You are on page 1of 24

Soft Computing (2020) 24:155–178

https://doi.org/10.1007/s00500-019-04289-5 (0123456789().,-volV)(0123456789().
,- volV)

FOCUS

Multi-objective optimization of cost-effective and customer-centric


closed-loop supply chain management model in T-environment
Arindam Garai1 • Tapan K. Roy2

Published online: 19 August 2019


 Springer-Verlag GmbH Germany, part of Springer Nature 2019

Abstract
This article presents one real-life-based cost-effective and customer-centric closed-loop supply chain management model.
The review of the existing literature identifies the classical performance indicators to any supply chain management model
as the aggregate revenue, the customer satisfaction and the environmental concern. However, this review fails to find a
single optimization-based supply chain management model that considers these three indicators, simultaneously. In this
article, the proposed model maximizes the customer-satisfaction index and the aggregate revenue both under the envi-
ronmental considerations via the reverse chain, whereas many existing studies took the reverse chain and the associated
subsidies into account; this is the first mathematical model that optimizes the customer-satisfaction index, at the same time.
This article employs the T-set that represents the inherent impreciseness to objective functions to the proposed model. The
corresponding optimal values are superior than stipulated goals to both the objective functions in T-environment. The
managerial insights extracted from sensitivity analysis of parameters suggest the managers to stabilize the environmental
concern and the customer satisfaction, while ensuring the cost-effectiveness in real-life-based T-environment. Also, this
analysis finds that the subsidy assists any supply chain to sustain, only if it is offered without any break and within the
optimally determined bounds.

Keywords Customer satisfaction  Environmental concerns  Closed-loop supply chain  T-set  Subsidy 
Multi-objective optimization

1 Introduction Beamon 2000). In corporate sectors, the organizations, like


planning, manufacturing, distribution, marketing and pur-
A supply chain is a set of facilities, provisions, clients, chasing, operate independently. However, the traditional
goods and methods of regulating inventory, acquiring and role of one organization often conflicts with another (Li
distribution. Traditionally, supply chain begins with the 2014). So, the development of a comprehensive mecha-
acquisition of unrefined raw materials by suppliers and nism by integrating all diverse functionaries turns out to be
ends with the consumptions of finished goods by customers essential. Accordingly, researchers introduced the supply
(Anderson 2014). In a supply chain, this flow of goods chain management (SCM) models to accomplish various
between suppliers and customers passes through the layers. tasks, such as to forecast the forthcoming demands, to
Each layer consists of several components (Sabri and schedule synchronized production of goods and to design
new products (Attaran and Sharmin 2007; Anderson 2014).
A survey of 421 organizations in ten countries by Deloitte
Communicated by O. Castillo, D. K. Jana.
(2019), at which SCM leaders reported an above average
& Arindam Garai revenue growth of 79%, augments strength to SCM
fuzzy_arindam@yahoo.com concepts.
Tapan K. Roy The livelihood of many hundreds of thousands of
roy_t_k@yahoo.co.in stakeholder hinges on the performance of supply chain. So,
1
supply chain network (SCN) design problem is a major
Department of Mathematics, Sonarpur Mahavidyalaya, comprehensive strategic decision-making problem.
Kolkata 700149, India
2
Accordingly, the sustainability that pushes for long-term
Department of Mathematics, IIEST, Howrah 711103, India

123
156 A. Garai, T. K. Roy

Table 1 Summary of recent SCM researches on CS and EC


No. References Key findings Research gaps

A. CS in SCM studies
1. Chuang and Lin (2013) Customer information quality impacts CS Used only subjective measures. Based in Taiwan,
only
2. Wei et al. (2013) Developed unique marketing strategies to SCM model Focused on specific customer data in single
country
3. Choudhury and Harrigan Discussed changes in communications between Small sample size. Skipped organization size and
(2014) customers and marketers employees’ strength
4. Garrido-Moreno and Garcı́a- Found effects of the organizational commitments Variable measurement based on perceptions of
Morales (2014) only general managers
5. Bahari and Elayidom (2015) Predicted the behaviour of customers Focused only on neuro-fuzzy classifiers
6. Zablah et al. (2016) Established reciprocal relation between the CS and job Could hurt morale to frontline employees
satisfaction
7. Agnihotri et al. (2017) Deep insights to firms on optimized sales and service Employee effectiveness impacted outcomes. No
outcomes longitudinal study
8. Aquilani et al. (2017) Three different clusters of articles on CS Limited number of databases, search engines and
keywords
9. Campos et al. (2018) Psychology focused definition of on-site customers’ Neither empirical nor analytical. Fewer
experience evaluation of tourism experiences
10. Secchi et al. (2018) Reliable multi-item measurement scale of customer Less applicable to budget ended hotels. More
service analysis required
11. Zhang et al. (2019) Mass customization and product modularity improved Only empirically tested conceptual model. No
quality integration analytical study
12. Sarkar (2019) Decreased transportation and setup time Difficult to implement in real life
B. EC in SCM studies
13. Gaussin et al. (2013) Deliberated on environmental footprint index of goods No future research direction towards the ideal
sustainability index
14. Kronborg et al. (2013) Standard carbon footprint measurement methods of No answer to challenges of suppliers than the
goods secondary data sources
15. Govindan et al. (2014) Found research gaps through 382 articles in GSCM Less attention to multi-objective models and
integration of various levels
16. Chin et al. (2014) Review of EC of GSCM models Considered only three types of threats to
organizations
17. Jabbour and Jabbour (2016) Relationships between green human resource No solution on green human resource
management and GSCM management and GSCM issues
18. Uygun and Dede (2016) Measured GSCM performance by fuzzy-based Solely one type of company. No integration of
approaches mathematical methods
19. Geng et al. (2017) Conducted meta-analysis of 25,680 effect sizes from No relationship between adoption of GSCM goals
11,127 companies on GSCM and performance metrics
20. Mumtaz et al. (2018) Linear regression framework of CLSCM to measure Only empirical results. Based on a country, not
firm performance in Pakistan well-implemented CLSCM
21. Kaur et al. (2018) DEMATEL-based GSCM approach in seven Canadian Several drawbacks in DEMATEL approach.
firms Considered only Canada
22. Alinaghian and Zamani Bi-objective model for green inventory routing Less applicable in real-life industries due to no
(2019) problem impreciseness in data
23. Kusi-Sarpong et al. (2018) Many insights into sustainable innovation Based on 5 Indian manufacturing companies only
management in SCM study
Present study This article formulates one CLSCM model by simultaneously incorporating the three major performance
indicators, viz. the AR, the EC and the CS. This determines the Pareto optimal solution to the proposed
model by employing the T-set-based multi-objective optimization method. Besides, the sensitivity
analysis of parameters presents several managerial insights

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 157

multi-dimensional viability of organizations is the single CLSCM model can be ethical, like to reflect the values of
major scientific, economic, political, social and environ- decision-makers (DM), or commercial, like to achieve
mental indicator to measure the performance of any supply advantageous position by signalling the EC to customers
chain. Researchers recognize several performance indica- (Testa and Iraldo 2010; Bloemhof and Soysal 2017).
tors to SCM models, such as aggregate revenue (AR), However, there are very few attempts to optimize the
environmental concerns (EC), customer satisfaction (CS), performance indicators through much general and broader
adoption of recent technologies, corporate social respon- look at overarching core topics of CLSCM (Aven 2016).
sibility (CSR) and collaboration. These indicators deter- To the best of the present knowledge, there is not a single
mine the efficiency of existing SCM models by comparing optimization-based study of CLSCM model, in which the
with alternative models and thereby design more effective researchers considered the above three performance indi-
SCM models. Among all the indicators, researchers clas- cators, simultaneously. This article proposes to develop an
sify the AR, CS and EC to be the classical and most integrated logistics, planning control and multi-objective
influential indicators (Seuring and Muller 2008). optimization-based analytical model that systematically
Firstly, the focus is on the revenue of supply chain. determines the maximum AR of chain by augmenting the
Since time immemorial, one single organization and the RC layer, along with the highest percentage of demands to
cluster of organizations mainly focus on the economic be met within stipulated access time.
aspects, i.e. to earn more revenue. There are countless Again, this simultaneous consideration of performance
existing studies that had put thrust on maximum AR of criteria yields a multi-objective optimization problem
organizations (Khansalar et al. 2015). So, this article (MOOP). Review of well-established articles on MOOP
emphasizes on the two other major aspects, like the CS and finds numerous methods to systematically and simultane-
the EC, to CLSCM models, in succession. Researchers find ously optimize the collection of objective functions. In
that long-term success and competitiveness of organiza- many of these optimization methods, researchers use fuzzy
tions breathe through the CS (Hennig and Klee 1997). High set theory (Bellman and Zadeh 1970; Ebrahimnejad and
satisfaction post-purchasing of products and associated Verdegay 2018), whereas the advantages of applying fuzzy
services is one central determinant to customer retention set-based optimization methods to solve MOOP are man-
and to subsequent addition of fresh customers (Rust and ifold (Wu et al. 2015); these authors documented several
Zahorik 1993). In case, customers make considerable effort drawbacks to classical fuzzy set (Garai et al. 2016, 2017).
to get products, and the satisfaction index is higher than Subsequently, a new set, viz. T-set, was born to supersede
that associated with modest effort (Cardozo 1965). Besides, fuzzy set. This T-set represents the impreciseness much
quality and in-time delivery failing short of expectation efficiently than classical fuzzy set (Garai et al. 2016, 2017).
have stronger impacts on satisfaction and repurchase In addition, these authors presented one generalized multi-
intentions of customers than quality exceeding the antici- objective optimization method to determine the Pareto
pation (Anderson et al. 1994; Angelova and Zekiri 2011). optimal solution to MOOP in T-environment (Garai et al.
Bequeathing a habitable earth to future generations turns 2017).
out as major concern to civilized society. Of late, the The most important contributions to fill in existing
existing and potential customers are ever more learning the research gaps are as follows
consciences followed by organizations. When the organi-
• To formulate one CLSCM model by simultaneously
zations adopt various environment-friendly policies, they
incorporating the three major performance indicators,
receive attractive financial and other incentives from a
like the AR, the EC and the CS;
number of agencies (Attaran and Sharmin 2007). More-
• To determine the Pareto optimal solution to the
over, the recent technological progresses contribute to
proposed model by employing the T-set-based multi-
rapid advancements to the reusability of goods. So, in
objective optimization method;
recent years, the EC comes to the forefront of debate of
• To perform the sensitivity analysis of parameters to the
global and local social interests. Also, the number of
proposed model and present the key managerial
organizations to contemplate the integration of environ-
insights.
mentally sound practices into strategic plans and daily
operations has steadily grown over years and across The organization of rest of this article is as follows.
geographies. Section 2 reviews the existing literature of SCM model
When any SCM study considers the EC, this is a closed- based on the CS and the EC aspects. Also, this section
loop supply chain management (CLSCM) model. Here, the briefly surveys the multi-objective optimization models in
word ‘closed-loop’ refers to the integration of various fuzzy environment. This forms the basis of the present
activities of the manufacturing chain layer (MC layer) and research. Section 3 presents the problem definitions and the
the reverse chain layer (RC layer). The motivation behind underlying assumptions. The specifications to the proposed

123
158 A. Garai, T. K. Roy

model are in Sect. 4. Section 5 formulates the proposed studies. The authors considered the principal logistics
multi-objective CLSCM model by assembling two objec- and SCM journals with time horizon of 20 years. They
tive functions and constraints. Next, Sect. 6 presents the found SCM research to evolve from a perspective and
existing definitions necessary to solve the proposed model. stand-alone study in social and environmental areas
Besides, this section discusses the multi-objective opti- through CSR perspective.
mization method in T-environment. The numerically • Walker and Jones (2012) described the organizations to
illustration to the proposed model is in Sect. 7. This section be mapped onto a typology of approaches to SCM. DM
compares the optimal values to objective functions could get useful lessons to smoothly implement the
between T-environment and fuzzy environment. The sen- SCM policies. Also, this article made several predic-
sitivity analysis of parameters to the proposed model along tions on future of SCM policies within leading
with several key managerial insights is in Sect. 8. Sec- enterprises.
tion 9 presents the conclusions and the future scopes of • Seuring (2013) summarized the research on quantitative
research. models of forward supply chains, which were published
in last fifteen years. The author could identify only 36
quantitative models from more than 300 articles on
2 Review of literature SCM strategies. Around same time, Seuring and Gold
(2013) asked for more research on SCM policies by
The development of electronic data interchange system adding the financial, social and environmental concerns.
initiated the era of SCM study around 1960. The intro- • In recent years, Wieland et al. (2016) presented the
duction of enterprise resource planning system advanced three-phase research study of SSCM models. The
this research through 1990s (Cardozo 1965). Keith Oliver, authors began with open-ended collection of opinions
a consultant at Booz Allen Hamilton, introduced the term, on research themes from 102 researchers. This article
SCM, to public domain (Markova and Tatjana 2013). identified several under-researched SCM areas, like
Subsequently, the SCM policies rose to prominence across employees, ethical issues and internal integration.
organizations. Operations managers began to use the term • Dubey et al. (2016) raised the voice in favour of using
‘SCM’ in the headings with increasing regularity (Markova total interpretive structural modelling in SCM models.
and Tatjana 2013). However, the SCM strategies turned The authors determined different drivers at various
much acceptable only after the publication of a seminal levels. They found that the transitive links of one
book, entitled, ‘Introduction to Supply Chain Management’ particular driver influenced another.
by Handfield and Ernest (1999). Around beginning of this • Melnyk et al. (2018) applied Dickens’ contrast between
millennium, researchers initiated systematic investigation the best of times and the worst of times as the frame.
of economic, social and environmental impacts of SCM Then, the authors hence assessed the current state of
strategies (Janssen et al. 2015). Across several decades, empirical research in operations and SCM. The goal
researchers performed numerous studies on various aspects was to find how the operations and SCM researchers
of CLSCM. Following list contains several existing and could learn from worst of times, so as to take
well-established CLSCM studies advantages of the best of times.
• Brandenburg et al. (2018) summarized the findings of
• Svensson (2007) presented an empirical study in order
several reviews of the SCM literature. The authors
to illustrate the aspects of SCM through the expansion
presented the propositions and guidelines of future
of existing theories. The author introduced several new
SCM research. Whereas they categorized the SCM
terms, like first-, second- and n-order supply chains.
models by dimensions, sustainability and research foci,
Besides, he observed the first-order supply chain of
there was a useful comparison between analytical and
brand-new clothing to go beyond the traditional point of
empirical research on SCM.
consumption.
• Jonkman et al. (2019) presented a general description of
• Seuring and Muller (2008) reviewed the existing SCM
SCN design problem in agri-food industrial chains. The
studies by considering 191 articles, published between
authors considered the role of seasonality, harvesting
1994 and 2007. The authors found a considerable rise of
decisions, perishability and timely processing in the
academic and corporate interests in CLSCM policies
Netherlands. Besides, they considered the imprecise-
within this time. Here, they integrated the three
ness in demand and harvest in stochastic version of this
classical dimensions of sustainability, viz. financial
model.
performance, CS and EC.
• In a multi-stage production system, Sarkar (2019)
• Carter and Easton (2011) conducted a systematic,
diminished the wastes and thereby minimized the total
rigorous and methodologically valid review of CLSCM
cost to the model. The author found that one large

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 159

investment plan was superior decision to reduce setup respond to customers’ pressures and to encourage
cost. The author did not consider any particular goods partner-focused GSCM efforts in environmental
in this study. However, this article was not a SCM policies.
study. • Fernandes et al. (2017) suggested that the synergies of
quality management and SCM could promote the
Therefore, the above review of existing and well-
integration of existing approaches. In addition, the
established articles identifies the AR as the primary con-
author showed several other significant organizational
cern, along with the two major characteristics, like the CS
benefits of CS.
and the EC. Rest of this section deliberates on these two
• Singh et al. (2018) found that on one side, Indian
aspects, in succession.
retailers had good knowledge of the influences of CS on
SCM, and on the other side, these people understood
2.1 Customer satisfaction
less on how to match the competitive advantages of
supply chain practices with organizational performance.
The review of existing and well-established SCM articles
However, this article was based on only one country,
finds the thrust on the CS to kick off with the article by
India.
Cardozo (1965). The article of Cardozo (1965) was a
• Kandampully et al. (2018) presented a comprehensive,
detailed and experimental study of customer effort,
systematic review of the existing literature on CS. The
expectation and satisfaction. Next, Churchill and Sur-
authors presented the detailed mechanisms to properly
prenant (1982) investigated the effects of disconfirmation
manage customer experiences in the hospitality indus-
on CS in SCM studies. In next decade, Fornell (1992)
try. They integrated state-of-the-art customer experi-
reported the results of large-scale Swedish efforts. He
ence management knowledge in the generic business
showed the CS to be the quality of total consumption
context, along with principles of hospitality
process. Around same time, Anderson and Sullivan (1993)
management.
reported the antecedents and the consequences of CS. As
• Ali and Haseeb (2019) examined the effective supply
well, Anderson et al. (1994) demonstrated that the expec-
chain activities through radio frequency identification
tation, quality and price could affect CS and thereby the
(RFID). The authors collected the data from the surveys
profitability of supply chain in Swedish environment.
in textile and apparel companies. They found that RFID
Below, this article lists several existing researches on the
was contributing majorly in SCM operations and thus
influences of CS in SCM studies.
had significantly positive effects on performance of
• Siddiqi (2011) presented a study on the retail banking supply chain.
sector in Bangladesh, which was a South Asian • Imran et al. (2019) developed a comprehensive frame-
emerging economy in recent years. The authors studied work for e-logistics CS. The authors conducted an
the interrelation between service quality attributes, CS e-mail-based survey of 300 questionnaires. They found
and customer loyalty in banking sector. the information technology to be the main contributory
• Flint et al. (2011) examined three key parameters, like element between effective payment method and CS.
customer value anticipation, CS and loyalty in the
Additionally, Part A of Table 1 lists the several recent
context of supply chain. This article was an empirical
researches on the CS in supply chain.
study. There was no mathematical model presented in
Therefore, the present review of the existing literature
this article.
identifies several empirical forecasting methods as well as
• Ryu et al. (2012) examined the behavioural intention
the analytical models, in which researchers found the
along with the CS in restaurant chains. The authors
higher satisfaction of customers to financially boost the
studied the influence of several parameters, like qual-
organizations. However, this review could not find a single
ities of physical environment, foods, services and so on.
analytical CLSCM study, in which the goal of maximizing
• Recently, Saeidi et al. (2015) considered the mediating
the CS index had same importance as the goal of maxi-
role of reputation and CS in supply chain. The authors
mizing AR under the consideration of environmental con-
also presented the impact of corporate social responsi-
cerns via the RC layer to mathematical model.
bility in firms’ financial performance.
• Chavez et al. (2016) explored the links to implement
2.2 Environmental concerns
the customer-centric GSCM with operational perfor-
mance and CS. The authors suggested that customer
In several case studies and other researches, researchers
pressure had a positive effect on the implementation of
found that green supply chain management (GSCM)
customer-centric GSCM. They presented the guidelines
practices could not necessarily have direct influence on the
to implement the customer-centric GSCM in order to

123
160 A. Garai, T. K. Roy

economic performance of organizations. However, this improve the overall performance of suppliers in terms
could greatly achieve indirect advances. Sarkis (2003) of carbon management.
discussed the components and elements of GSCM. That • Zhu et al. (2013) showed that institutional pressures had
article showed the various components of GSCM to serve driven the adoption of internal GSCM practices by
as the foundation of decision framework. Here, he explored manufacturers. This in turn resulted in the adoption of
the applicability of dynamic nonlinear multi-attribute external GSCM practices. That research contributed to
decision-making model within green supply chain. Next, the literature on institutional theory in corporate
Srivastava (2007) identified the lack of broad reference environmental practices.
frame for CLSCM practices. That article presented one • Kannan et al. (2013) presented one comparative GSCM
integrated and fresh look into the area of CLSCM by study in fuzzy environment. Here, the authors inte-
deeply covering conceptualization in the reverse logistics grated the fuzzy multi-criteria decision-making method
angle. Rest of this subsection presents several other exist- and the multi-objective optimization approach for
ing and well-established articles based on EC of CLSCM supplier selection and order allocation problem.
studies. • Chen et al. (2017) presented one in-depth review
article. The authors took a systematic literature review
• Zhu et al. (2008) presented an empirical investigation.
and a quantitative bibliometric analysis. The results
The authors presented the construct and the scale in
showed that the environmental and financial consider-
order to evaluate the implementation of GSCM prac-
ations dominated the present research.
tices among manufacturers. In addition, they studied
• In an extensive review, Rajeev et al. (2017) found that
how the globalization and entry of China into the World
the studies, which focused on multiple dimensions of
Trade Organization helped to promote GSCM practices
sustainability, were comparatively scarce. The authors
in manufacturing enterprises.
found that the models on pollution making industries
• Bowen et al. (2009) presented one very useful study.
remained largely unaddressed, particularly in the
Here, the authors presented the benefits and downsides
emerging economies. So, the authors suggested to have
of GSCM policies in real life. They showed why
more focus on the industry-specific studies.
managers should avoid GSCM-related complex initia-
• Petljak et al. (2018) studied the environmental and
tives in cases of not having the capability to implement
financial performance of food retail markets. The
them.
authors identified many positive relations between the
• Sarkis et al. (2011) categorized and reviewed the recent
green in-store activities and the CLSCM practices.
CLSCM literature under nine broad organizational
• Chatterjee et al. (2018) thoroughly investigated the
theories. The authors specially emphasized the adop-
GSCM studies. The authors considered fifteen criteria
tion, diffusion and outcomes of GSCM practices. They
and then categorized these into five dimensions. Also,
identified other organizational theories as future
they proposed one evaluation framework in order to
research scopes.
select the suitable green suppliers.
• Eltayeb et al. (2011) noted the positive effects of
• Sarvestani et al. (2019) presented an integrated frame-
reverse logistics on cost reductions. The authors
work of one supplier selection, order acceptance and
presented the ideal approaches for business organiza-
scheduling problem. Here, the authors focused on the
tions to generate benefits through reduced waste and
profit in single-machine environment by considering
better resource utilization. These benefits were in
many customers.
addition to the economic benefits and thereby reduced
• Tseng et al. (2019) presented one detailed review of the
the cost to organizations.
existing literature published between 1998 and 2017.
• Around same time, Green et al. (2012) found that the
The authors collected the data from Scopus and ISI
adoption of GSCM practices by the manufacturing
Web of Science databases and objectively selected 880
industries could lead to an improved environmental,
papers to conduct the metadata analysis. They found
economic and operational performance.
that the research on drivers of GSCM had a declining
• Elhedhli and Merrick (2012) considered a SCN design
trend while there was a growing trend to apply
problem. Here, the authors considered the CO2 emis-
optimization models for enhancing the decision making
sions into the model. Besides, they considered the fixed
in pursuit of environmental performance.
and variable location and production costs to the
analytical model. The Part B of Table 1 lists several recent researches on
• Hsu et al. (2013) employed the decision-making trial the EC in SCM. Thus, the present review finds that the RC
and evaluation laboratory approach to recognize the layer to any CLSCM model boasts significant positive
influential criteria of carbon management in GSCM to effects on the environment. In addition, the green policies

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 161

Fig. 1 Flow chart of the logistics flows and induced monetary flows to the proposed model

carry good vibes and lead to revenue growth, job creations, et al. (2015) introduced the redefined membership func-
image building exercises and so on. However, this review tions to objective functions in fuzzy environment. How-
finds not a single optimization-based mathematical model ever, in 2016, authors of this article recognized several
of CLSCM that simultaneously considers the three classi- drawbacks to redefined membership functions (Garai et al.
cal performance indices, like AR, CS and EC. 2016, 2017). ‘‘Appendix B’’ describes several drawbacks
of existing membership function of classical fuzzy set as
2.3 Solution to MOOP in fuzzy environment well as redefined membership function.
In order to overcome those drawbacks, present authors
By concurrently considering the above three performance proposed to discard both the upper bound and lower bound
indicators, this article turns into a MOOP. Again, in order of classical membership function of fuzzy set. Thus, a new
to match with real life, this article assumes the imprecise set, viz. T-set, was born to supersede fuzzy set. The
aspiration levels to objective functions to the proposed T-characteristic function of T-set represents the impre-
MOOP, whereas the researchers devised numerous meth- ciseness more efficiently than membership function of
ods to characterize the Pareto optimal solution to MOOP fuzzy set (Garai et al. 2016, 2017). Besides, these authors
(Dubois and Prade 2015); these authors find the fuzzy set- found several analogous drawbacks in classical fuzzy
based methods to be very popular for solving MOOP optimization method of Zimmermann (1976, 1978) and in
(Ebrahimnejad and Verdegay 2018). Zimmermann other fuzzy set-based optimization methods (Garai et al.
(1976, 1978) established the classical method to solve 2016, 2017). The resultant generalized T-set-based multi-
MOOP in fuzzy environment. Many researchers, such as objective optimization method could determine ‘more
Zhou et al. (2007), Sakawa and Matsui (2012), Sakawa preferable’ Pareto optimal solution to MOOP in real-life-
et al. (2013), Wei et al. (2014), Salehi et al. (2015), Pra- based T-environment (Garai et al. 2016). Here, this article
manik et al. (2015), Garai et al. (2016, 2017), Dutta and proposes to employ the T-set-based method to determine
Jana (2017), Basu et al. (2019) and many others, applied the Pareto optimal solution to the proposed CLSCM model
various optimization methods in order to solve MOOP in in real-life-based T-environment.
fuzzy environment.
Again, in 2009, Jimenez and Bilbao showed that fuzzy
efficient solutions were not Pareto optimal, when one of the 3 Problem definition and assumptions
fuzzy goals was fully achieved. The authors generalized the
two-phase methods of Guu and Wu (1999). However, Wu This section describes the problem definition and
et al. (2015) argued that the method of Jimenez and Bilbao assumptions to the proposed CLSCM model.
(2009) was not a general procedure to obtain the Pareto
optimal solution to MOOP in fuzzy environment. Also, Wu

123
162 A. Garai, T. K. Roy

3.1 Problem definition recycles plants, secondary material markets, landfills


at any time. In addition, DM know the maximum
This article proposes one cost-effective and customer- operable units of plants and DCs at any time [see, for
centric CLSCM model by incorporating various conceiv- reference, Altiparmak et al. (2006), Grimm and
able components into the structure. The proposed model, Wosten (2018)].
not specific to any particular goods, is applicable to v. This article emphasizes on the social stability in
numerous businesses. This multi-objective mathematical order to achieve sustainability to the proposed
model optimizes the three major classical performance CLSCM model. So, the chain members stay away
indicators of supply chain, such as the AR, the EC and the from various conflicts, such as labour unrest, polit-
CS, simultaneously. This is one unique characteristic to the ical interferences and degradation of brand values.
proposed model. On one side, the model proposes to Accordingly, this article presumes that once a new
recycle the used goods via RC layer, and on other side, the plant or DC become operational, this remains
model emphasizes to maximize the CS index. The recy- operational throughout the time.
cling of used goods makes the best use of subsidies that the vi. One customer purchases the products from only one
Governments and various Environmental Protection DC at any time. As well, DM know the time-
Agencies (EPA) provide as assistance. Figure 1 illustrates dependent quantity of customers’ demand at any time.
the model graphically, in which the solid and dashed lines vii. The recycle plants plan the operation as per the
explain the directional relationships in terms of logistics return ratios of used products. Here, DM are familiar
flows and induced monetary flows, respectively. to the corresponding time-dependent values at any
Again, the proposed CLSCM model employs the real- time.
life-based T-set in order to determine the corresponding
Pareto optimal solution (Garai et al. 2016, 2017). The
T-characteristic functions to objective functions signify the
4 Specifications of the proposed SSCM
inherent impreciseness within the objective functions to the
model
proposed model. Lastly, the sensitivity analysis of param-
eters to the proposed model along with the graphical
This section formulates the proposed CLSCM model.
illustrations yields several key managerial insights.
4.1 Chain components
3.2 Assumptions
Based on the above deliberations, this article develops a
The following assumptions simplify the formulation of the
comprehensive conceptual framework to formulate an
proposed model
integrated logistics operational CLSCM model. The pro-
i. The proposed model considers solely one finished posed model involves eight potential chain components,
product at all the time. So, this is a single product scattered in two layers, viz. MC layer and RC layer. In MC
scenario. Besides, each component to the proposed layer, the potential chain members are as follows
model has one single channel or single crane
i. Suppliers (of unrefined or raw materials),
warehouse [see, for reference, Sheu et al. (2005),
ii. Plants,
Sarkar (2019)].
iii. Distribution centres (DCs) and
ii. The flow of information and money is instantaneous.
iv. Customers.
This article presumes all the transportations and
productions to be immediate. There is no allotment In RC layer, the potential chain members are as follows
of time to any of these tasks [see, for reference,
i. Collecting points,
Green et al. (2012), Grimm and Wosten (2018)].
ii. Recycle plants,
iii. The time-dependent inventory capacities to various
iii. Secondary material markets and
components to the proposed CLSCM model are fixed
iv. Final disposal locations (i.e. landfills).
and are well known to DM at any time. Also, the
time-dependent maximum quantity of products to be Besides, this article proposes to include the Govern-
manufactured at the plants and goods to be treated at ments or independent EPAs to act as another key anchor to
the recycle plants are fixed at any time [see, for the proposed model. The EPAs make various regulations to
reference, Khansalar et al. (2015), Sarkar (2019)]. be followed by several components to RC layer. These
iv. DM are acquainted with the time-dependent numbers authorities infuse taxpayers’ hard earned money in form of
of available suppliers, customers, collecting points, the aids and subsidies. This, in turn, stimulates the

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 163

recycling of used products in the recycle plants to RC layer Secondly, the combined time-dependent quantity of
to the proposed model. revenue to the RC layer within given total time T ðTDrc Þ is
as follows
4.2 Variables and parameters 0 P P  1
Rix ðlÞ  ðqrÞix ðlÞ þ
Table 11 in ‘‘Appendix A’’ lists all the symbols along with B i PxP C
XT BB x t ðRxt ðlÞ  rxt ðlÞÞþ C
C
the descriptions to be used in the proposed model. In TDrc ¼ B PP C ð4Þ
B ðRtm ðlÞ  rtm ðlÞÞþ C
addition, the proposed model employs following binary l¼1 B C
@ Pt Pm A
integer variables Rmk ðlÞ  ðrqÞmk ðlÞ
 m k
1; if DC j is operational
8j 2 J; zj ¼ Thirdly, the combined time-dependent quantity of sub-
0; otherwise
 sidies received by RC layer within given total time TðTSrc Þ
1; if plant k is operational
8k 2 K; pk ¼ is as follows
0; otherwise !
 X T XX
1; if customer i is served by DC j
8i 2 I; 8j 2 J; yji ¼ TSrc ¼ ðStm ðlÞ  rtm ðlÞÞ ð5Þ
0; otherwise l¼1 t m
ð1Þ
5.1.2 Customer satisfaction

5 Formulation of model The proportion of customers’ demand to be met within the


stipulated access time governs the CS index. The maxi-
This section formulates the proposed CLSCM model. mization of the total percentage of demand of customers to
be encountered within total given time T boosts up the
5.1 Objective functions value of CS index (Fernandes et al. 2017). Accordingly, the
combined time-dependent CS index within given total time
The two objective functions to the proposed CLSCM T ðCSSSCM Þ is as follows
model are as follows !
XT XX  X
CSSSCM ¼ zj  qji ðlÞ = di ðlÞ ð6Þ
5.1.1 Aggregate revenue l¼1 j i i

The AR to the proposed model has three major parts as


5.2 Constraints
follows
i. Combined revenue associated with given MC layer This article proposes the logistics conditions to be restric-
ðTRmc Þ, ted by the operational requirements or to be compelled by
ii. Combined revenue associated with given RC layer various regulations. There are five groups of constraints to
ðTDrc Þ, and the proposed model as follows
iii. Total subsidies associated with given RC layer
• Inventory constraints,
ðTSrc Þ.
• Return resource constraints,
So, the time-dependent AR ðARSSCM Þ is as follows • Operational constraints,
ARSSCM ¼ TRmc þ TDrc þ TSrc ð2Þ • Capacity constraints and
• Production constraints.
Firstly, the combined time-dependent quantity of rev-
enue to the MC layer within given total time TðTRmc Þ is as
follows 5.2.1 Inventory constraints

X
T XX 5.2.1.1 MC layer
TRmc ¼ ðRsk ðlÞ  bsk ðlÞÞ
l¼1 s k
! • Suppliers of unrefined materials (s)
XX  XX 
þ Rkj ðlÞ  f kj ðlÞ þ Rji ðlÞ  qji ðlÞ To each supplier s, the time-dependent quantity of
k j j i
unrefined material in inventory at any time l is equal to the
ð3Þ sum of corresponding residual in inventory at previous

123
164 A. Garai, T. K. Roy

time ðl  1Þ and time-dependent quantity acquired by that time, minus corresponding combined outbound time-
given supplier at that time, minus corresponding combined dependent quantity of finished goods transported to various
outbound flow of unrefined material transported to various customers at that time. Moreover, this time-dependent
plants at that time. Moreover, this time-dependent quantity quantity in inventory at given DC j complies with corre-
in inventory complies with corresponding predetermined sponding predetermined upper bound and lower bound at
upper bound and lower bound at given time l. Therefore, given time l. Therefore, the time-dependent inventory
the time-dependent inventory restriction owing to suppliers restriction owing to DCs to the proposed CLSCM model is
to the proposed CLSCM model is as follows as follows
ð8s; 8lÞ : inf s ðlÞ  ðqiÞs ðlÞ X ð8oD ; 8lÞ : inf ðlÞ  ðqiÞj ðlÞ
j X
¼ ðqiÞs ðl  1Þ þ qs ðlÞ  bsk ðlÞ  supðlÞ
s ¼ ðqiÞj ðl  1Þ þ fkj ðlÞ
k
X k
ð7Þ  qji ðlÞ  supðlÞ ð9Þ
i j

The above equation holds for any DC, provided that this
• Plants (k)
remains open at given time l.
To each plant k, the time-dependent quantity of both
unrefined materials and finished goods in inventory at any 5.2.1.2 RC layer
time l is equal to the sum of corresponding residual in
• Collecting points (x)
inventory at previous time ðl  1Þ and combined inbound
time-dependent quantity of unrefined materials sourced To each collecting point x, the time-dependent quantity
from various suppliers at that time and combined inbound of untreated goods in inventory at any time l is equal to the
time-dependent quantity of unrefined materials sourced sum of corresponding inventory quantity of untreated
from various secondary material markets at that time and goods, residual at previous time ðl  1Þ and combined
time-dependent quantity of products manufactured therein inbound time-dependent quantity of untreated goods
at that time, minus corresponding time-dependent quantity sourced from various customers at that time, minus cor-
of unrefined materials used to manufacture goods at given responding combined outbound time-dependent quantity of
plant at that time, minus combined outbound time-depen- untreated goods transported to various recycle plants at that
dent quantity of finished goods transported to various DCs time. Moreover, this time-dependent quantity in inventory
at that time. Moreover, this time-dependent quantity in at given collecting point x complies with corresponding
inventory at given plant k complies with corresponding predetermined upper bound and lower bound at that time.
predetermined upper bound and lower bound at given time Therefore, the time-dependent inventory restriction owing
l. Therefore, the time-dependent inventory restriction to collecting points to the proposed CLSCM model is as
owing to plants to the proposed CLSCM model is as follows
follows ð8x; 8lÞ : inf ðlÞ  ðriÞx ðlÞ
  x X
8k 2 op ; 8l : inf k ðlÞ  ðqiÞk ðlÞ ¼ ðriÞx ðl  1Þ þ ðqrÞix ðlÞ
¼ ðqiÞk ðl  1Þ ! X i
X X  rxt ðlÞ  supðlÞ ð10Þ
þ bsk ðlÞ þ ðrqÞmk ðlÞ þ qk ðlÞ t x
s m
X
 sr=m  ðqk ðlÞÞ  fkj ðlÞ  supðlÞ
j k
• Recycle plants (t)
ð8Þ
Recycle plants make use of three types of inventories,
The above relation holds for any plant, provided that the such as (i) to store the untreated goods received from
plant remains open at given time l. collecting points, (ii) to store the treated goods to be
• DCs (j) transported to secondary material markets and (iii) to store
wasted goods to be transported to final disposal locations.
To each DC j, the time-dependent quantity of finished
goods in inventory at any time l is equal to the sum of i. Untreated goods
corresponding residual in inventory at previous time ðl  To each recycle plant t, the time-dependent quantity of
1Þ and combined inbound time-dependent quantity of fin- untreated goods received from collecting points in inven-
ished goods sourced from various plants at given DC j at tory at any time l is equal to the sum of corresponding

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 165

quantity in inventory residual at previous time ðl  1Þ and dependent inventory restriction owing to wasted goods in
combined inbound time-dependent quantity of untreated recycle plants to the proposed CLSCM model is as follows
goods sourced from various collecting points at that time, w
minus corresponding time-dependent quantity of untreated ð8t; 8lÞ : inf ðlÞ  ðritwÞt ðlÞ
t
goods put to use to be recycled at that time. Moreover, this ¼ ðritwÞt ðl  1Þ þ su=w  ðrtÞt ðlÞ
X w
time-dependent quantity in inventory at given recycle plant  rtw ðlÞ  supðlÞ ð13Þ
t complies with corresponding predetermined upper bound w t
and lower bound at that time. Therefore, the time-depen-
Here, the physical transformation rate satisfies the con-
dent inventory restriction owing to untreated goods in
recycle plants to the proposed CLSCM model is as follows dition that su=m þ su=w ¼ 1:
u • Secondary material markets (m)
ð8t; 8lÞ : inf ðlÞ  ðriÞt ðlÞ
t X u To each secondary material market m, the time-depen-
¼ ðriÞt ðl  1Þ þ rxt ðlÞðrtÞt ðlÞ  supðlÞ
x t dent quantity of goods in inventory at any time l is equal to
the sum of corresponding inventory quantity of goods
ð11Þ
residual at previous time ðl  1Þ and combined inbound
time-dependent quantity of goods sourced from various
recycle plants at that time, minus corresponding combined
ii. Treated goods
outbound time-dependent quantity of goods transported to
To each recycle plant t, the time-dependent quantity of various plants at that time. Moreover, this time-dependent
treated goods in inventory at any time l is equal to the sum quantity in inventory at given secondary material market m
of corresponding quantity of treated goods residual at complies with corresponding predetermined upper bound
previous time ðl  1Þ and time-dependent quantity of and lower bound at given time l. Therefore, the time-de-
recycles goods at that time, minus corresponding combined pendent inventory restriction owing to secondary material
outbound time-dependent quantity of treated goods trans- markets to the proposed CLSCM model is as follows
ported to various secondary material markets at that time.
ð8m; 8lÞ : inf ðlÞ  ðriÞm ðlÞ
Moreover, this time-dependent quantity in inventory at m X
given recycle plant t complies with corresponding prede- ¼ ðriÞm ðl  1Þ þ rtm ðlÞ
termined upper bound and lower bound at that time. X t
 ðrqÞmk ðlÞ  supðlÞ ð14Þ
Therefore, the time-dependent inventory restriction owing m
k
to treated goods in recycle plants to the proposed CLSCM
model is as follows
m
• Final disposal locations (w)
ð8t; 8lÞ : inf ðlÞ  ðritmÞt ðlÞ
t
¼ ðritmÞt ðl  1Þ þ su=m  ðrtÞt ðlÞ To each final disposal location w, the time-dependent
X m quantity of wasted goods in inventory at any time l is equal
 rtm ðlÞ  supðlÞ ð12Þ
m t to the sum of corresponding inventory quantity of wasted
goods residual at previous time ðl  1Þ and combined
inbound time-dependent quantity of wasted goods sourced
iii. Wasted goods from various recycle plants at that time, minus corre-
sponding combined time-dependent quantity of wasted
To each recycle plant t, the time-dependent quantity of goods disposed at that time. Moreover, this time-dependent
wasted goods to be transported to final disposal locations in quantity in inventory at given final disposal location w
inventory at any time l is equal to the sum of corre- complies with corresponding predetermined upper bound
sponding inventory quantity residual at previous time ðl  and lower bound at given time l. Therefore, the time-de-
1Þ and corresponding time-dependent quantity of wasted pendent inventory restriction owing to final disposal loca-
goods at that time, minus corresponding combined out- tions to the proposed CLSCM model is as follows
bound time-dependent quantity transported to various final
disposal locations at that time. Moreover, this time-de- ð8w; 8lÞ : inf ðlÞ  ðriÞw ðlÞ
w X
pendent quantity in inventory at given recycle plant t ¼ ðriÞw ðl  1Þ þ rtw ðlÞ  ðrf Þw ðlÞ  supðlÞ
w
complies with corresponding predetermined upper bound t
and lower bound at that time. Therefore, the time- ð15Þ

123
166 A. Garai, T. K. Roy

X
5.2.2 Return resource constraints ð8i; 8lÞ : qji ðlÞ  di ðlÞ  yji ðlÞ ð21Þ
j
The time-dependent quantity of the used product returned
by various customers at any time l is the product between and
X
combined time-dependent demand from customers at that ð8i; 8lÞ : yji ðlÞ ¼ 1 ð22Þ
time and used-product return flow ratio c. So, therefore, the j
time-dependent return resource constraint to the proposed
CLSCM model is as follows
! 5.2.4 Capacity constraints
XX X
ð8lÞ : Rrc ðlÞ ¼ ðqrÞix ðlÞ ¼ c  di ðlÞð  0Þ
i x i 5.2.4.1 Suppliers To any supplier s, the combined time-
dependent quantity of acquired materials is more than or
ð16Þ
equal to the corresponding combined outbound time-de-
Here, the statistical evidences predetermine the value of pendent quantity of unrefined materials transported to
parameter c.  So, the time-dependent
various plants within any time T.
capacity constraint owing to suppliers to the proposed
5.2.3 Operational constraints CLSCM model is as follows
! !
XT X T X
5.2.3.1 Operational plants There is a limit to number of ð8T  T; 8sÞ : qs ðlÞ  bsk ðlÞ ð23Þ
plants that operate at any time l. Moreover, each plant k l¼l1 l¼l1 k
either operates or remains closed during entire time T. The
total number of operational plants does not exceed the
corresponding maximum allowable number at any time l. 5.2.4.2 Plants To any plant k, the combined time-de-
So, the time-dependent operational constraint owing to pendent quantity of unrefined materials used to manufac-
plants to the proposed CLSCM model is as follows ture goods does not exceed the sum of corresponding
X combined time-dependent quantity of unrefined materials
ð8l; 8kÞ : pk ðlÞ  P ð17Þ received from several secondary material markets and
k2Op  In addition, the time-
various suppliers within any time T.
 
8k; li ðtime interval), 8lj ðtime interval) : pk ðli Þ dependent quantity of goods manufactured at any plant k is
¼ pk ðlj Þ ð18Þ more than or equal to the corresponding combined time-
dependent quantity of finished goods transported to various
DCs. So, the time-dependent capacity constraint owing to
5.2.3.2 Operational DCs There is a limit to number of plants to the proposed CLSCM model is as follows
DCs that operate at any time l. Moreover, each plant k ð8T  T; 8kÞ :
either operates or remains closed during entire time T. The T T
X X X
total number of operational DCs does not exceed the cor- Unrefined materials s r=m
 qk ðlÞ  ð bsk ðlÞ
responding maximum allowable number at any time l. So, l¼l1 l¼l1 s
the time-dependent operational constraint owing to DCs to X
þ ðrqÞmk ðlÞÞ
the proposed CLSCM model is as follows m
X
ð8lÞ : zj ðlÞ  W ð19Þ ð24Þ
j2OD
 
8k; li ðtime interval), 8lj ðtime interval) : zj ðli Þ ¼ zj ðlj Þ
ð20Þ
Table 2 Structural overview of the proposed CLSCM model
5.2.3.3 On relations between DCs and customers The Facts Numbers
time-dependent quantity of products transported to cus-
tomer i from various DCs is less than or equal to the time- Number of objective functions 2
dependent demand of customer i at any time l. Moreover, Types of constraints 27
this article presumes that one customer buys goods from Types of decision variables 22
one DC only. So, the time-dependent operational constraint Types of binary variables 3
owing to customers to the proposed CLSCM model is as
follows

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 167

T
X T X
X dependent capacity constraint owing to treated goods in
Manufactured goods qk ðlÞ  fkj ðlÞ ð25Þ recycle plants to the proposed CLSCM model is as follows
l¼l1 l¼l1 j
T X
X T
X
ð8T  T; 8tÞ : rtm ðlÞ  su=m  ðrtÞt ðlÞ ð29Þ
l¼l1 m l¼l1
5.2.4.3 DCs To any DC j, the combined outbound time-
dependent quantity of manufactured goods purchased by
many customers is less than or equal to the corresponding
combined inbound time-dependent quantity of manufac- • Wasted goods

tured goods sourced from various plants within any time T. The combined outbound time-dependent quantity of
So, the time-dependent capacity constraint owing to DCs to wasted goods transported to various final disposal locations
the proposed CLSCM model is as follows is less than or equal to the corresponding combined time-
T X
X T X
X dependent quantity of wasted goods processed within any
ð8T  T; 8jÞ : fkj ðlÞ  qji ðlÞ ð26Þ  So, the time-dependent capacity constraint owing
time T.
l¼l1 k l¼l1 i to wasted goods in recycle plants to the proposed CLSCM
model is as follows
5.2.4.4 Collecting points To any collecting point x, the T X
X T
X
combined inbound time-dependent quantity of untreated ð8T  T; 8tÞ : rtw ðlÞ  su=w  ðrtÞt ðlÞ ð30Þ
l¼l1 w l¼l1
goods sourced from various customers is more than or
equal to the combined outbound time-dependent quantity
of untreated goods transported to different recycle plants 5.2.4.6 Secondary material markets To any secondary
within any time T. So, the time-dependent capacity con- material market m, the combined outbound time-dependent
straint owing to collecting points to the proposed CLSCM quantity of recycled materials transported to various plants
model is as follows is less than or equal to the corresponding combined
T X
X T X
X inbound time-dependent quantity of treated goods sourced
ð8T  T; 8xÞ : ðqrÞix ðlÞ  rxt ðlÞ ð27Þ from various recycle plants within any time T.  So, the
l¼l1 i l¼l1 t time-dependent capacity constraint owing to secondary
material markets to the proposed CLSCM model is as
follows
5.2.4.5 Recycle plants Following relationships hold to
 T X
X T X
X
any recycle plant t within any time T.
ð8T  T; 8mÞ : ðrqÞmk ðlÞ  rtm ðlÞ ð31Þ
• Untreated goods l¼l1 k l¼l1 t

The combined inbound time-dependent quantity of


untreated goods sourced from various collecting points is 5.2.5 Production constraints
more than or equal to the corresponding combined time-
dependent quantity of goods taken for recycling within any 5.2.5.1 Plants The time-dependent quantity of goods
 So, the time-dependent capacity constraint owing
time T. produced in any plant k does not exceed the corresponding
to untreated goods in recycle plants to the proposed manufacturing capacity at any time l. So, the time-de-
CLSCM model is as follows pendent production constraint owing to plants to the pro-
T X
X T
X posed CLSCM model is as follows
ð8T  T; 8tÞ : rxt ðlÞ  ðrtÞt ðlÞ ð28Þ ð8k; 8lÞ : qk ðlÞ  Capak ðlÞ ð32Þ
l¼l1 x l¼l1

5.2.5.2 Recycle plants The time-dependent quantity of


• Treated goods goods recycled at any recycle plant t does not exceed the
corresponding recycling capacity at any time l. So, the
The combined outbound time-dependent quantity of
time-dependent production constraint owing to recycle
treated goods transported to various secondary material
plants to the proposed CLSCM model is as follows
markets is less than or equal to the corresponding inbound
time-dependent quantity of treated goods obtained by ð8t; 8lÞ : ðrt)t ðlÞ  Capat ðlÞ ð33Þ
recycling untreated goods within any time T. So, the time-
Here, the proposed analytical model considers two
objective functions along with twenty seven types of

123
168 A. Garai, T. K. Roy

constraints. This is to note that the actual quantities of Table 4 Unit revenues of MC layer and RC layer to the proposed
constraints, decision variables, binary variables, parameters model
and all that are subject to the values of various indices Parameters Unit revenues ($)
associated with chain components. Table 2 describes the
Lower bounds Upper bounds
vastness to the proposed CLSCM model.
MC layer
Suppliers of unrefined materials 29 55
6 Mathematical preliminaries Plants 360 610
Distribution centres 521 825
This section presents the existing mathematical develop- Customer 0 5
ment employed to determine the Pareto optimal solution to RC layer
the proposed model. Collecting points 4.3 6.8
Recycle plants 6.1 9.5
6.1 Definitions Secondary material markets 15. 30

‘‘Appendix C’’ presents the existing and useful definitions


to this article.
8
Definition A point x 2 X is said to be T-Pareto optimal > 1; zi ðxÞ  Li
<
solution to MOOP if there does not exist another x 2 X Ui  zi ðxÞ
li ðzi ðxÞÞ ¼ ; Li  zi ðxÞ  Ui ; 8i ¼ 1. . .k
such that Ti ðci xÞ  Ti ðci xÞ; 8i ¼ 1. . .k, i = j and >
: U i  Li
Ti ðcj xÞ\Ti ðcj xÞ, for at least one j. Here, ‘*’ denotes 0; zi ðxÞ  Ui
optimality. ð35Þ
Here Li and Ui are the goal and goal plus tolerance
6.2 T-sets-based multi-objective optimization
values, respectively, to ith objective function zi ðxÞ; i ¼
1. . .k (Zadeh 1965). The existing literature prefers to
The general form of MOOP with k imprecise objective
denote the interval ½Li ; Ui  by Xi ; 8i ¼ 1. . .k (Ebrahimne-
functions zi ðxÞ; i ¼ 1. . .k is as follows (Zimmermann
jad and Verdegay 2018).
1976, 1978)
In real-life-based environment, the linear T-character-
min ðz1 ðxÞ; z2 ðxÞ. . .zk ðxÞÞ istic function Ti ðzi ðxÞÞ; i ¼ 1. . .k to ith objective function
subject to ð34Þ zi ðxÞ; i ¼ 1. . .k to model (34) is as follows (Garai et al.
x 2 X: 2016, 2017)
Ui  zi ðxÞ
Here, X ¼ fx 2 Rn : Ax  b; x  0g; b ¼ ðb1 ; b2 . . .bm Þ 2 8i ¼ 1. . .k; Ti ðzi ðxÞÞ ¼ ; 8zi ðxÞ ð36Þ
m Ui  Li
Z and A is m  n matrix. Under the assumption of
imprecise aspiration level to each objective function By considering v ¼ maxfTi ðzi ðxÞÞg; i ¼ 1. . .k and
i
zi ðxÞ; i ¼ 1. . .k to the above model (34), there are numer- by employing the min–max operator of Zimmermann
ous methods to characterize the corresponding Pareto (1976) to model (34), the single objective optimization
optimal solutions (Zimmermann 1985; Ebrahimnejad and model to model (34) in T-environment is as follows (Garai
Verdegay 2018). et al. 2017)
In case the objective functions zi ðxÞ; i ¼ 1. . .k to model max v
(34) are fuzzy in nature, the corresponding linear mem-
subject to ð37Þ
bership function to ith objective function zi ðxÞ; i ¼ 1. . .k is
as follows (Zimmermann 1976, 1978) Ti ðzi ðxÞÞ  v; i ¼ 1. . .k; x 2 X:

Table 3 Demand of goods and


Time intervals Demands of goods from customers Used-product returns ðc ¼ 0:25Þ
return of used products to the
proposed model 1 4290 1072
2 4263 1065
3 4049 1012
4 4115 1028

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 169

Table 5 Numerical values of different parameters


Parameters Values Parameters Values

Subsidy 8.7 Physical transformation rate of finished goods 1


Physical transformation rate of treated goods 0.8 Physical transformation rate of wasted goods 0.2
Used-product return flow ratio 0.25 Weights associated with MC layer/RC layer 1
Maximum allowable number of open plants 1 Maximum allowable number of open DCs 1
Minimum inventory capacities Maximum inventory capacities
Suppliers 0 Suppliers 5000
Plants 0 Plants 5000
DCs 0 DCs 5000
Collecting points 0 Collecting points 3000
Untreated goods at recycle plants 0 Untreated goods at recycle plants 1000
Treated goods at recycle plants 0 Treated goods at recycle plants 1000
Wasted goods at recycle plants 0 Wasted goods at recycle plants 1000
Secondary material markets 0 Secondary material markets 2000
Disposal locations 0 Disposal locations 1000

Table 6 Individual maximum and minimum values to objective Table 8 Sensitivity analysis of the return flow of used products in
functions T-environment
Objective functions Maximum values Minimum values c Aggregate revenue Customer-satisfaction index

AR $22,970,400 $20,074,550 1.0 $23,031,794 5.78


CS index 7.70 3.11 0.8 $22,992,843 6.45
0.5 $22,984,889 7.31
0.33 $22,978,169 7.30
0.2 $22,965,303 7.56
0.1 $22,940,888 7.56
When the DM assign necessary numerical values, the
0.0 $22,914,669 7.20
software, like LINGO, SUPERLINGO, MATLAB,
MATHEMATICA, can determine the optimal solution to
the above single objective optimization model and hence
the Pareto optimal solution to the proposed CLSCM model.
7.1 Setting up the data

7 Numerical application The motivation behind the data used in this section comes
mainly from the mathematical models of several existing
This section numerically illustrates the proposed multi- articles, like Sheu et al. (2005) and Altiparmak et al.
objective CLSCM model in T-environment and thereby (2006). The key strength of these articles is that both
determines the corresponding Pareto optimal solution. articles are well established, and the data used in these
articles tasted the time. Again, Sheu et al. (2005) presented
one optimization-based mathematical model, in which the
authors dealt with various integrated logistics operational

Table 7 Comparison between optimal values in T-environment and fuzzy environment


Objective T- Fuzzy Remarks
functions environment environment

AR $22,970,400 $22,000,000 Optimal values to both the objective functions of the proposed CLSCM model in T-environment
CS index 7.5 6 are superior than that of classical fuzzy environment. The optimal values in T-environment are
preferable than goals to both objectives

123
170 A. Garai, T. K. Roy

problems of GSCM. Here, this model considered several Table 9 Sensitivity analysis of subsidy in T-environment
factors, like the used-product return ratio, Governments Stm ðlÞ Aggregate revenue Customer-satisfaction index
subsidies to RC layer and all that. However, this model did
not consider the CS index, one most fundamental perfor- 13 $22,968,000 7.31
mance index to any SCM model. Besides, the author solved 10 $22,970,400 7.49
the proposed model in crisp environment. 8 $22,970,400 7.49
Altiparmak et al. (2006) proposed a genetic algorithm- 6 $22,970,400 7.40
based solution procedure to determine the set of Pareto 3 $22,968,268 7.28
optimal solutions to a mixed-integer multi-objective SCN 0 $22,968,268 7.27
design problem. The authors used the actual data from one
company that produced the plastic products in Turkey.
Also, the solution method of Altiparmak et al. (2006)
consisted two different weight-based approaches. That products returned by customers at any time. Table 4 sets
mathematical model simultaneously determined the mini- the estimated unit revenues to MC layer and RC layer.
mum costs of supply chain and the maximum CS index in Next, Table 5 displays the other numerical values. This
crisp environment. However, there was no deliberation on numerical application considers around 125 constraints and
the corresponding environmental impact. 87 variables, including 3 binary integer variables.
Although the present article considers the data that are
partly inspired from Altiparmak et al. (2006) and are 7.2 Computation
moderately inspired from Sheu et al. (2005), the constraints
to the proposed model are much analytical and different Firstly, analysts consider one objective function arbitrarily.
from those existing studies. This article numerically solves Next, they compute the individual maximum and minimum
the proposed CLSCM model in real-life-based T-environ- values, subject to given constraints. Table 6 displays the
ment. Besides, the advantages of considering SCM models results.
in the real-life-based environments are well established. Based on the above values, DM predetermine the goals
Hence, this article does not compare present numerical and tolerance values to objective functions. This article
results with those of Sheu et al. (2005) or Altiparmak et al. assumes the goal of AR ðARSSCM Þ as $22; 000; 000 with
(2006). acceptable tolerance level $20; 500; 000; and the goal of
• Values CS index ðCSSSCM Þ as 6 with acceptable tolerance level 4.
So, the T-characteristic functions to objective functions are
Here, T = 1 year (i.e. four quarters in a year are four as follows
time intervals). For the sake of simplicity alone, this article
ARSSCM  20; 500; 000:
considers the number of members to various components to TðARSSCM Þ ¼ ; 8ARSSCM
1; 500; 000:
be unity in each case. Table 3 presents the demands of
manufactured goods along with the quantities of used and

Fig. 2 Graphical illustration of changes in the return flow of used products in T-environment

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 171

Fig. 3 Graphical illustration of changes in the subsidy in T-environment

Table 10 Sensitivity analysis of the physical transformation rate of The above optimal solution shows that the optimal value
treated goods from untreated goods in T-environment
to AR is superior than the corresponding goal. Likewise,
su=m Aggregate revenue Customer-satisfaction index the optimal value to the CS is better than the corresponding
goal.
1.0 $22,968,495 7.29
0.9 $22,974,997 7.41
7.3 Comparison between optimum values in T-
0.7 $22,971,353 7.51
environment and fuzzy environment
0.6 $22,972,022 7.30
0.5 $22,971,563 7.29
Table 7 compares the optimal values to objective functions
between fuzzy environment and T-environment. Here, the
proposed model finds an enhancement of 4.41% to AR and
an enhancement of 25% to CS index, while simultaneously
CSSSCM  4 addressing the EC in T-environment. This result confirms
TðCSSSCM Þ ¼ ; 8CSSSCM
64 the superiority of T-sets-based Pareto optimal solution over
Therefore, LINGO 17 determines the Pareto optimal the fuzzy counterpart to CLSCM models.
solution to the proposed CLSCM model as follows
v ¼ 1:65; ARSSCM  ¼ $22; 970; 400; SSSCM  ¼ 7:5:

Fig. 4 Graphical illustration of changes in the physical transformation rate of treated goods from untreated goods in T-environment

123
172 A. Garai, T. K. Roy

8 Sensitivity analysis and managerial above result is important to the political establishments
insights and philanthropical organizations. Figure 3 portrays all
these findings graphically.
This section performs the sensitivity analysis in detail. • Table 10 presents the phenomenon of percentage
changes of physical transformation rate of the treated
8.1 Vulnerability of key parameters goods from untreated goods. This analysis finds that
fully (100%) recyclable goods generate the less rev-
The modifications of system parameters within present enue, and this reduces the satisfaction of customers to
study alter the optimal values, which are obtained in sub- supply chains. This is a contradiction to the general
section 7.2 of this article, to both the objective functions to phenomena that the fully recyclable goods invite
the proposed model. In addition, to maintain the sensible highest level of CS. Again, since the less recyclable
length, this article performs the sensitivity analysis to goods reduce the CS index faster, this article suggests
solely the following key parameters by taking one param- the use of customer centric and recyclable materials,
eter at a time and keeping the remaining parameters at both, in optimal ratios to manufacture goods. More
original levels. research is necessary to optimally determine the CS
index as a function of quantity of recyclable goods.
i. Physical transformation rate of the treated goods Moreover, this strategy increases the AR to SCM
 
from untreated goods su=m , model. Figure 4 portrays all these findings graphically.
ii. Subsidies of the Governments or EPA ðStm ðlÞÞ, and
iii. Return flow of used products ðcÞ. The above analysis presents a profound knowledge to
DM. However, the DM should think of the findings of
• Table 8 presents the sensitivity analysis of the return Oliver Wight (2019), who found that whereas an optimized
flow of used products to the proposed CLSCM model in SCN reduced the corresponding operating costs by up to
T-environment. This analysis finds that the collection of 50% along with the perfect order ratings by 17%, around
used products at higher rate leads to more revenue to 66% of change initiatives failed due to lack of management
the proposed model. In addition, this yields that a col- skills and 75% of supply chains were unsuccessful to
lection of 100% of used products from the customers deploy the planned policies. Hence, the DM should act so
results in the maximum AR to the proposed model. that the proposed CLSCM model is customer friendly as
However, the relatively higher return flow negatively well as environmentally and financially sustainable at any
impacts the CS index to the proposed model. When the time.
customers are bound to return the used products within
some given time, this analysis finds a subsequent drop 8.2 Implications and managerial insights
to CS index. The authors observe that various measures,
like to strategically place the collecting points, to There are well-established results in real-life-based envi-
announce promotional schemes and all that make the ronments that the goal of maximizing the CS makes posi-
customers willing to return the used products within tive impacts on the AR, and the EC reflected through RC
given time. This article suggests putting more thrust to layer certainly increases the brand value and all that to any
collect the used products from customers. Moreover, SCM model. However, this article demonstrates that the
this article proposes more in-depth exploration of return AR and the CS index both improve under consideration of
flow of used products. The authors used MATLAB to the EC constraints to the proposed CLSCM model in real-
depict all these findings graphically in Fig. 2. life-based T-environment. So, the DM should consider all
• Table 9 illustrates the sensitivity analysis of the subsidy three classical performance indices to any SCM model
as received by the RC layer to the proposed model in simultaneously.
T-environment. This analysis finds that both of the The sensitivity analysis finds that the moderately high
inflated and the low subsidies result in much lesser percentage of return of used products by the customers
satisfaction level to customers along with lower AR to yields more AR and much higher CS index. And, very high
supply chain. This finding adds strength to the matter percentage of return of used products within specified time
that the amounts of subsidies and thereby the quantity reduces the CS index. Therefore, the policy makers should
of treated goods at various recycle plants should always function so as to make the customers more sincere and
lie within some predetermined limits. Besides, this more enthusiastic to timely return the used products under
suggests that there should have intervallic subsidies to the influences of higher ethical values and/or various pro-
the businesses, who optimally recycles the used goods, motional schemes. So, the proposed optimal blending of
in order to protect the environment. Undeniably, the CS and EC increases the AR and thus makes any SCM

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 173

model to be sustainable. The DM can take analogous of return of used products by customers can yield more AR
blending in their organizations. and higher CS index. But, very high percentage of return
Besides, the DM should avoid the negative conse- within specified time decreases the CS index. Therefore,
quences due to break in operations or due to shut down of this model blends the EC and the CS index while ensuring
any plant, DC. So, the number of plants and DCs to be the cost-effectiveness. Besides, this analysis finds that the
newly established or to be operational at any time should subsidies offered without any break and within the opti-
be optimally predetermined. Moreover, to earn through mally determined bounds to the RC layer can help the SCN
subsidy increases the enthusiasm in managers of any sup- to achieve the EC related targets.
ply chain. However, this study finds that the amount of There are many scopes to carry future researches based
subsidy received by recycle plants should remain within on the present study. One can consider the optimization of
one predetermined interval. The entire above-mentioned net profit as one objective function. Future models can
circumstances play key role in the decision-making process impose the additional structures to the CS index. More
and thus provide significant insights to supply chain studies can help to fine-tune the constraints to present
managers. model. Future mathematical models can consider several
recent indicators, like CSR, agility, data security and
sharing, IoT and all that, along with the classical perfor-
9 Conclusions mance indicators to mathematical models. Also, future
researchers can consider more general numerical data sets
This article presents one CS- and EC-based analytical and thereby solve the MOOP. In future, the researchers
CLSCM model in real-life-based environment. The review should simulate the results in order to make the industry-
of the existing literature in Sect. 2 identifies the AR, the CS ready model. One can consider higher number of members
and the EC to be the classical performance indicators to to any chain component. One can formulate analogous
any SCM model. However, this review fails to find a single models to specific industry. And last but not the least, the
optimization-based SCM model that simultaneously con- present study remains distinctive to the crisp environment,
siders these three classical indicators. This article considers classical fuzzy environment and other real-life-based
the AR and the CS as objective functions. The RC layer to imprecise environments.
the proposed model reflects the EC through various con-
straints. Even though several existing studies take the RC
layer into account, the present study is the first to optimize Funding This study was not funded by any agency or Government.
None of the authors received any kind of financial grant or support for
the CS index at the same time. The accompanying subsidy this study.
obtained from the Governments and EPA to the recycle
plants at RC layer boosts the financial outcome to the
proposed model. Compliance with ethical standards
Here, the objective functions to the proposed model turn
the mathematical model into one MOOP. This article Conflict of interest The authors declare that they have no conflict of
interest.
determines the corresponding Pareto optimal solution in
real-life-based T-environment. ‘‘Appendix C’’ lists the Ethical approval This article does not contain any studies with human
advantages of T-set over fuzzy set in order to represent participants or animals performed by any of the authors.
impreciseness. The numerical application in Sect. 6 shows
that the optimal values to both the objective functions, viz.
the AR and the CS index, are more preferable to DM than Appendix A: Symbols and descriptions
respective goals. The sensitivity analysis along with
graphical illustrations in Sect. 7 to the parameters to the See Table 11.
proposed model finds that the moderately high percentage

Table 11 Symbols with the descriptions used in the proposed CLSCM model
Symbols Descriptions

Indices
s2S Supplier s among total S suppliers of unrefined materials
k2K Plant k among total K plants
j2J DC j among total J DC

123
174 A. Garai, T. K. Roy

Table 11 (continued)
Symbols Descriptions

i2I Customer I among total I customers


x2X Collecting point x among total X collecting points
t 2 PR Recycle plant t among total PR recycle plants
m2M Secondary material market m among total M secondary material markets
w2W Final disposal location w among total W final disposal locations
Objective functions
ARSSCM AR to the proposed CLSCM within given total time T
TRmc Combined time-dependent quantity of revenue to MC layer within given total time T
TDrc Combined time-dependent quantity of revenue to RC layer within given total time T
TSmc Combined time-dependent quantity of subsidy to RC layer within given total time T
CSSSCM Combined time-dependent value of CS index within given total time T
Coefficients to objective functions
Rsk ðlÞ Revenue of supplier s associated with one unit of unrefined material transported to plant k at any time l
Rkj ðlÞ Revenue of plant k associated with one unit of finished product transported to DC j at any time l
Rji ðlÞ Revenue of DC j associated with one unit of finished product transported to customer I at any time l
Rix ðlÞ Compensation received by customer I associated with one unit of used products returned to collecting point x at any time l
Rxt ðlÞ Revenue of collecting point x associated with one unit of used products transported to recycle plant t at any time l
Rtm ðlÞ Revenue of recycle plant t associated with one unit of treated item transported to secondary material market m at any time l
Rmk ðlÞ Revenue of secondary material market m associated with one unit of treated item transported to plant k at any time l
Stm ðlÞ Subsidy associated with one unit of treated item sent to secondary material market m, received by recycle plant t from Governments and/or EPA at any
time l
Decision variables and parameters
qs ðlÞ Time-dependent quantity of unrefined material acquired by any given supplier s at any time l
qk ðlÞ Time-dependent quantity of products manufactured at any given plant k from unrefined materials therein at any time l
ðqiÞs ðlÞ Time-dependent quantity of unrefined materials in inventory associated with any given supplier s at any time l
ðqiÞk ðlÞ Time-dependent quantity of both unrefined materials and finished goods in inventory related to each given plant k at any time l
ðqiÞj ðlÞ Time-dependent quantity of finished goods in inventory associated with any given DC j at any time l
ðriÞx ðlÞ Time-dependent quantity of used goods in inventory associated with any given collecting point x at any time l
ðriÞt ðlÞ Time-dependent quantity of untreated goods in inventory associated with any given recycle plant t at any time l
ðrtÞt ðlÞ Time-dependent quantity of untreated goods in inventory put to use for recycling related to each given recycle plant t at any time l
ðritmÞt ðlÞ Time-dependent quantity of treated goods in inventory associated with any given recycle plant t at any time l
ðritwÞt ðlÞ Time-dependent quantity of wasted goods in inventory associated with any given recycle plant t at any time l
 
ðriÞm ðlÞ Time-dependent quantity of treated goods in inventory associated with any given secondary material market m at any time l
 
ðriÞw ðlÞ Time-dependent quantity of wasted goods in inventory associated with any given final disposal location w at any time l
 
ðrf Þw ðlÞ Time-dependent quantity of wastages disposed at any given final disposal location w at any time l
bsk ðlÞ Quantity of unrefined materials sourced from supplier s to plant k at any time l
f kj ðlÞ Quantity of finished goods transported sourced from plant k to DC j at any time l
qji ðlÞ Quantity of finished goods transported sourced from DC j to customer I at any time l
ðqrÞix ðlÞ Quantity of used goods sourced from customer I to collecting point x at any time l
rxt ðlÞ Quantity of used goods sourced from collecting point x to recycle plant t at any time l
rtw ðlÞ Quantity of untreated goods sourced from recycle plant t to final disposal location w at any time l
rtm ðlÞ Quantity of treated goods sourced from recycle plant t to secondary material market m at any time l
ðrq)mk ðlÞ Quantity of treated goods sourced from secondary material market m to plant k at any time l
di ðlÞ Time-dependent demand of finished goods from customer I at any time l
T Total time
l Any given time
Op Number of open plants
OD Number of open DCs
sr=m Physical transformation rate of finished goods from unrefined materials
u=m Physical transformation rate of treated goods from untreated goods
s
su=w Physical transformation rate of wasted goods from untreated goods

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 175

Appendix B: Drawbacks of classical fuzzy set Appendix C: T-set and related definitions

The drawbacks of the membership function to classical The characteristic function of crisp subset assigns either 1
fuzzy set as well as the redefined membership function in or 0 to elements of universal set and thereby discriminates
Wu et al. (2015) are as follows (Garai et al. 2016, 2017) between the members and non-members. Again, member-
ship function to fuzzy subset generalizes the characteristic
• Firstly, the employments of redefined membership
function and the values assigned to elements of universal
functions in Wu et al. (2015) were not as per the
set fall within the specified range of closed unit interval [0,
definitions presented in mathematical model as well as
1]. However, Garai et al. (2016, 2017) observed that the
in numerical examples of that article.
membership function failed to discriminate between yes
• Secondly, the mathematical model in Wu et al. (2015)
and certainly yes by assigning same value ‘unity’ to both
employed strictly monotonic redefined membership
cases and failed to discriminate between no and certainly
functions over the entire real line. However, the
no by assigning same value ‘zero’ to both cases (Garai
definitions in that article specified otherwise.
et al. 2016, 2017).
• In addition, the researchers could only employ the part
In other words, elements to universal set must be or be
of redefined membership functions, at which the
or partly be or not be or never be lying in a subset. The
corresponding objective values lied within the goals
concept of being, partly being or not being is well mea-
and goal plus tolerances, to the minimization type of
sured by membership function of fuzzy subset. However,
objective functions. However, so far, researchers did
this cannot explain the cases of must being (since mem-
not consider the above prerequisite (in form of
bership value is unity, at all times) as well as never being
constraints) within the mathematical models.
(since membership value is zero, at all times) for elements
• Thirdly, Wu et al. (2015) discarded the upper bound at
of fuzzy subset. Garai et al. (2017) presented the following
unity to membership functions of fuzzy objective
example to illustrate all these as follows
functions. However, the corresponding lower bound to
membership functions of fuzzy objective functions • The task is to pick up one tall policeman among all
remained intact at zero. Here, the authors find this to be policemen for the guard of honour ceremony of the
arbitrary and biased. President of USA during the official visit to North
• Most importantly, the classical fuzzy membership Korea. To assign the job solely to some tall person in
function could provide satisfying solution, only when force is a bold step to North Korea. If one policeman
the extreme ends of imprecise information lied within with height of more than 60 is considered to be tall, he
zero and one. However, Garai et al. (2016, 2017) can be assigned unity as the membership value in fuzzy
showed that the proper mathematical representation of subset of tall policemen.
impreciseness cannot be confined within any closed and
Assume that Mr. Bansal is 60 400 tall and Mr. Chowd is
bounded interval of real line, at all times. 0 00
6 6 tall. In fuzzy set theory, both are given unity as the
Figure 5 illustrates the drawbacks of membership same membership value. However, if the DM were all
function to classical fuzzy set (Garai et al. 2017). machines with zero emotion, this would have been

Fig. 5 Drawbacks of the


membership function to
classical fuzzy set

123
176 A. Garai, T. K. Roy

acceptable to choose any one of two persons (or in fact, Alinaghian M, Zamani M (2019) A bi-objective fleet size and mix
any policeman taller than 60 ). Nevertheless, as human green inventory routing problem, model and solution method.
Soft Comput 23(4):1375–1391. https://doi.org/10.1007/s00500-
beings, the mind plays key role. DM can have full happi- 017-2866-2
ness only after selecting the tallest person. Here, all the Altiparmak F, Mitsuo G, Lin L, Paksoy T (2006) A genetic algorithm
persons taller than 60 are assigned the same membership approach for multi-objective optimization of supply chain
value (unity). So, DM fail to find the most suitable person networks. Comput Ind Eng 51(1):196–215
Anderson D (2014) Design for manufacturability: how to use
in force for the prestigious ceremony. concurrent engineering to rapidly develop low-cost, high-quality
This marks that the upper bound of membership func- products for lean production. CRC Press, Boca Raton
tion of fuzzy subset at unity causes the dilemma. Anderson E, Sullivan M (1993) The antecedents and consequences of
Consequently, Garai et al. (2016, 2017) introduced the CS for firms. Mark Sci 12(2):125–143
Anderson E, Fornell C, Lehmann D (1994) Customer satisfaction,
T-characteristic function and then T-set as follows market share, and profitability: findings from Sweden. J Mark
58(3):53–66
Definition 1 Let S be universal set and A is any subset of
Angelova B, Zekiri J (2011) Measuring CS with service quality using
S. Then, T-characteristic function of A is denoted by TA and American CS model (ACSI model). Int J Acad Res Bus Soc Sci
is defined as TA : S ! IR. So, this assigns real number 1(3):232–258
TA ð xÞ to each element x 2 S. Higher the value of TA ð xÞ, Aquilani B, Silvestri C, Ruggieri A, Gatti C (2017) A systematic
literature review on total quality management critical success
larger the value of membership of x 2 S in A. Here IR factors and the identification of new avenues of research. TQM J
denotes set of real numbers. 29(1):184–213. https://doi.org/10.1108/TQM-01-2016-0003
Attaran M, Sharmin A (2007) Collaborative supply chain manage-
Definition 2 Let S be universal set. Then, T-subset A of S ment. Bus Process Manag J 13(3):390–404
is defined as the ordered pair A ¼ fðx; TA ðxÞÞ : x 2 Sg, Aven T (2016) Risk assessment and risk management: review of
where T-characteristic function TA : S ! IR assigns real recent advances on their foundation. Eur J Oper Res 253(1):1–13
number TA ð xÞ as membership value to each x 2 S: Bahari F, Elayidom S (2015) An efficient CRM-data mining
framework for the prediction of customer behaviour. Procedia
The universal set S is not necessarily an ordered set in Comput Sci 46:725–731
imprecise environment. However, the function TA : S ! IR Basu S, Pramanik S, Dey S, Panigrahi G, Jana DK (2019) Fire
imposes an ordering over elements of S based on the value monitoring in coal mines using wireless underground sensor
of TA ð xÞ8x 2 S. network and interval type-2 fuzzy logic controller. Int J Coal Sci
Technol 6(2):274–285
Definition 3 The union of two T-subsets A and B of S, Bellman RE, Zadeh LA (1970) Decision-making in a fuzzy environ-
ment. Manag Sci 17(4):141–164
which is denoted by A [ B, is defined as Bloemhof JM, Soysal M (2017) Sustainable food supply chain design.
A [ B ¼ fðx; TA[B ðxÞÞ : x 2 Sg, where T-characteristic In: Bouchery Y, Corbett C, Fransoo J, Tan T (eds) Sustainable
function TA[B ðxÞ of A[B is defined as supply chains, springer series in supply chain management, vol
TA[B ðxÞ ¼ maxfTA ðxÞ; TB ðxÞg; 8x 2 S. 4. Springer, Cham, pp 395–412
Bowen EF, Paul CD, Richard LC, Farukt AC (2009) The role of
Definition 4 The intersection of two T-subsets A and B of supply management capabilities in green supply. Prod Oper
S, which is denoted by A \ B, is defined as Manag 10(2):174–189
Brandenburg M, Gerd HJ, Rebs T (2018) Social and environmental
A \ B ¼ fðx; TA\B ðxÞÞ : x 2 Sg, where T-characteristic dimensions of organizations and supply chains: trade-offs and
function TA\B ðxÞ of A\B is defined as synergies. Springer, Cham
TA\B ðxÞ ¼ minfTA ðxÞ; TB ðxÞg; 8x 2 S. Campos AC, Mendes J, Valle PO, Scott N (2018) Co-creation of
tourist experiences: a literature review. Curr Issues Tour
Definition 5 The complement of T-subset A of S, denoted 21(4):369–400
by A or Ac , is defined as A ¼ fðx; TAðxÞÞ : x 2 Sg, where Cardozo RN (1965) An experimental study of customer effort,
expectation, and satisfaction. J Mark Res 2(3):244–249
T-characteristic function TA ð xÞ of A or Ac is defined as Carter CR, Easton PL (2011) Sustainable supply chain management:
TAðxÞ ¼ 1  TA ðxÞ; 8x 2 S: evolution and future directions. Int J Phys Distrib Logist Manag
41(1):46–62
Chatterjee K, Pamucar D, Zavadskas EK (2018) Evaluating the
performance of suppliers based on using the R’AMATEL-
MAIRCA method for green supply chain implementation in
References electronics industry. J Clean Prod 184:101–129
Chavez R, Yu W, Feng M, Wiengarten F (2016) The effect of
Agnihotri R, Trainor KJ, Itani OS, Rodriguez M (2017) Examining customer-centric green supply chain management on operational
the role of sales-based CRM technology and social media use on performance and customer satisfaction. Bus Strategy Environ
post-sale service behaviors in India. J Bus Res 81:144–154 25(3):205–220
Ali A, Haseeb M (2019) Radio frequency identification (RFID) Chen L, Zhao X, Tang O, Price L, Zhang S, Zhu W (2017) Supply
technology as a strategic tool towards higher performance of chain collaboration for sustainability: a literature review and
supply chain operations in textile and apparel industry of future research agenda. Int J Prod Econ 194:73–87
Malaysia. Uncertain Supply Chain Manag 7:215–226 Chin TA, Hamid ABA, Rasli A, Tat HH (2014) A literature analysis
on the relationship between external integration, environmental

123
Multi-objective optimization of cost-effective and customer-centric closed-loop supply chain… 177

uncertainty and firm performance in Malaysian SMEs. Procedia Green KW, Zelbst PJ, Meacham J, Bhadauria VS (2012) Green
Soc Behav Sci 130:75–84 supply chain management practices: impact on performance.
Choudhury M, Harrigan P (2014) CRM to social CRM: the Supply Chain Manag Int J 17(3):290–305
integration of new technologies into customer relationship Grimm D, Wosten H (2018) Mushroom cultivation in the circular
management. J Strateg Mark 22(2):149–176. https://doi.org/10. economy. Appl Microbiol Biotechnol 102(18):7795–7803
1080/0965254X.2013.876069 Guu S-M, Wu Y-K (1999) Two-phase approach for solving the fuzzy
Chuang SH, Lin HN (2013) The roles of infrastructure capability and linear programming problems. Fuzzy Sets Syst 107(2):191–195
customer orientation in enhancing customer-information quality Handfield RB, Ernest NL (1999) Introduction to supply chain
in CRM systems: empirical evidence from Taiwan. Int J Inf management. Prentice Hall, Upper Saddle River
Manag 33:271–281 Hennig T, Klee A (1997) The impact of customer satisfaction and
Churchill G, Surprenant C (1982) An investigation into the determi- relationship quality on customer retention: a critical reassess-
nants of customer satisfaction. J Mark Res 19(4):491–504 ment and model development. Psychol Mark 14(8):737–764
Deloitte (2019) Deloitte touche tohmatsu limited. http://www.deloitte. Hsu C-W, Kuo T-C, Chen S, Hu AH (2013) Using DMTEL to
com. Accessed 21 July 2019 develop a carbon management model of supplier selection in
Dubey R, Gunasekaran A, Papadopoulos T, Childe S, Shibin K, green supply chain management. J Clean Prod 56:164–172
Wamba S (2016) Sustainable supply chain management: frame- Imran M, Hamid S, Aziz A, Hameed W (2019) The contributing
work and further research directions. J Clean Prod factors towards e-logistic customer satisfaction: a mediating role
142(2):1119–1130 of information technology. Uncertain Supply Chain Manag
Dubois D, Prade H (2015) The legacy of 50 years of fuzzy sets: a 7:63–72
discussion. Fuzzy Sets Syst 281:21–31. https://doi.org/10.1016/j. Jabbour C, Jabbour A (2016) Green human resource management and
fss.2015.09.004 green supply chain management: linking two emerging Agendas.
Dutta A, Jana D (2017) Expectations of the reductions for type-2 J Clean Prod 112:1824–1833
trapezoidal fuzzy variables and its application to a multi- Janssen PB, Johnson MP, Schaltegger S (2015) 20 years of perfor-
objective solid transportation problem via goal programming mance measurement in sustainable supply chain management—
technique. J Uncertain Anal Appl. https://doi.org/10.1186/ what has been achieved? Supply Chain Manag Int J
s40467-017-0057-4 20(6):664–680. https://doi.org/10.1108/SCM-06-2015-02
Ebrahimnejad A, Verdegay JL (2018) Fuzzy sets-based methods and Jimenez M, Bilbao A (2009) Pareto-optimal solutions in fuzzy multi-
techniques for modern analytics. studies in fuzziness and soft objective linear programming. Fuzzy Sets Syst
computing. Polish Academy of Sciences, Warsaw 160(18):2714–2721
Elhedhli S, Merrick R (2012) Green supply chain network design to Jonkman et al (2019) Selecting food process designs from a supply
reduce carbon emissions. Transp Res Part D Transp Environ chain perspective. J Food Eng 195:52–60
17(5):370–379 Kandampully J, Zhang TC, Jaakkola E (2018) Customer experience
Eltayeb TK, Zailani S, Ramayah T (2011) Green supply chain management in hospitality: a literature synthesis, new under-
initiatives among certified companies in Malaysia and environ- standing and research agenda. Int J Contemp Hosp Manag
mental sustainability: investigating the outcomes. Resour Con- 30(1):21–56. https://doi.org/10.1108/IJCHM-10-2015-0549
serv Recycl 55(5):495–506 Kannan D, Khodaverdi R, Olfat L, Jafarian A, Diabat A (2013)
Fernandes AC, Sampaio P, Sameiro M, Truong HQ (2017) Supply Integrated fuzzy multi criteria decision making method and
chain management and quality management integration: a multi-objective programming approach for supplier selection and
conceptual model proposal. Int J Qual Reliab Manag order allocation in a green supply chain. J Clean Prod
34(1):53–67. https://doi.org/10.1108/IJQRM-03-2015-0041 47:355–367
Flint JD, Christopher BP, Boutin PJ (2011) Customer value antici- Kaur J, Sidhu R, Awasthi A, Chauhan S, Goyal S (2018) A
pation, CS and loyalty: an empirical examination. Ind Mark DEMATEL based approach for investigating barriers in green
Manag 40(2):219–230 supply chain management in Canadian manufacturing firms. Int J
Fornell C (1992) A national CS barometer: the Swedish experience. Prod Res 56(1-2):312–332
J Mark 56(1):6–21 Khansalar E, Bayez MLD, Safari R (2015) Maximizing profit in
Garai A, Mandal P, Roy TK (2016) Intuitionistic fuzzy T-sets based reverse supply chain. Int Bus Res 8(11):111–118. https://doi.org/
optimization technique for production-distribution planning in 10.5539/ibr.v8n11p111
supply chain management. Opsearch 53(4):950–975 Kronborg JJ, Munksgaard KB, Arlbjorn JS (2013) Chasing value
Garai A, Mandal P, Roy TK (2017) Multipollutant air quality offerings through green supply chain innovation. Eur Bus Rev
management strategies: t-Sets Based optimization technique 25(2):124–146. https://doi.org/10.1108/09555341311302657
under imprecise environment. Int J Fuzzy Syst 19(6):1927–1939 Kusi-Sarpong S, Gupta H, Sarkis J (2018) A supply chain sustain-
Garrido-Moreno A, Garcı́a-Morales LV (2014) Paving the way for ability innovation framework and evaluation methodology. Int J
CRM success: the mediating role of knowledge management and Prod Res. https://doi.org/10.1080/00207543.2018.1518607
organizational commitment. Inf Manag 51(8):1031–1042 Li X (2014) Operations management of logistics and supply chain:
Gaussin M, Hu G, Abolghasem S, Basu S, Shankar MR, Bidanda B issues and directions. Discrete Dyn Nat Soc. https://doi.org/10.
(2013) Assessing the environmental footprint of manufactured 1155/2014/701938
products: a survey of current literature. Int J Prod Econ Markova S, Tatjana P-M (2013) Social media and supply chain.
146(2):515–523 Amfiteatru Econ J 15(33):89–102
Geng R, Mansouri S, Aktas E (2017) The relationship between green Melnyk SA, Flynn BB, Awaysheh A (2018) The best of times and the
supply chain management and performance: a meta-analysis of worst of times: empirical operations and supply chain manage-
empirical evidences in Asian emerging economies. Int J Prod ment research. Int J Prod Res 56(1):164–192
Econ 183(A):245–258 Mumtaz U, Ali Y, Petrillo A (2018) A linear regression approach to
Govindan K, Kaliyan M, Kannan D, Haq A (2014) Barriers analysis evaluate the green supply chain management impact on indus-
for green supply chain management implementation in Indian trial organizational performance. Sci Total Environ 624:162–169
industries using analytic hierarchy process. Int J Prod Econ Oliver Wight (2019) https://www.oliverwight-eame.com. Accessed
147(B):555–568 21 July 2019

123
178 A. Garai, T. K. Roy

Petljak K, Zulauf K, Stulec I, Seuring S, Wagner R (2018) Green Siddiqi O (2011) Interrelations between service quality attributes, CS
supply chain management in food retailing: survey-based and customer loyalty in the retail banking sector in Bangladesh.
evidence in Croatia. Supply Chain Manag Int J 23(1):1–15. Int J Bus Manag 6(3):12–36
https://doi.org/10.1108/SCM-04-2017-0133 Singh R, Sandhu HS, Metri BA, Kaur R (2018) Supply chain
Pramanik S, Jana DK, Mondal SK, Maiti M (2015) A fixed-charge management practices, competitive advantage and organiza-
transportation problem in two-stage supply chain network in tional performance: a confirmatory factor model. Oper Serv
Gaussian type-2 fuzzy environments. Inf Sci 325:190–214 Manag. https://doi.org/10.4018/978-1-5225-3909-4.ch054
Rajeev A, Pati R, Padhi S, Govindan K (2017) Evolution of Srivastava SK (2007) Green supply-chain management: a state-of-
sustainability in supply chain management: a literature review. the-art literature review. Int J Manag Rev 9(1):53–80
J Clean Prod 162(20):299–314 Svensson G (2007) Aspects of sustainable supply chain management
Rust RT, Zahorik AJ (1993) Customer satisfaction, customer (SSCM): conceptual framework and empirical example. Supply
retention, and market share. J Retail 69(2):193–215 Chain Manag Int J 12(4):262–266
Ryu K, Lee HR, Kim WG (2012) The influence of the quality of the Testa F, Iraldo F (2010) Shadows and lights of GSCM (green supply
physical environment, food, and service on restaurant image, chain management): determinants and effects of these practices
customer perceived value, customer satisfaction, and behavioral based on a multi-national study. J Clean Prod 18(10-11):953–962
intentions. Int J Contemp Hosp Manag 24(2):200–223 Tseng M, Islam M, Karia N, Fauzi F (2019) A literature review on
Sabri EH, Beamon BM (2000) A multi-objective approach to green supply chain management: trends and future challenges.
simultaneous strategic and operational planning in supply chain Resour Conserv Recycl 141:145–162
design. Omega 28(5):581–598 Uygun O, Dede A (2016) Performance evaluation of green supply
Saeidi SP, Sofian S, Saeidi P, Saeidi SP, Saaeidi SA (2015) How does chain management using integrated fuzzy multi-criteria decision
corporate social responsibility contribute to firm financial making techniques. Comput Ind Eng 102:502–511
performance? the mediating role of competitive advantage, Walker H, Jones N (2012) Sustainable supply chain management
reputation, and customer satisfaction. J Bus Res 68(2):341–350 across the UK private sector. Supply Chain Manag Int J
Sakawa M, Matsui T (2012) An interactive fuzzy satisficing method 17(1):15–28
for multi-objective stochastic integer programming with simple Wei JT, Lee MC, Chen HK, Wu HH (2013) Customer relationship
recourse. Appl Math 3:1245–1251 management in the hairdressing industry: an application of data
Sakawa M, Yano H, Nishizaki I (2013) Linear and multi-objective mining techniques. Expert Syst Appl 40(18):7513–7518
programming with fuzzy stochastic extensions. Springer, New Wei G, Wang H, Zhao X, Lin R (2014) Hesitant triangular fuzzy
York information aggregation in multi-attribute decision making.
Salehi S, Selamat A, Mashiinchi MR, Fujita H (2015) The synergistic J Intell Fuzzy Syst Appl Eng Technol 26(3):1201–1209
combination of particle swarm optimization and fuzzy sets to Wieland A, Handfield RB, Durach CF (2016) Mapping the landscape
design granular classifier. Knowl Based Syst 76:200–218 of future research themes in supply chain management. J Bus
Sarkar B (2019) Mathematical and analytical approach for the Logist. https://doi.org/10.1111/jbl.12131
management of defective items in a multi-stage production Wu Y-K, Liu CC, Lur YY (2015) Pareto-optimal solution for multi-
system. J Clean Prod 216:896–919. https://doi.org/10.1016/j. objective linear programming problems with fuzzy goals. Fuzzy
jclepro.2019.01.078 Optim Decis Mak 14(1):43–55
Sarkar B, Guchhait R, Sarkar M, Pareek S, Kim N (2018) Impact of Zablah AR, Carlson BD, Donavan DT, Maxham JG, Brown T (2016)
safety factors and setup time reduction in a two-echelon supply A cross-lagged test of the association between CS and employee
chain management. Robot Comput Integr Manuf. https://doi.org/ job satisfaction in a relational context. J Appl Psychol
10.1016/j.rcim.2018.05.001 101(5):743–755
Sarkis J (2003) A strategic decision framework for green supply chain Zadeh LA (1965) Fuzzy sets. Inf Control 8(3):338–353
management. J Clean Prod 11(4):397–409 Zhang M, Guo H et al (2019) Linking supply chain quality integration
Sarkis J, Qinghua Z, Lai KH (2011) An organizational theoretic with mass customization and product modularity. Int J Prod
review of green supply chain management literature. Int J Prod Econ. https://doi.org/10.1016/j.ijpe.2017.01.011
Econ 130(1):1–15 Zhou S, Lam J, Zheng WX (2007) Control design for fuzzy systems
Sarvestani HK, Zadeh A, Seyfi M, Barzoki MR (2019) Integrated based on relaxed non-quadratic stability and H! performance
order acceptance and supply chain scheduling problem with conditions. IEEE Trans Fuzzy Syst 15(2):188–199
supplier selection and due date assignment. Appl Soft Comput Zhu Q, Sarkis J, Lai KH (2008) Confirmation of a measurement
75:72–83. https://doi.org/10.1016/j.asoc.2018.10.045 model for green supply chain management practices implemen-
Secchi E, Roth A, Verma R (2018) The impact of service tation. Int J Prod Econ 111(2):261–273
improvisation competence on customer satisfaction: evidence Zhu Q, Sarkis J, Lai KH (2013) Institutional-based antecedents and
from the hospitality industry. Prod Oper Manag. https://doi.org/ performance outcomes of internal and external green supply chain
10.1111/poms.12969 management practices. J Purch Supply Manag 19(2):106–117
Seuring S (2013) A review of modeling approaches for sustainable Zimmermann H-J (1976) Description and optimization of fuzzy
supply chain management. Decis Support Syst 54(4):1513–1520 systems. Int J Gen Syst 2(4):209–215
Seuring S, Gold S (2013) Sustainability management beyond Zimmermann H-J (1978) Fuzzy programming and linear program-
corporate boundaries: from stakeholders to performance. ming with several objective functions. Fuzzy Sets Syst
J Clean Prod 56:1–6 1(1):45–55
Seuring S, Muller M (2008) From a literature review to a conceptual Zimmermann H-J (1985) Applications of fuzzy set theory to
framework for sustainable supply chain management. J Clean mathematical programming. Inf Sci 36(1–2):29–58
Prod 16(15):1699–1710
Sheu BJ, Chou YH, Hu CC (2005) An integrated logistics operational Publisher’s Note Springer Nature remains neutral with regard to
model for green-supply chain management. Transp Res Part E jurisdictional claims in published maps and institutional affiliations.
Logist Transp Rev 41(4):287–313

123

You might also like