[ieee 2013 5th international conference on modeling, simulation and applied optimization (icmsao...

5
978-1-4673-5814-9/13/$31.00 ©2013 IEEE Selecting Configuration Of Reverse Logistics Network Using Sustainability Indicators Soumaya DHIB, Taicir LOUKIL Higher Institute of Industrial Management of Sfax, Sfax, Tunisia [email protected] Sid-Ali ADDOUCHE, Abderrahman EL MHAMEDI LISMMA – Univ. Paris8, 140 rue de la nouvelle France, 93100 Paris, France Abstract—The network of the reverse logistics (RL) aims to treat and re-inject into the supply chain all that can be valorized from products which are defective, at the end of life, at the end of warranty, at high obsolescence level, etc. The design of that network must take into account many things like the uncertainty about the expected volumes of these products, the forecast of consumer needs customers, producers returns projections, recycling systems, etc. In general, products are not always accompanied with complete data. Those data are often imprecise, hypothetical, inconsistent ... Our literature review is interested mainly in economic viability of an RL organization for a family of products in the context of uncertainty. It shows that almost all research papers do not take into account the incompleteness of the data, do not capitalize on the practices and the historical data of the network and, then, do not consider any indicator of sustainable development (SD). In this paper, we develop a model to select the best reverse logistic network under uncertainty of products returns. This model uses mathematic model and Bayesian network to detect the distribution of used product, integrated in Arena Software to simulate different configurations. Keywords—Reverse Logistics; network configuration; sustainability indicators; Bayesian network I. INTRODUCTION Recently, Reverse Logistics (RL) networks become more complex, due to the rapid changes, competition and technical innovation in market. Successful enterprises recognize the establishment of more and more complex flow networks of returns and goods to be recovered. Precisely, RL refers to the distribution activities involved in product returns, source reduction, conservation, recycling, substitution, reuse, disposal, refurbishment, repair and remanufacturing [1]. In other words, RL is the movement of the goods from a consumer towards a producer in a channel of distribution Murphy [2]. One can also consider RL definition of Revlog [3], “the process of planning, implementing and controlling flows of raw materials, in process inventory, and finished goods, from the point of use back to the point of recovery or point of proper disposal”. Thus, good management of reverse logistics can make the reintroduction of raw materials and recycled goods economically viable. This contributes to sustainable development of the supply chain. So, various frameworks Seitz [4] De Brito [5] have developed designing RL based on sustainable development criteria: economic, environmental and social ones. More frequently, researchers focused on economic criteria to make a strategic decision as well as the design of RL network … Stock [1] Si-bo and Wei- Lai [6] Kim and al. [7]. The design of reverse logistic network requires an economical and environmental quantification. It concerns, for example, the choice of deposit locations, their capacities, quantity of returned product, etc. However, in real situations, data related to reverse logistics are often imprecise, hypothetical and inconsistent. Our framework focuses on choosing the best network reverse logistics under uncertainty. Based on development sustainable indicators, decision maker can choose the suitable decision system. The rest of the paper is organized as follows: In section 2, literature review presents the different reverse logistic networks. Section 3 describes our proposed methodology to choose the best RL configuration. Section 4 describes numerical example including test scenarios and numerical results II. LITERTURE REVIEW Many previous researches on the problems of logistics network design that consider the facility locations and allocation models, have been widely discussed Si-bo and Wei- Lai [6] Kim and al. [7] Lu and al. [8]. These models allow industry actors to tack a variety of decisions as determining the number of these manufacturing centers to find a minimum cost of location-allocation. Based on two main centers of network, Lu and al. [8] developed a problem of multi-objectifs of reverse logistic have network to determine processing center. In their research, Zhou [9] have proposed Mixed Linear model in localization allocation problem in strategic situation that minimize the number of installation center of forward chain and to increase numbers of recovery. Most studies of the existing network models Pourmohammadi and al. [10], Meng, [11] and Mutha and Pokhal [12] have considered the economic viability to select the best reverse network. Pourmohammadi and al. [10] have developed a mixed integer linear model for recovery industrial waste of aluminum in the city of Los Angelos in the United states that minimize the cost to select the best reverse network, considering fixed availability of process area. Meng (2008) has focus on designing and configuration network to minimize the cost of the electronics waste using a mathematical formulation of probabilistic parameters. Mutha and Pokhal [12] has considered the best reverse network which can be

Upload: abderrahman

Post on 27-Jan-2017

218 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: [IEEE 2013 5th International Conference on Modeling, Simulation and Applied Optimization (ICMSAO 2013) - Hammamet (2013.4.28-2013.4.30)] 2013 5th International Conference on Modeling,

978-1-4673-5814-9/13/$31.00 ©2013 IEEE

Selecting Configuration Of Reverse Logistics Network Using Sustainability Indicators

Soumaya DHIB, Taicir LOUKIL Higher Institute of Industrial Management of Sfax,

Sfax, Tunisia [email protected]

Sid-Ali ADDOUCHE, Abderrahman EL MHAMEDI

LISMMA – Univ. Paris8, 140 rue de la nouvelle France, 93100

Paris, France

Abstract—The network of the reverse logistics (RL) aims to

treat and re-inject into the supply chain all that can be valorized from products which are defective, at the end of life, at the end of warranty, at high obsolescence level, etc. The design of that network must take into account many things like the uncertainty about the expected volumes of these products, the forecast of consumer needs customers, producers returns projections, recycling systems, etc. In general, products are not always accompanied with complete data. Those data are often imprecise, hypothetical, inconsistent ... Our literature review is interested mainly in economic viability of an RL organization for a family of products in the context of uncertainty. It shows that almost all research papers do not take into account the incompleteness of the data, do not capitalize on the practices and the historical data of the network and, then, do not consider any indicator of sustainable development (SD). In this paper, we develop a model to select the best reverse logistic network under uncertainty of products returns. This model uses mathematic model and Bayesian network to detect the distribution of used product, integrated in Arena Software to simulate different configurations.

Keywords—Reverse Logistics; network configuration; sustainability indicators; Bayesian network

I. INTRODUCTION Recently, Reverse Logistics (RL) networks become more complex, due to the rapid changes, competition and technical innovation in market. Successful enterprises recognize the establishment of more and more complex flow networks of returns and goods to be recovered. Precisely, RL refers to the distribution activities involved in product returns, source reduction, conservation, recycling, substitution, reuse, disposal, refurbishment, repair and remanufacturing [1].

In other words, RL is the movement of the goods from a consumer towards a producer in a channel of distribution Murphy [2]. One can also consider RL definition of Revlog [3], “the process of planning, implementing and controlling flows of raw materials, in process inventory, and finished goods, from the point of use back to the point of recovery or point of proper disposal”. Thus, good management of reverse logistics can make the reintroduction of raw materials and recycled goods economically viable. This contributes to sustainable development of the supply chain. So, various frameworks Seitz [4] De Brito [5] have developed designing RL based on sustainable development criteria: economic,

environmental and social ones. More frequently, researchers focused on economic criteria to make a strategic decision as well as the design of RL network … Stock [1] Si-bo and Wei-Lai [6] Kim and al. [7]. The design of reverse logistic network requires an economical and environmental quantification. It concerns, for example, the choice of deposit locations, their capacities, quantity of returned product, etc. However, in real situations, data related to reverse logistics are often imprecise, hypothetical and inconsistent. Our framework focuses on choosing the best network reverse logistics under uncertainty. Based on development sustainable indicators, decision maker can choose the suitable decision system. The rest of the paper is organized as follows: In section 2, literature review presents the different reverse logistic networks. Section 3 describes our proposed methodology to choose the best RL configuration. Section 4 describes numerical example including test scenarios and numerical results

II. LITERTURE REVIEW Many previous researches on the problems of logistics network design that consider the facility locations and allocation models, have been widely discussed Si-bo and Wei-Lai [6] Kim and al. [7] Lu and al. [8]. These models allow industry actors to tack a variety of decisions as determining the number of these manufacturing centers to find a minimum cost of location-allocation. Based on two main centers of network, Lu and al. [8] developed a problem of multi-objectifs of reverse logistic have network to determine processing center. In their research, Zhou [9] have proposed Mixed Linear model in localization allocation problem in strategic situation that minimize the number of installation center of forward chain and to increase numbers of recovery. Most studies of the existing network models Pourmohammadi and al. [10], Meng, [11] and Mutha and Pokhal [12] have considered the economic viability to select the best reverse network. Pourmohammadi and al. [10] have developed a mixed integer linear model for recovery industrial waste of aluminum in the city of Los Angelos in the United states that minimize the cost to select the best reverse network, considering fixed availability of process area. Meng (2008) has focus on designing and configuration network to minimize the cost of the electronics waste using a mathematical formulation of probabilistic parameters. Mutha and Pokhal [12] has considered the best reverse network which can be

Page 2: [IEEE 2013 5th International Conference on Modeling, Simulation and Applied Optimization (ICMSAO 2013) - Hammamet (2013.4.28-2013.4.30)] 2013 5th International Conference on Modeling,

selected by minimizing cost of fixed ability of the process area using Linear Mathematical model the design of reverse logistics network. Cruz and Ertel [13] He and al. [14] Zhaohua and Jianhua [15] have proposed various application fields like vehicle and municipal solid waste to implement their quantitative models according to real contexts (end of life of vehicles in Mexico, waste of the aluminum in the city of Los Angeles in the United States, Recovery computer in the city of Taiwan in Shandong China, …) Cruz and Ertel [13] have proposed a prognostics location model of collections centers to 2025 of end of life vehicles applied in Mexico. The model has aimed to determine the minimum number of collection centers that Minimize the transport distance of vehicles with maximum coverage of saved areas have used the simulator approach. In their research, He and al. [14] have treated a model of locate problem in context recovery of municipal solid waste. Based on genetic algorithm fuzzy logic, the author has developed a model to determine a processing center. Zhaohua and Jianhua [15] have proposed a mixed integer linear model including four entities: Client, though reprocessing center, treatment centre and production facilities applied in city of Taiwan Shandong China of recovered computer. Used a method of Lagrangian relaxation aimed to minimize the total cost. Most study studies of an exciting network model consider the economic viability to select the best network. Zhaohua and Jianhua [15] have developed a mixed integer linear model for recovery industrial waste of aluminum in the city of Los Angelos in the United states that minimize the cost to select the best reverse network, considering fixed availability of process area. Meng [11] has focused on designing and configuration network to minimize the cost of the electronics waste using a mathematical formulation of probabilistic parameters. Mutha and Pokhal [12] have considered the best reverse network which can be selected by minimizing the fixed ability of cost in the process area using Linear Mathematical model.

III. REASONING PROPOSAL

The purpose of design reverse logistic network is to get the best configuration not only according to the economic criteria but also to other aspect of human life and the environment resources. Network structure design involves the integration of closed flows of returned products (defective, end of life, at the end of operation…). The real flow managing of used product needs to be convoy at the different physical location and facility. It is considered to allow the best configuration and make an effective decision according to sustainable indicators. The methodology of our proposed research is summarized in four steps in figure1.

Fig. 1. Reasoning Proposed

A. Step 1: Mapping of RL structure an flow

The literature review allows identifying the whole flows associated to product returns. This demonstrats that in general, retailers collect the first, returned products. Then, they ship the collected products to warehouses that are usually specialized (white goods, brown, car ...). Incidentally, these retailers can also sell to the secondhand market, products that are in good working order and with satisfactory market conditions. The warehouse, then, serves re processing centers. They can also make disposal certain products when they are in excess compared to market demand. The reprocessing center is the key factor in the RL. This is where the destinations of the whole end-of-life products are really directed. The reprocessing centers allow us sending products:

- to landfill, - to incineration or to the material recycling

processes(like plastic fractions), - as spare parts after is assembly operations, - to remanufacturing centers to be refurbished and re-

qualified in order to give new life or new uses to the returned products.

Note that remanufacturing centers need new spare parts and fractions from classical suppliers to remanufacture returned products. Figure 2 summarizes the organization flows according to our point of view. It corresponds especially to the representation of the RL presented in research works of Mutha and Pokhal [12]. We have also considered assumptions about flow patterns. In summary, we have to identify and quantify the RL actors in this first step of reasoning proposal. We have also determined flows of the whole products, modules, fractions and parts in the given RL networks. The final task is to express precisely, all the assumptions we consider to simplify or make a more realistic model of flow organization of a given RL.

Identifying the configurations of the RL structure

Indicator system design

Probablistic modeling of reverse logistics

Arena simulation providing alternative network design

Page 3: [IEEE 2013 5th International Conference on Modeling, Simulation and Applied Optimization (ICMSAO 2013) - Hammamet (2013.4.28-2013.4.30)] 2013 5th International Conference on Modeling,

Fig 2. Proposed structure of logistic networ

B. Step 2: Indicator system design In order to make a suitable decision in selecting the best

configuration, sustainable development is a key factor of social, environment and economic issues. Thus, the decision should take into account three-side considerations to evaluate the proposed configuration:

a) Economic consideration: Economic indicators aim to bring benefits to the manufacturing by saving cost of the waste management, resource reduction, improvement in customer/supplier relations, etc. RL perceived economic benefits relating to all the recovery options. Example of Economic indicators: inventory cost, re-processing cost, spare parts returns, transport cost, etc.

b) Environment consideration: Concerning green issues, the RL has gained benefits of environment by options for take-back and recovery of their used products. Manufactures should ensure the environmental respects because of trade regulation and the agreement of environment preservation. Example of environment indicators: rate energy consumption, Carbon footprint, valorization rate, reprocessing cost, etc

C) Social consideration: The measurement of social impacts and the calculation of indicators are less developed than economic indicator. In general, they are based on 4 points of view: internal human resource, external population, stake holder participation and macro social performance. Example of societal responsibility indicators: Cost of employee health and safety, Total revenues, Local suppliers rate…

In the present work, only inventory cost and reprocessing cost are taken into consideration. Concerning these costs, we adopt the economical model of Mutha and Pokhal [12].

1) Inventory cost We consider such cost at warehouses, reprocessing centers

and remanufacturing ones. The inventory cost of product p at whole of W warehouses at time t is given as:

(1)

: Quantity of product p at warehouses w at time t : Inventory carrying cost of product p at warehouses w

at time t

: Quantity of product p at RPC j at time t

: Inventory carrying cost of product p at RPC j at time t The inventory cost of modules n obtained from returned product p at whole at J RPC at time t is given as:

(2)

: Quantity of modules n obtained from returned

product p at RPC j at time t : Inventory carrying cost of module n from product p at

RPC j at time t : Quantity of modules n from product p stored at RPC j

at time t 2) Reprocessing cost

Cost of reprocessing concerns processing of modules of product p for both spare markets s and remanufacturing centers u.

: Quantity of modules n obtained from returned

product p sent to Spare Market s at time t : Quantity of modules n obtained from returned

product p sent to remanufacturing center u s at time t : Unit reprocessing cost for module n of product p at

RPC j at time t : Quantity of modules n from product p sent to Spare

Markets at time t : Quantity of modules n from product p sent to

remanufacturing centers at time t

: Quantity of modules n obtained from returned

product p sent to remanufacturing center u s at time t : Unit reprocessing cost for module n of product p at

RPC j at time t : Quantity of modules n from product p sent to Spare

Markets at time t : Quantity of modules n from product p sent to

remanufacturing centers at time t 3) Probabilistic modeling of RL

The data related to the element of managing reverse logistic operation are not completely known, due to the variability in delivery as well as quality and quantity of retuned product. Consequently, this variability is often imprecise and uncertain. Bayesian Networks provide high potential to represent ambiguous knowledge and for performing reasoning under uncertainty by probabilistic dependencies among the corresponding random variables. Bayesian Networks (BN) approach offers a directed acyclic graphs (DAGs) represented

Page 4: [IEEE 2013 5th International Conference on Modeling, Simulation and Applied Optimization (ICMSAO 2013) - Hammamet (2013.4.28-2013.4.30)] 2013 5th International Conference on Modeling,

by nodes (vertices), which indicates the existence of statistical dependency between them, and the arcs (edges) given parents pa (xi) child relation in the graph. Due to the characteristic of inter-causal reasoning, BN provide to propagate information in a whole network. A Bayesian network model depends on set of variable V={X1,X2,…,XN}. It is providing conditional distribution describing dependency between nodes p (X1/pa (Xi)). Generally, BN modeling can be summarized as follows: a) Identification of the aim of the model the first result in terms

of the relationships between the different variables and their strength

b) Data preparation and refinement. It consists in preparing data based on general BN which have been initially implemented for discrete random variables

c) Model learning: it builds the model from available information. It corresponds to qualitative component (graph structure) estimating the parameters

d) Validation: the validation methods check the representatively of the model and the accuracy of the inference results.

e) Software is proposed to compute the distributions if the problem is medium.

IV. ILLUSTRATIVE EXAMPLE: SIMULATION AND IMPLEMENTATION

Simulation package can provide alterative networks configuration. Considering different configurations generate from various scenarios, users test a different configurations reverse logistic to allow maker to take a better decision as well. The simulation models assist the decision maker in taking a better decision by considering different changes in the structure of the reverse network, due to the complexity of the reverse network model. The simulation is known as an effective approach for adopting process, implementing uncertain data and evaluating different proposed changes. It allows the possibility to respond for any number of “what-if” scenarios. Considering assessing several structure of reverse logistic, simulation model aims to detect the critical points related to the economic, social and environment viability.

C3

C2

W1

W2

W3

RCP1

RCP2

RC DC

C1

Fig 3 . Bayesian networks for facility location

We suggest an example of Electric and Electronic Waste in reverse logistic networks according to proposal RL structure shown in Figure 2. The proposed model includes five actors: customer (C), Warehouses (W) as collecting used products from reprocessing centers (RCP) and recycling centre (RC). The re-processing centers (RC) and recycling center have an unlimited capacity are assumed unlimited. An appropriate probabilistic graphical model describes the structure of the reverse logistics network. This graphical model corresponds to

the Bayesian network. The set of nodes Xi represent customers (C), Warehouses (W), Re-processing Center (RCP), Recycling Center (RC) and Distribution center (DC). The direct edges represent a statistical dependence between different nodes. The Bayesian network structure gives a conditional probability attached to the parents’ nodes p (Xi/Pa (Xi)). The joint distribution of the different nodes of the network is located in conditional probability tables (CPT) (Table 2). These CPT include the volumes of used product calculated in different point in reverse logistic

TABLE I. CONDITIONAL PROBABILITY TABLES

After determining the probability based on the distribution provided from conditional probability of Bayesian network, we propose Arena/simulation package, which allows capturing the probabilistic quantities of the used products generated by Bayesian model. The aim of this paper is to investigate the most appropriate factors to select the best reverse logistic configuration. Through using arena 9.0 simulation packages, four scenarios of configuration have been developed. When designing a network structure of RL, the number of locations in the network is considered a main factor. This scenario includes 3 warehouses (W), 2 reprocessing centers (RCP) in the first configuration. The second scenario, the configuration is based on, 2 (W) and 2 (RCP). The third one corresponds to 3 (W) and 1 (RCP). The last configuration network is based on 2 (W) and 1 (RCP). The number of the other center has not been changed in different scenarios. We identify the duration or the simulation time at which the simulation should stop. Thus, we specify a transition period of about 4800 minutes and a simulation period of 9600minutes for a number of replication equal to 20. Based on sustainable indicators that are: inventory

C1 C2 C3 W2 Probability 0 0 0 0 0,02 1 0 0 0 0,07 0 1 0 0 0,02 1 1 0 0 0,07 0 1 1 0 0,02 1 1 1 0 0,07 0 1 0 1 0,02 0 1 1 1 0,07 1 1 1 1 0,02

C1 C2 C3 W1 Probability 0 0 0 0 0,02 1 0 0 0 0,07 0 1 0 0 0,02 1 1 0 0 0,07 0 1 1 0 0,02 1 1 1 0 0,07 0 1 0 1 0,02 0 1 1 1 0,07 1 1 1 1 0,02

C1 C2 C3 W3 Probability 0 0 0 0 0,02 1 0 0 0 0,07 0 1 0 0 0,02 1 1 0 0 0,07 0 1 1 0 0,02 1 1 1 0 0,07 0 1 0 1 0,02 0 1 1 1 0,07 1 1 1 1 0,02

W1 W2 RCP Probability 0 0 0 0,11 1 0 0 0,11 0 1 0 0,11 1 1 0 0,11 0 1 1 0,11 1 1 1 0,11 0 1 0 0,11 0 1 1 0,11 1 1 1 0,11

W1 RCP2 Probability 0 1 0,375 1 0 0,125

0 0 0,125 1 1 0,375

Page 5: [IEEE 2013 5th International Conference on Modeling, Simulation and Applied Optimization (ICMSAO 2013) - Hammamet (2013.4.28-2013.4.30)] 2013 5th International Conference on Modeling,

cost, re-processing cost, rate of energy consumption and number of hours work; Table 3 reports the simulation result of the four alternatives configurations. According to the simulation results, the inventory cost is calculated using the simulation model. The inventory cost will be approximately1,5$/unit in the warehouse, 1$/unit in the re-processing center and 0.5$/unit in the recycling center. Values of inventory costs lie within 98.5 and 137.08.The lowest value of inventory cost is derived by locating 3 warehouses and 2 reprocessing centers. The highest value is obtained by the configuration 4, which is composed of 2 Warehouses and 1 Re-processing center. Inventory cost is increased when the number of storage location is reduced. Similarly, in the scenarios 2 which is composed of 2 warehouses and 2 re-processing centers, the inventory cost is increased approximately the double. According to the re-processing cost, the configuration 3 records the lowest value (11,795$). It is anticipated that the number of reprocessing-center has a relationship with the number of warehouses. The scenario 3 is affected by energy of consumption that is reduced to 74%. Due to the limited number of re-processing centers the energy of consumption is reduced. Compared with to the result of the forth scenario, a highest value is obtained (92%), the consumption of energy is not de-pended on the number of location in the reverse network. The result of the simulation models concerning the number of hours of work is insignificant. This is can be dedicated to the economic factors that have the biggest impact on the choice of the suitable configuration.

TABLE II. SIMULATION RESULTS

V CONCLUSION AND FUTURE RESEARCH Simulation model is developed in this study to allow users to test different alternatives by changing the number of location centers. This paper proposes a model to choose the best configuration of reverse network. Considering incompleteness

of data, we suggest using a Bayesian network to model the ambiguity according to the volume of returned production in our case. We evaluate alternative configuration under uncertainty considering sustainable indicators that are: economic, environment and social viability. The model provides different structures of reverse network. Simulation/Arena package provide the user to analyze factors which are important. However, simulation model does not provide significant result concerning number of hours. Our future research may consider the development model of multiple criteria decision making model based on Bayesian network.

REFERENCES [1] J. K. Stock, “Reverse logistics, white paper, council of logistics

management”. IL: Oak Brook. 1992. [2] P. Murphy, “A preliminary study of transportation and warehousing

aspects of reverse distribution” Trans-portation Journal, 34(1), p.48–56, 1986 [3] REVLOG,www.fbk.eur.nl/OZ/REVLOG/PROJECTS/TERMINOLOG

Y/Definitions.html [4] M Seitz,., “A critical assessment of motives for product recovery: the

case of engine remanufacturing”, Journal of Cleaner Production 15 (11e12), 1147-1157. 2006.

[5] M. P. De Brito, Managing Reverse Logistics or Reversing Logistics Management? Erasmus University Rot-terdam, PhD thesis,2004.

[6] D. Si-bo, and H. Wei-Lai, “Optimal design of multi-echelon reverse logistics using genetic algorithm”, IEEE Forth International Conference on Wireless Communication, Networking and Mobil Computing, 12-14 Oct (2008)

[7] K. Kim, I. Song, and J. Kim, and B. Jeong, “Supply planning model for remanufacturing system in reverse logistic environment”, Computers and IndustrialEngineering,51,pp.279-287,2008

[8] Y. Lu, P. Lu, and L. Liang, “Multi-objective Optimisation of Reverse Logistics Network Based on Random Weight and Genetic Algorithm”, IEEE International Conference on Networking Sensing and Control, pp.1196-1200,6-8April,2008

[9] Y. Zhou, and S. Wang, “Generic model of reverse logistics network design”, Journal of Transportation Systems En-gineering and Information Technology, 8(3), p.71-78, 2008.

[10] H. Pourmohammadi, M. Rahimi, and M. Dessouky, “Sustainable reverse logistics for distribution of industrial waste by products: Ajoint optimization of operation and environmental cost”, Supply Chain Forum, 9 (1). 2008

[11] Meng, “Network design on reverse Logistic of Electronic Wastes Recycling”, IEEE International Conference on Automation and Logistics, 1-3 September 2008

[12] A. Mutha, and S. Pokharel, “Strategic design for reverse logistics and remanufacturing using new and old product modules”, Computers and Industrial Engineering, 2008

[13] R. Cruz, R. and J.Ertel,” Reverse Logistics Network Design for the Collection of End-of-Life Vehicles in Mexico”, European Journal of OperationalResearch,196,pp.930-9392009

[14] B., He, C. Yang and M Ren. “A Fuzzy Multi-objective Programming for Optimization of Reverse Logistics for Solid Waste through Genetic Algorithms”, IEEE Forth International Conference on Fuzzy System and KnowledgeDis-covery,2007

[15] W. Zhaohua, and Y. Jianhua, “Modeling for Facility Location Optimization of Spent Computer’s Reverse Logis-tics”, IEEE Forth International Conference on Wireless Communication, Networking and Mobil Computing, 12-14 Octobre, 2008

Inventory cost

Reprocessing cost

Rate of energy

Hours of work

Configuration1 98,5 13,05 84% 160 Configuration2 137,08 13,254 74% 160 Configuration3 117,12 11,795 74% 160 Configuration4 136,98 13,795 92% 160