An integration of environmental awareness into flexible supply chains: a trade-off between costs and environmental pollution.

Today, environmental awareness is highly interested in supply chains and logistics networks with regard to sustainable development goals. This proposes a bi-objective linear mathematical model comprising supply chain flexibility dimensions. The proposed model is to integrate environmental considerations into a flexible supply chain as an optimization framework. The first objective function is to minimize the costs, while the second one minimizes the environmental impacts of automotive industry. The goal of this paper is to find a trade-off between the total cost and the environmental pollution with regard to the supply chain flexibility dimensions. We suggest adding four different supply chain flexibility dimensions to the model which are budget for transportation, trained labor team to help the packaging process, number of active plants, and outsourcing the painting process flexibilities to curb harmful emissions from factories while reducing the costs. Six flexibility scenarios are proposed in this study to do the sensitivity analysis. The model is applicable with the use of a real data set derived from an automotive parts factory located in Iran. We use an improved augmented ε-constraint method to address the proposed bi-objective optimization framework. The results show that choosing the model with all flexibility dimensions is the best initiative to promote sustainable development, since it leads to a significant reduction in costs and environmental pollution.

Introduction

Today, there is a great deal of concern with the environmental pollution in the automobile industry as one of great industries in Iran (Fahimnia et al. 2013). The environmental impact of this industry is viewed as a particular environmental concern in recent years (Fard et al. 2017). Environmentalists believe that car factories can generate as much pollution as a car being driven (Liu et al. 2020). Companies need to be able to measure their carbon footprints (Fathollahi-Fard et al. 2020a). It is a demanding job to reduce the size of carbon dioxide while producing car parts (Karampour et al. 2020). Both manufacturing and the exhaust pipe emissions of cars are two factors which must be controlled. This leads to reduce the impact of the use and production of vehicles on the environment (Saha et al. 2021). Hence, we can improve the air quality and public health (Fathollahi-Fard et al. 2020b).

It goes without saying that the age automotive manufacturers are developing technologies to help reduce these impacts by applying the most advanced manufacturing processes in economically developed countries (Yu et al. 2021; Zhang et al. 2020). Yet there is still a need in developing countries to promote sustainable development and raise awareness of environmental issues (Golmohamadi et al. 2017). For example, while delivering end products to end users, more fuel-efficient cars can be used (Fathollahi-Fard et al. 2020c).

In this paper, we put forward a suggestion on the issue of environmental impact of automobile industry and propose a flexible supply chain to curb harmful emissions from factories producing toxic waste. In our case study, there is no chance of existence of carbon offset projects. The model with flexibility dimensions is a bi-objective model in which the first objective function is to minimize the total costs of the company while the second one minimizes the level of pollution caused by car part factories. We compare six different scenarios to find out the best flexibility model with the minimum level of pollution and costs. We apply the improved version of augmented ε-constraint. The results show the need for flexibility in our case study. The results are analyzed using some collected data from an automotive parts factory located in Iran.

The flexibility dimensions of the model are budget for transportation, trained labor team to help the packaging process, number of active plants, and outsourcing the painting process. We believe that being flexible is a strategy that companies can map out to cut pollution.

The rest of this study is: Section 2 presents a survey on the important and recent works. Section 3 establishes the proposed model. In Section 4, the solution method is clarified (the improved version of augmented ε-constraint method named AUGMECON2). In Section 5, we illustrate the results of the model. In the last section, we draw the conclusions and make recommendations for future research.

Literature review

In the area of supply chain management, flexibility has become one of the main research topics. Previous studies have reviewed a number of advances in supply chain management (Hajiaghaei-Keshteli and Sajadifar 2010; Hajiaghaei-Keshteli et al. 2011). Moreover, the need to study flexibility in supply chains has been highlighted. What matters the most to car manufacturers is being able to compete with new rivals if they are to survive and thrive.

Being flexible is defined as the ability of firms to adapt quickly and efficiently to changes in the environment (Swafford et al. (2006), Mavrotas and Florios (2013)) and to respond to unpredictable changes such as customers’ demands and competitor actions (Moon et al. (2012)). Former papers have mainly focused on strategic supply chain flexibility with a more emphasis on non-modeling studies (Esmaeilikia et al. (2014b)). Being flexible can be defined in different ways for the reason that it has different dimensions. For instance, Archer et al. (2006) proposed the concept of supply chain flexibility (SCF) with optimization model and demand uncertainty. Generally, the SCF aims to respond quickly to any unpredictable changes in customers’ demand or supply and demand (e.g., Merschmann and Thonemann (2011), Moon et al. (2012), Malhotra and Mackelprang (2012)).

Vickery et al. 1997 and Martínez Sánchez and Pérez Pérez (2005) have classified flexibility dimensions as follows: product flexibility, volume flexibility, routing flexibility, delivery, trans-shipment, postponement flexibility, sourcing flexibility, launch flexibility, and access flexibility.

Ceryno et al. (n.d.) reviewed papers and classified them into three categories: the theoretical, empirical, and theoretical empirical papers. Stevenson and Spring (2007) also did a review of existing literature on supply chain flexibility and clearly defined flexibility in the context of supply chains. According to the mentioned articles, there is still a need for empirical studies in the field of supply chain flexibility (Esmaeilikia et al. (2014a)).

The SCF can be measured; for instance, Gong (2008) presented a supply chain flexibility model with different flexibility dimensions in which total system flexibility was measured by an economic index. The author indicated that some flexibility factors were more effective than others. Esmaeilikia et al. (2014b) also suggested a tactical supply chain planning model with several flexibility options.

The outcomes from earlier researches show that most review papers mainly stressed the modeling characteristics, but they did not thoroughly discuss the topic of flexibility (Fahimnia et al. (2013), Min and Zhou (2002), Mula et al. (2010), Schütz and Tomasgard (2011)). The main focus of their attention was non-modeling studies on strategic SCF.

The green supply chain management (GSCM) is also imported in the area of SCF from many previous studies. The GSCM can be defined as be the ability to integrate environmental thinking into supply chain management; it can be applied in different areas of study such as product design, material sourcing, manufacturing process, and product delivery.

Former researchers have reviewed the papers on green supply chain management. Malviya and Kant (2015) presented and categorized a review of the existing studies on green supply chain. Min, H. and Kim, I. (2012) also in their paper traced the evolution of green supply chain research with the flexibility and resiliency dimensions.

Ahi and Searcy (2013) proposed a survey on the recent trends of GSCM with sustainable development goals. The study of Sarkis et al. (2011) is also another review paper in the field of sustainable supply chain and GSCM. Srivastava (2007) presented a state-of-the-art literature review of GSCM integrating the whole gamut of activities in the area. Their literature review highlighted the ongoing integration process in GSCM. Sarkis and Zhu et al. (2011) reviewed the literature on GSCM with a focus on identifying applicable and explanatory organizational theories that have been utilized to expand understanding and knowledge of this research field.

Similarly, Alexander et al. (2014) did a review on the existing literature and used a systematic and interdisciplinary approach in tandem with the lead time perspective. Shan and Wang (2018) visualized the research on the GSCM with an introduction to the SCF. There are various papers that integrated mathematical modeling into green supply chain. As a case in point Kannan et al. (2015) proposed a model to evaluate the criteria of GSCM with a supplier assessment.

Likewise, Mohammed and Wang (2017) investigated a three-echelon meat supply chain in a fuzzy environment. Their model includes four minimization objectives, i.e., the total transportation and implementation cost, the amount of CO2 emissions, and the distribution time of productsKumar et al. (2017) used fuzzy analytical hierarchy process to assess a supplier selection with the use of a fuzzy multi-objective linear programming.

Kannan et al. (2015) in their paper suggested the criteria that impacted on GSCM implementation built on former studies. Based on their proposal, companies can reach balanced economic and environmental performance. Our study builds on earlier work of two research (Gong (2008), Esmaeilikia et al. (2014a)) concerning environmental issues.

Recently, Nezhadroshan et al. (2020) proposed a robust possibilistic optimization model for the relief products. They optimized the total cost and travel time of emergency logistics with resiliency dimensions. Karampour et al. (2020) proposed a GSCM with an introduction to the total cost and green emissions of the back order. They applied multi-objective red deer algorithm, Keshtel algorithm, and the non-dominated sorting genetic algorithm. Fathollahi-Fard et al. (2020d) proposed the red deer algorithm for the routing optimization and SCF with the fixed charge. At last but not the least, an adaptive social engineering optimizer was used for a routing and scheduling of the healthcare systems with resiliency dimensions.

More recently, Sabuj et al. (2021) provided an extensive framework to analyze the relationships of the environmental sustainability based on circular economy concept. Ali et al. (2021) developed an integrated model combining fuzzy analytical hierarchy process with Delphi method to assess the drivers and factors for supply chain disruption. Karmaker et al. (2021) developed an integrated model for improving the supply chain sustainability in the context of COVID-19 pandemic in Bangladesh. Mahtab et al. (2021) considered the multi-vehicle supposition for the evolution of supply chain resiliency for a flood in Bangladesh. Shahed et al. (2021) evaluated the supply chain mitigation for COVID-19 pandemic in Bangladesh. Islam et al. (2021) proposed a novel prediction model based on a grey particle swarm optimization for ranking the warehouse performance. Marimuthu et al. (2021) evaluated the triple bottom line of sustainability for mining industry in Bangladesh for implication of resource policies in emerging economics. Salehi-Amiri et al. (2021) developed a sustainable closed-loop supply chain for walnut industry in Iran. Zahedi et al. (2021) proposed a closed-loop supply chain with multi-task and carbon tax. Finally, Zahedi et al. (2021b) evaluated a supply chain resiliency with internet of things technologies in the case of COVID-19 pandemic.

The main research gap is the lack of supply chain flexibility in corporation with supply chain environmentally. In this regard, the objective of our study is to gain a profound managerial insight into production process of green supply chains that are less harmful to the environment.

Proposed problem

Problem description

Figure 1 shows the supply chain in this study with four different levels: suppliers, production plants, warehouses, and end users. The following parameters are assumed known in this study:

Fig. 1

Schematic supply chain network

Values of each end user’s demand forecast for each product, number, location and capacities of plants and warehouses, respectively; capacity limitations imposed by machinery allowed operation times; restrictions on capacity of raw material supply; limitations on storage capacity in plants and warehouses; transportation capacities; and fixed operation costs of plants. The components of products are purchased from different suppliers. End products will either be warehoused and then be delivered directly to the end user or taken out of the plants to the warehouses. They are also delivered either to end users or alternatively taken to the warehouses before delivery.

In this study, a mixed integer linear programming model for a supply chain with multiple dimensions of flexibility is proposed. This bi-objective model minimizes the total production cost and the amount of carbon dioxide and other toxic waste in the environment. Total production cost successively comprises costs of opening manufacturing plants, regular time production, overtime production, holding components in plants, holding products in plants, transportation from plants to end users, transportation from warehouses to end users, warehouse storage and handling, stock-out, and labor regular time and overtime wages. Also, the costs related to environmental issues that are cost of outsourcing the painting process, using more advanced cars for delivering end product and the penalty the company has to pay to regulators while harming the environment during its production process. In the second objective function, we aim at a reduction in the emission of carbon dioxide into the environment as well as toxic wastes.

Notations

Notations of the model are given in Tables 1, 2, and 3 for indices, parameters, and decision variables.

Optimization model

A bi-objective flexibility model is developed to examine the role of our flexibility in our model. This can also help the company provide highly adaptable decisions. In our study, we have developed a mathematical model based on the models of Esmaeilikia et al. (2014a) and Gong (2008).

The two objective functions and constraints are shown:

Subject to:

Equation (1) and Eq. (2) show the two objective functions. Equation (3) and Eq. (4) show material balance constraint in production. Equation (5) shows warehouses balance constraint. Equation (6) ensures demand satisfaction. Equation (7) shows the minimum supplier contribution constraint. Equation (8) shows restriction on outsourcing capacity. Equation (9) and Eq. (10) show restriction on production capacity in regular time and overtime production. Equation (11) shows restriction on outsourcing the products. Equation (12) shows restriction on number of held components in plants. Equation (13) shows restriction on number of held products in plants. Equation (14) shows restriction on number of products in warehouses. Equation (15) shows the cost of not meeting the customers’ demands that are lost sales. Equation (16) represents restriction on transportation from plants to end users. Equation (17) shows restriction on transportation from plants to warehouses. Equation (18) shows restriction on transportation from warehouses to end users. Equation (19) indicates restriction on transportation capacity. Equation (20) and Eq. (21) are total machine working time constraints; Eq. (22) and Eq. (23) are total labor working time constraints, respectively, in regular time and overtime. Equation (24) and Eq. (25) are production time and labor working time constraints, in that order. Equation (26) and Eq. (27) show restriction on time for production correspondingly in regular time and overtime. Equation (28) to Eq. (36) show the non-negativity constraints of decision variables. In the revision, we apply the improved version of augmented ε-constraint method (AUGMECON2) proposed by Mavrotas and Florios (2013). Since we have novelty in this method, the details about this exact algorithm is not provided in the text.

Case study

Forgekar is an Iranian car parts producer that provides two main local auto manufacturers and after-markets with mainly primarily seat runners for family cars. It is situated in Qazvin (150 km northwest of Tehran, where end users’ plants base) and produces are existed. Two labor teams work in double shifts which are the regular time and the overtime shifts. Located in the factory premises, there are three production plants where two main products are made. A number of machine types are used in the production plants (e.g., hydraulic press, milling machine, drill, guillotine, blow press, compressor). And raw materials can be bought from three different suppliers for production plants. We gathered information and collected data by interviewing the managers, employees, labors, and engineers to obtain the data by administering questionnaires that were completed by them.

To examine the role of flexibility dimensions in raising awareness of environmental issues such as a carbon dioxide emissions and toxic wastes. We believe that being flexible can help companies produce less pollution and offset carbon dioxide emissions to achieve sustainable development. We can see they are both the pollution produced from the production process or from the vehicles delivering the final product. The products are usually packaged in boxes of 10. We suggest a new way of packaging the products in which almost all the products are located in one package as the to end user the package is considered as trash and we can preserve natural resources by changing the packaging method.

In this study, we compare six different scenarios shown in Table 4 to find the best fitted one to our case study, and also make an analysis. The results are shown in the following section.

Table 4

Different flexibility scenarios

Scenario number	Flexibility dimension
1	All
2	None
3	Plant
4	Labor team for packaging and
5	Outsourcing the painting process
6	Budget for transportation and vehicles

Results

In this section, the results of the model are expressed numerically in Table 5. The proposed mathematical model is coded in LINGO 11.0. For different scenarios of flexibility dimensions, the created model is solved. In the first scenario, all flexibility constraints exist in the model. In the second one, all flexibility constraints have been omitted from the model. In the rest, only one flexibility dimension exists respectively. The objective function values (costs) are also shown in Table 5. The results in Fig. 2 show that the scenario including all four flexibility dimensions brought quite a great improvement comparing to the other ones, although there is another scenario with a slight difference in value. The scenario with labor team flexibility dimension has fairly a greater value comparing to the other scenarios. As it was stated earlier, adding flexibility to manufacturing systems can be costly. In order to keep costs down and have a more flexible system, there are other scenarios which can be chosen as the effective ones for the company. If the company is able to add just one flexibility dimension, the scenario with labor team flexibility can be considered as one of the best cases, for it does not conclude budget flexibility and its value is close to the least. In this problem, we have a bi-objective problem with payoff values shown in Table 6. We used AUGMECON2 to overcome all the weak points of the conventional ε-constraint. In this method, different values are considered for epsilon (grid points), which is the environmental impacts in this model, to find the optimum point for objective functions avoiding unnecessary repetitions.

Table 5

Costs in different scenarios

Scenario number	Costs (IRR)
1	1953723
2	2569578
3	2556097
4	1959875
5	2510720
6	2514526

Fig. 2

Model results: effect of adding flexibility dimensions to the model

Table 6

Payoff values

	f1	f2
Minf1	19537230	548
Minf2	19319930	0

In Fig. 3a to Fig. 3d, we compare some decision variables in each scenario. Figure 3a demonstrates that the scenario with all flexibility dimensions has the greatest number of components which means manufacturing more products as well as Fig. 3b. It is illustrated in Fig. 3c that the number of products to be stored is less in scenario with all flexibility dimensions; this leads to a reduction in costs of storage. In Fig. 3d, scenarios with all flexibility dimensions and labor team flexibility show the company’s needs to buy less products from its rival company which means having a great number of products made by the company itself and leads to a decrease in costs.

Fig. 3

a The effect of adding flexibility dimensions on the number of components purchased from suppliers for plants. b The effect of adding flexibility dimensions on the number of products made by labor teams in plants in regular time. c The effect of adding flexibility dimensions on the number of products held in warehouses. d The effect of adding flexibility dimensions on the number of products outsourced by plants

Sensitivity analysis of two effective parameters on supply chain flexibility is shown in Fig. 4a and b, while other parameters are assumed fixed. The results are hardly surprising; for larger values of lost sales cost of one unit of product i in plant p in period t illustrated in Fig. 4a result in marked cost aggregation in the final attribute set. The same occurs for the time needed to make one unit of product i by labor team l in plant p in period t, which is demonstrated in Fig. 4b.

Fig. 4

a Sensitivity analysis of the cost of lost sales of one unit of product i. b Sensitivity analysis of environmental impacts to make one unit of product

In order to conduct sensitivity analysis of products’ environmental impacts, the main model with all flexibility dimensions has been selected. To increase understanding of the relationships between environmental impacts and costs, we performed a comparative sensitivity analysis of environmental impacts (epsilon). The first objective function is divided to four separate sections which are production and purchase costs, warehouse costs, costs of production, and fixed and labor costs, respectively. From the obtained results which are shown in Table 7, it can be realized that effective sections of the first objective function are sections one to three that imply the existence of an indirect correlation. This means the correlation coefficient is − 1. It can be reasonably inferred that the higher environmental impacts values result in less costs and vice versa. For further elaboration, we have calculated the slopes of the lines of each section (lines one to four) and compared their slopes. Using this data, it can be deduced that slight changes in environmental impacts contribute to significant changes in storage costs. Two other lines, which are related to production and purchase costs, can be interpreted in the same way, but the slope is gentler. As a final point, the last section with slope of zero that shows the fixed and labor costs is not related to environmental impacts at all.

Table 7

Sensitivity analysis for environmental impacts

Environmental impacts	Section 1 (IRR)	Section 2 (IRR)	Section 3 (IRR)	Section 4 (IRR)
800	59753140	12252000	19531570	28412700
1000	59688140	12033820	19527080	28412700
1200	59623140	11815860	19522590	28412700
1400	59558140	11597900	19518090	28412700
1600	59493140	11379600	19513600	28412700
1800	59428140	11161200	19509110	28412700
2000	59363140	10942800	19504620	28412700

Conclusions

In this section, we first discuss the results of our research on supply chain flexibility and then outline the area for future study. Supply chain flexibility has become a major topic of research for academics. The objective of this paper was to explore the effect of adding flexibility to a supply chain model to improve the supply chain performance and reduce the pollution caused by production process; hence, our primary contribution is that we found the empirical support for adding flexibility dimensions to a supply chain model. A supply chain flexibility model including trained labor team flexibility for packaging, sourcing flexibility, production line flexibility, and budget flexibility was developed to find the best scenario among all these flexibility dimensions. To establish a link between flexibility and supply chain performance, we proposed a bi-objective model and solved it applying AUGMECON2. Our case study which was an automotive parts factory located in Iran evidenced an application of flexibility. Furthermore, this demonstrated the different dimensions of flexibility in a supply chain and reinforced the idea that the relationship between flexibility and supply chain performance in view of the environmental effects is of a major consideration. Having compared the first objective function values in different flexibility scenarios, the need for being flexible in this production plant arose. The model with all flexibility dimensions had less objective function values, which meant a reduction in costs of the company and pollution as well. For this specific case study, choosing the scenario with all flexibility dimensions appeared to be reasonable as it had the least objective function among all.

Our results can provide important managerial insights for supply chain managers in the automobile industry. For example, the company could choose either to have all flexibility dimensions or the best one among the others in terms of costs and environmental issues. In our case study, choosing labor flexibility could also be a rational choice to make for the reason that it is the second reasonable scenario of all. Our secondary contribution was that we found a relationship between flexibility and environmental issues. Since the products needed to be produced with less harm to the environment, the company had to be flexible in the production parts that can damage the environment so as not to be charged heavy financial penalties for pollution they make.

Some of the flexibility dimensions have been incorporated to the model. But there are still other flexibility dimensions that can be added to our model. A fertile area for future research would be time horizon as one of the effective elements in modeling. To extend the discussion of this paper, it is worth modeling the supply chain in an uncertain environment with uncertain variables. Demand as one of the major factors of modeling could be given uncertain values. Moreover, the model can be generalized to other industries to develop a general flexibility model. Another possible future research is to study other factors related to environmental issues on the systems (Fathollahi-Fard et al. 2020e).

Table 1

Notations of indices

I	Product type
P	Production plant
C	Component type
T	Time
E	End user
S	Supplier
L	Labor team
M	Machine type
W	Warehouse

Table 2

Notations of parameters

Parameter	Definition
diet	Demand for product i by end user e in period t
fpt	Fixed costs of plant p in period t
aipt	Overhead production cost of one unit of product i in plant p in regular time in period t
aipt′	Overhead production cost of one unit of product i in plant p in overtime in period t
ript	Cost to buy one unit of product i in plant p in period t (outsourcing cost)
hcpt	Cost to hold one unit of component c in plant p in period t
hipt′	Cost to hold one unit of product i in plant p in period t
bipt	Cost of lost sales of one unit of product i in plant p in period t
b′ipt	Maximum number of lost sales of product i in plant p in period t
noci	Number of units of component used to make one unit of product i
wcpt	Maximum capacity of plant p to hold one unit of component c in period t
wipt′	Maximum capacity of plant p to hold one unit of product i in period t
uipt	Maximum production capacity of product i in plant p in regular time in period t
uipt′	Maximum production capacity of product i in plant p in overtime in period t
uipt′′	Maximum number of product i that can be outsourced for plant p in period t
pccspt	Cost to purchase one unit of component c from supplier s for plant p in period t
pqcsptmin	Minimum number of component c can be purchased from supplier s for plant p in period t
pqcsptmax	Maximum capacity of supplier s to sell component c to plant p in period t
εstmin	Minimum number of component can be purchased from supplier s in period t
T_ipet	Cost of transportation one unit of product i from plant p to end user e in period t
Tipwt′	Cost of transportation one unit of product i from plant p to warehouse w in period t
Tiwet′′	Cost of transportation one unit of product i from warehouse w to end user e in period t
tqipetmin	Minimum number of product i transportation from plant p to end user e in period t
tqipetmax	Maximum number of product i transportation from plant p to end user e in period t
tq^′iwetmin	Minimum number of product i transportation from warehouse w to end user e in period t
tq^′iwetmax	Maximum number of product i transportation from warehouse w to end user e in period t
tq^"ipwtmin	Minimum number of product i transportation from plant p to warehouse w in period t
tq^"ipwtmax	Maximum number of product i transportation from plant p to warehouse w in period t
ε′tmin	Minimum number of transportation in period t
kiwt	Cost to hold one unit of product i in warehouse w in period t
kiwt′	Maximum capacity of warehouse w for product i in period t
ttipet	Needed time to transport one unit of t product i from plant p to end user e in period t
ttiwet	Needed time to transport one unit of product i from warehouse w to end user e in period t
waiplt	Wage of labor team l for making one unit of product i in plant p in regular time in period t
wa′iplt	Wage of labor team l for making one unit of product i in plant p in overtime in period t
nlpt	Number of labor team l in plant p in period t
nmpt	Number of machines m in plant p in period t
nccmt	Number of component c processed in machine m in period t
βiplt	Time needed to make one unit of product i by labor team l in plant p in period t
δmt	The probability of machine m working properly in period t
tmpmt	Total machine m working time in plant p in period t
tlplt	Total labor l working time in plant p in regular time in period t
tl′plt	Total labor l working time in plant p in overtime in period t
tccpmt	Time needed to process one unit of component c by machine m in plant p in period t
αipmlt	Maximum capacity of production time of product i in regular time on machine m in plant p by labor l in period t
α′ipmlt	Maximum capacity of production time of product i in overtime on machine m in plant p by labor l in period t
Ds1	Distance between plant p and end user e in period t
Ds2	Distance between warehouse w and end user e in period t
DS3	Distance between plant p and warehouse win period t
Vlpipt	Volume of toxic waste produced by painting product i in plant p in period t
Vlwipt	Volume of waste produced by product i in plant p in period t
Cpipt	Cost of outsourcing the painting process of product i in plant p in period t
Ccipt	Cost of using advanced cars with less CO2 emission to deliver product i from plant p in period t
Peipt	Cost of damaging the environment by production process of product i in plant p in period t
Prp	Probability of outsourcing the painting process
Prc	Probability of using advanced cars with less CO2 emission
Vco2ipt	Volume of CO2 produced by transportation of product i from plant p in period t

Table 3

Notations of decision variables

Decision variable	Definition
QCSPt	Number of component c purchased from supplier s for plant p in period t
Aiplt	Number of product i made by labor team l in plant p in regular time in period t in period t
Aiplt′	Number of product i made by labor team l in plant p in overtime in period t in period t
Ript	Number of product i outsourced by plant p in period t
Xcpt	Number of component c held in plant p in period t
Xipt′	Number of product i held in plant p i in period t
Jipet	Number of product i transported from plant p to end user e in period t
Jipwt′	Number of product i transported from plant p to warehouse w in period t
Jiwet′′	Number of product i transported from warehouse w to end user e in period t
Yiwt	Number of product i held in warehouse w in period t
Bipt	Number of product i backordered in plant p in period t
UC_cspt=1,If componentcis bought from suppliersfor plantpiin periodt0,If componentcis not bought from suppliersfor plantpiin periodt
V_ipet=1,If productiis transported from plantptoenduserein periodt0,If productiis not transported from plantptoenduserein periodt
Vipwt′=1,If productiis transported from plantpto warehousewin periodt0,If productiis not transported from plantpto warehousewin periodt
Viwet′′=1,If productiis transported warehousewtoenduserein periodt0,If productiis not transported warehousewtoenduserein periodt
G_cspt=1,If plantpis open and produces productiin periodt0,If plantpis not open in periodt
ENV_cspt=1,IfCO2emission exceeds the emission performance standard0,IfCO2emission doesn'texceed the emission performance standard