research for green logistics – an
integer nonlinear programming algorithms and case study on hydrocarbon biofuels. overview of aspects, issues, contributions and challenges. European Journal of
ACS Sustainable Chemistry and Engineering 2013;1(8):1003–14. Operational Research 2012;219(3):671–9.
84 Yue D, Slivinsky M, Sumpter J, You F. Sustainable design and operation of 112 Dua V, Pistikopoulos EN. An algorithm for the solution of multiparametric
cellulosic bioelectricity supply chain networks with life cycle economic, mixed integer linear programming problems. Annals of Operations Research
environmental, and social optimization. Industrial and Engineering Chemistry 2000;99(1–4):123–39.
Research 2014;53(10):4008–29. 113 Fleischmann M, Bloemhof-Ruwaard J, Dekker R, Laan E, van Nunen J, Van
85 Zamboni A, Bezzo F, Shah N. Spatially explicit static model for the strategic Wassenhove L. Quantitative models for reverse logistics: a review. European
design of future bioethanol production systems. II. Multi-objective environ- Journal of Operational Research 1997;103(1):1–17.
mental optimization. Energy and Fuels 2009;23(10):5134–43. 114 Goedkoop M, Spriensma R. The Eco-Indicator 99, a damage oriented method
86 Zamboni A, Murphy R, Woods J, Bezzo F, Shah N. Biofuels carbon footprints: for life cycle impact assessment: Methodology report, third edition. Technical
whole-systems optimisation for ghg emissions reduction. Bioresource Technol- Report. PRé Consultants; 2000.
ogy 2011;102(16):7457–65. 115 Goedkoop M, Heijungs R, Huijbregts M, De Schryver A, Struijs J, Van Zelm R.
87 Zhang M, Wiegmans B, Tavasszy L. Optimization of multimodal networks including ReCiPe 2008, a life cycle impact assessment method which comprises harmo-
environmental costs: a model and findings for transport policy. Computers in nised category indicators at the midpoint and the endpoint level. first edition.
Industry 2013;64(2):136–45. report i: Characterisation. Technical Report, Rijksinstituut voor Volksgezond-
heid en Milieu, The Netherlands; 2009.
116 Guide VDR, Van Wassenhove LN. Closed-loop supply chains: an introduction to
the feature issue. Part I. Production and Operations Management 2006;15
(3):345–50.
117 Guide VDR, Van Wassenhove LN. Closed-loop supply chains: an introduction to
Further references the feature issue. Part II. Production and Operations Management 2006;15
(4):471–2.
88 Ahi P, Searcy C. An analysis of metrics used to measure performance in green 118 Guillén-Gosálbez G. A novel MILP-based objective reduction method for multi-
and sustainable supply chains. Journal of Cleaner Production 2015;86:360–77. objective optimization: application to environmental problems. Computers
89 Akçali E, Çetinkaya S, Üster H. Network design for reverse and closed-loop and Chemical Engineering 2011;35(8):1469–77.
supply chains: an annotated bibliography of models and solution approaches. 119 Gupta S, Palsule-Desai O. Sustainable supply chain management: review and
Networks 2009;53(3):231–48. research opportunities. IIMB Management Review 2011;23(4):234–45.
90 Aras N, Boyaci T, Verter V. Designing the reverse logistics network. In: Ferguson 120 Hahn R. ISO 26000 and the standardization of strategic management processes
M, Souza G, editors. Closed loop supply chains. Taylor and Francis, CRC Press; for sustainability and corporate social responsibility. Business Strategy and the
2010. p. 67–98, http://www.crcpress.com/product/isbn/9781420095258. Environment 2013;22(7):442–55.
91 Arioglu Salmona MO, Selam A, Vayvay O. Sustainable supply chain manage- 121 Hakimi SL. Optimum locations of switching centers and the absolute centers
ment: a literature review. In: International conference on value chain sustain- and medians of a graph. Operations Research 1964;12(3):450–9.
ability (ICOVACS 2010), Valencia, Spain; 2010. 122 Hassini E, Surti C, Searcy C. A literature review and a case study of sustainable
92 Ashby A, Leat M, Hudson-Smith M. Making connections: a review of supply supply chains with a focus on metrics. International Journal of Production
chain management and sustainability literature. Supply Chain Management: Economics 2012;140(1):69–82.
An International Journal 2012;17(5):497–516. 123 Heckmann I, Comes T, Nickel S. A critical review on supply chain risk -
93 Autry C, Goldsby T, Bell J, Moon M, Munson C, Watson M, Lewis S, Cacioppi P, definition measure and modeling. Omega 2015;52(0):119–32.
Jayaraman J. The definitive guide to modern supply chain management 124 Ilgin M, Gupta S. Environmentally conscious manufacturing and product
(collection). Upper Saddle River, New Jersey: Pearson Education; 2013. recovery ECMPRO: a review of the state of the art. Journal of Environmental
94 Awudu I, Zhang J. Uncertainties and sustainability concepts in biofuel supply Management 2010;91:563–91.
chain management: a review. Renewable and Sustainable Energy Reviews 125 IPCC, Fourth assessment report: climate change 2007: working group I report:
2012;16(2):1359–68. the physical science basis. Geneva: IPCC; 2007.
95 Azapagic A. Life cycle assessment and its application to process selection, 126 ISO. Environmental management-life cycle assessment-principles and frame-
design and optimisation. Chemical Engineering Journal 1999;73(1):1–21. work; 2006.
96 Barbosa-Póvoa AP. Process supply chains management – where are we? where 127 Johnsen T, Miemczyk J, Macquet M. Sustainable purchasing and supply
to go next? Frontiers in Energy Research 2014;2(23). management. Supply Chain Management 2012;17(5):478–96.
97 Beamon BM. Supply chain design and analysis: models and methods. Interna 128 Jolliet O, Margni M, Charles R, Humbert S, Payet J, Rebitzer G, Rosenbaum R.
tional Journal of Production Economics 1998;55(3):281–94. Impact 2002þ: a new life cycle impact assessment methodology. The Inter-
98 Benoi\widehatt-Norris C. Data for social lca. The International Journal of Life national Journal of Life Cycle Assessment 2003;8(6):324–30.
Cycle Assessment 2014;19(2):261–5. 129 Jorgensen A. Social LCA – a way ahead? The International Journal of Life Cycle
99 Beske P, Land A, Seuring S. Sustainable supply chain management practices Assessment 2013;18:296–9.
and dynamic capabilities in the food industry: a critical analysis of the 130 Jorgensen A, Bocq A, Nazarkina L, Hauschild M. Methodologies for social life
literature. International Journal of Production Economics 2014;152:131–43. cycle assessment. The International Journal of Life Cycle Assessment
100 Bostel N, Dejax P, Lu Z. In: Langevin A, Riopel D, editors. Logistics systems: 2008;13:96–103.
design and optimization, chapter the design, planning and optimization of 131 Kumar A, Jain V, Kumar S. A comprehensive environment friendly approach for
reverse logistic networks. US: Springer: Logistics systems: design and optimi- supplier selection. Omega 2014;42(1):109–23.
zation; 2005. p. 171–212. 132 Meixell MJ, Gargeya VB. Global supply chain design: a literature review and
101 Boukherroub T, Fondrevelle J, Guinet A, Ruiz A. Multi-criteria decision making for critique. Transportation Research Part E: Logistics and Transportation Review
the supply chain design: A review with emphasis on sustainable supply chains. In 2005;41(6):531–50.
Creative logistics for an uncertain world. Proceedings of the 4th International 133 Masoumik SM, Abdul-Rashid SH, UdoncyOlugu E, Ghazilla RAR. Sustainable
Conference on Information Systems, Logistics and Supply Chain; 2012. supply chain design: a configurational approach. The Scientific World Journal
102 Brandenburg M, Govindan K, Sarkis J, Seuring S. Quantitative models for 2014;2014:16.
sustainable supply chain management: developments and directions. Eur- 134 Melachrinoudis E. The location of undesirable facilities. In: Eiselt HA, Marianov
opean Journal of Operational Research 2014;233(2):299–312. V, editors. Foundations of location analysis of international series in operations
103 Burton I. Our common future – world commission on environment and research and management science, vol. 155. US: Springer; 2011. p. 207–39.
development. Environment 1987;29(5):25–9. 135 Melo MT, Nickel S, Saldanha-daGama F. Facility location and supply chain
104 Castka P, Balzarova MA. Iso 26000 and supply chains – on the diffusion of the management – a review. European Journal of Operational Research 2009;196
social responsibility standard. International Journal of Production Economics (2):401–12.
2008;111(2):274–86. 136 Messac A, Ismail-Yahaya A, Mattson C. The normalized normal constraint
105 Chardine-Baumann E, Botta-Genoulaz V. A framework for sustainable perfor- method for generating the pareto frontier. Structural and Multidisciplinary
mance assessment of supply chain management practices. Computers and Optimization 2013;25(2):86–98.
Industrial Engineering 2014;76(0):138–47. 137 Miemczyk J, Johnsen T, Macquet M. Sustainable purchasing and supply
106 Chen L, Olhager J, Tang O. Manufacturing facility location and sustainability: a management: a review of definitions and measures at the dyad, chain and
literature review and research agenda. International Journal of Production network levels of analysis. Supply Chain Management: An International
Economics 2014;149(0):154–63. Journal 2012;17(5):478–96.
138 Montibeller G, Belton V, Ackermann F, Ensslin L. Reasoning maps for decision 152 Seuring S, Müller M. From a literature review to a conceptual framework for
aid: an integrated approach for problem-structuring and multi-criteria evalua- sustainable supply chain management. Journal of Cleaner Production 2008;16
tion. Journal of the Operations research Society 2008;59(5):575–89. (15):1699–710.
139 Nikolopoulou A, Ierapetritou M. Optimal design of sustainable chemical 153 Simchi-Levi D, Kaminsky P, Simchi-Levi E, editors. Managing the supply chain:
processes and supply chains: a review. Computers and Chemical Engineering the definitive guide for the business professional. New York: McGraw-Hill;
2012;44:94–103. 2004.
140 OECD/IEA. CO2 emissions from fuel combustion 2012-highlights. Technical 154 Snyder LV. Facility location under uncertainty: a review. IIE Transactions
Report. OECD/IEA, 2012. URL: https://www.iea.org/co2highlights/co2high 2004;38:547–64.
lights.pdf (accessed March 28, 2014). 155 Soysal M, Bloemhof-Ruwaard JM, Meuwissen MP, van der Vorst JG. A review
141 Owen SH, Daskin MS. Strategic facility location: a review. European Journal of on quantitative models for sustainable food logistics management. Interna-
Operational Research 1998;111(3):423–47. tional Journal on Food System Dynamics 2012;3(2):136–55.
142 Pati R, Jans R, Tyagi R. Green logistics network design: a critical review. In: 156 Srivastava S. Green supply-chain management: a state-of-the-art literature
24th Annual POMS conference. Denver, USA; 2013. review. International Journal Of Management Reviews 2007;9(1):53–80.
143 Pieragostini C, Mussati MC, Aguirre P. On process optimization considering LCA 157 Srivastava SK. Network design for reverse logistics. Omega 2008;36(4):535–48.
158 Terouhid SA, Ries R, Fard MM. Towards sustainable facility location – a
methodology. Journal of Environmental Management 2012;96(1):43–54.
literature review. Journal of Sustainable Development 2012;5(7).
144 Pishvaee M, Zanjirani Farahani R, Dullaert W. A memetic algorithm for bi-
159 Weber A, Peik G. Über den Standort der Industrie; 1909.
objective integrated forward/reverse logistics network design. Computers and
160 Wright LA, Kemp S, Williams I. Carbon footprinting: towards a universally
Operations Research 2010;37(6):1100–12.
accepted definition. Carbon Management 2011;2(1):61–72.
145 Pokharel S, Mutha A. Perspectives in reverse logistics: a review. Resources,
161 Yue D, You F, Snyder SW. Biomass-to-bioenergy and biofuel supply chain
Conservation and Recycling 2009;53(4):175–82.
optimization: overview, key issues and challenges. Computers and Chemical
146 Ramos TRP, Gomes MI, Barbosa-Póvoa AP. Planning a sustainable reverse
Engineering 2014;66(0):36–56 [selected papers from ESCAPE-23 (European
logistics system: balancing costs with environmental and social concerns. symposium on computer aided process engineering – 23). Lappeenranta,
Omega 2014;48:60–74. Finland; 9–12 June 2013].
147 Saaty TL. How to make a decision: the analytic hierarchy process. European 162 Yusuf YY, Gunasekaran A, Musa A, El-Berishy NM, Abubakar T, Ambursa HM.
Journal of Operational Research 1990;48(1):9–26. The UK oil and gas supply chains: an empirical analysis of adoption of
148 Sahin G, Süral H. A review of hierarchical facility location models. Computers sustainable measures and performance outcomes. International Journal of
and Operations Research 2007;34(8):2310–31. Production Economics 2013;146(2):501–14.
149 Sarkis J, Zhu Q, Lai K. An organizational theoretic review of green supply chain 163 Zadeh L. Fuzzy sets as a basis for a theory of possibility. Fuzzy Sets and Systems
management literature. International Journal of Production Economics 1978;1:3–28.
2011;130(1):1–15. 164 Zailani S, Jeyaraman K, Vengadasan G, Premkumar R. Sustainable supply chain
150 Seman N, Zakuan N, Jusoh A, Arif M. Green supply chain management: a management (SSCM) in Malaysia: a survey. International Journal of Production
review and research direction. International Journal of Managing Value and Economics 2012;140(1):330–40.
Supply Chains 2012;3(1). 165 Zanjirani Farahani R, Rezapour S, Drezner T, Fallah S. Competitive supply chain
151 Seuring S. A review of modeling approaches for sustainable supply chain network design: an overview of classifications models solution techniques and
management. Decision Support Systems 2013;54(4):1513–20. applications. Omega 2014;45(0):92–118.

Sustainable Supply Chain Network Design

Uploaded by

Copyright:

Available Formats

Sustainable Supply Chain Network Design

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sustainable Supply Chain Network Design

Uploaded by

Copyright:

Available Formats

Omega 54 (2015) 11–32

Contents lists available at ScienceDirect

Sustainable supply chain network design:

Article Facility location Sustainability Scope or special focus

Supply chain management

Terouhid et al. [158] Socially responsible location

Nikolopoulou and Ierapetritou [139] ○ ○ ○ 12

Fig. 2. Distribution of reference papers by journal.

in chemical and process industry. Many papers are published in

2.4. The 3 dimensions of sustainable development

The reference papers do not all address the 3 dimensions of

3.1.2. Life-cycle impact assessment

Fig. 5. LCA scopes.

CML92 Bloemhof-Ruwaard et al. [10]

ReCiPe Mota et al. [59]

Biochemical Oxygen Demand Mele et al. [55]

Water footprint Bernardi et al. [9]

Table 6 pollution. In Harris et al. [39], the quantity of energy used is a

Table 7 indicator. Since the social impact is often qualitative by nature, it is

Table 8 planning horizon. Other factors such as transportation costs or the

Table 9 consists of locations and technologies for transfer stations, mate-

In summary, a large variety of modeling techniques have been

Type of method Articles

Modeling Reference papers

Fig. 6. Review of industrial applications.

You might also like