DESIGNING A CLOSED-LOOP SUPPLY CHAIN FOR ALUMINUM ENGINE MANUFACTURING

DESIGNING A CLOSED-LOOP SUPPLY CHAIN FOR ALUMINUM ENGINE MANUFACTURING R.S. Lashkari and Yi Duan Department of Industrial and Manufacturing Systems ering University of Windsor 40 Sunset Ave, Windsor, ON N9B 3P4, Canada Email: lash@uwindsor.ca, duan4@uwindsor.ca ABSTRACT The closed-loop supply chain of an aluminum engine manufacturing operation is investigated. The push for the recycling of aluminum components in auto industry is motivated by profit incentives (i.e., the recovery of a valuable resource) as well as by legal obligations (i.e., to comply with the requirements of the Extended Product Responsibility legislation), which have, therefore, given rise to the development of closed-loop supply chains. The paper presents a planning model for the closed-loop process that includes purchasing, production, and end-oflife product collection and recycling/ remanufacturing in the context of an aluminum engine manufacturing and recycling operation. The model is a multi-echelon general integer linear program with the objective of minimizing the total costs in the network subject to structural and functional constraints. The model may be employed to make decisions regarding raw material procurement, production, recycling and inventory levels, and the transportation activities in the network. KEY WORDS Closed-Loop Supply Chains, Aluminum Recycling, Integer Programming. Introduction In the last two decades, the concept of Supply Chain Management has emerged as a manufacturing paradigm for improving the competitiveness of an enterprise. In order to improve responsiveness and flexibility of manufacturing organizations, the SCM is considered as a competitive strategy for integrating suppliers and customers []. Supply chain management is a set of approaches utilized to efficiently integrate suppliers, manufacturers, warehouses, and retailers, so that merchandise is produced and distributed at the right quantities, to the right locations, and at the right time, in order to minimize system-wide costs while satisfying service level requirement [2]. A supply chain is a network of facilities and distribution options that performs the function of procurement of materials, transformation of these materials into intermediate and finished products and the distribution of these products to customers [3]. A supply chain not only includes the manufacturer and suppliers, but also transporters, warehouses, retailers, and customers themselves. It consists of all the parties directly or indirectly involved in fulfilling the customer demand [4]. In recent years, the reuse of products and materials has gained renewed importance due to legal obligations (i.e., take-back legislation) and financial reasons (i.e., incentives to recover value in returned products), creating a new material flow from the users back to the suppliers, and giving rise to the concept of reverse logistics or reverse supply chain, which encompasses the logistics activities all the way from end-of-life products to remanufactured products again usable in the market [5]. In this paper we propose a closed-loop supply chain model of aluminum engine production, recycling and remanufacturing. The proposed model is based on the operations of a major automobile manufacturer in the U.S. (hereinafter referred to as the Company), which manufactures and distributes automobiles in 200 markets. The research is focused on the operations of its 5 engine plants in the U.S. The paper is organized as follows. Section 2 provides a brief description of the system constituting the elements of the supply chain under consideration. In section 3 the relevant literature is reviewed. Section 4 outlines the structure of the mathematical model; section 5 presents some numerical results; and section 6 concludes with some observations and comments.

2. System Description 2. The life cycle of aluminum engines The aluminum engine manufacturing, recycling, and remanufacturing under consideration in this paper is depicted in Figure. Other Aluminum Parts, including Cylinder Heads, Cylinder Blocks The engine plant produces engines depending on customer demand. It purchases other aluminum engine parts from outside suppliers. The engine plant maintains safety stocks for both the components and the assembled engines. The safety stock is normally equal to a two-day production level. The plant also supplies new engines to dealerships for replacing warranty returns, as need arises. The assembled engines are delivered to auto assembly plants to be installed in vehicles. After assembly, vehicles are sent to dealerships, where customers purchase their vehicles. Dealerships send warranty replacements from customers to collection centers, and they order new engines from the engine plants to replace the warranty returns. Aluminum Casting Plants Recycling Centers Other products containing aluminum Plants Collection Centers-- Decision point Auto Assembly Plants Dealerships /Garages Customers The end-of-life vehicles are returned to collection centers where an elaborate electronic system is used to help in deciding what to dismantle. If the engine can be rebuilt then the collection center takes the engine out and sends it to a remanufacturing center. If not suitable for remanufacturing, the engine will be left in the vehicle which will be flattened. Flattened hulks are shipped to shredders which pulverize them into fist-sized pieces in minutes. Valuable ferrous and non-ferrous metals are removed magnetically and using complex floatation systems, and the shredder fluff is sent to landfills. Remanufacturing Centers 2.2 The System Elements Cylinder blocks Warranty Returns New s Finished Vehicles New Vehicles End-of-life Vehicles The closed-loop model includes: - aluminum casting plant - 5 engine plants - 3 auto assembly plants - 3 collection centers - 2 recycling centers - 2 remanufacturing centers Flattened Hulks Aluminum Parts Recycled Aluminum Rebuilt s Rebuildable s Warranty Replacements Figure : The Closed-Loop Supply Chain for Aluminum Production and Recycling The aluminum casting plant purchases aluminum ingots and recycled aluminum as raw material to be used in making new aluminum cylinder blocks and cylinder heads, which are then delivered to the engine plants according to a schedule. Table presents the system under consideration. The engines are divided into three product families, aluminum-aluminum engines (i.e., aluminum cylinder head and aluminum cylinder block), aluminum-iron engines (i.e., aluminum cylinder head and iron cylinder block) and iron-iron engines (i.e., iron cylinder head and iron cylinder block). In this work, we only consider the first two families. There are 5 engine plants which produce 7 engine types within the 2 product families mentioned above. The engines are then sent to 3 auto assembly plants. For example, engine plant produces engine types and 2 within product family. type is sent to auto assembly plant 2, and engine type 2 is sent to auto assembly plant 8. Similarly, engine plant 5 produces engine types 6 and 7 which are sent,

respectively, to auto assembly plants 3,4,5,6,7, 3, and auto assembly plant. In addition to the one aluminum casting plant which is part of the current supply chain, there are two other casting plants which are part of a joint venture, operating outside the U.S. One plant sends its products to some of the 5 engine plants in the U.S. The aluminum casting plant sends engine parts for engine type 3 to engine plant plant, m Table : Products Produced at Plants Product family, q type, n. 2.0L I4 Auto assembly plants, a 2 2. 2.3L I4 8 s with 3. 2.5L V6 0 2 Aluminum Cylinder 4. 3.0L V6, 4, 9, 0 3 4 Head and Aluminum Cylinder Block 4. 3.0L V6 5. 3.5 L V6 2 5 6. 4.6L V8 5 2 s with Aluminum Cylinder Head and Cast Iron Cylinder Block 6. 4.6L V8 7. 5.4L V8 3, 4, 5, 6, 7, 3 2, for engine type 4 to engine plants 2 and 3, and for engine type 6 to engine plant 5. The rest of the aluminum engine parts are purchased by the Company from outside suppliers. These suppliers do not belong to the Company, and therefore are not included in our study. Auto assembly plants 9, 0, and and the aluminum casting plant are all outside the U.S.; however, they are considered a part of the chain in so far as they operate under the financial constraints set by the Company. Although there are many collection, recycling and remanufacturing centers within the U.S., we only use 3 collection centers, 2 recycling centers and 2 remanufacturing centers to represent the entire group. While they do not belong to the Company, they are nevertheless a part of the closed-loop supply chain that is organized to support the end-of-life vehicle returns, and therefore they have a financial incentive to minimize their cost of providing the raw material to the Company; that is, the costs in the reverse channel also have an effect on the forward channel. 3. Literature Review In this section we present a brief review of the relevant literature as it pertains to reverse logistics and closed-loop supply chains. 3. Reverse logistics Reverse distribution is the collection and transportation of used products, which may occur through the original forward channel, through a separate reverse channel, or through combinations of the forward and the reverse channels. Krumwiede and Sheu [6] investigated current industry practices in reverse logistics business, and employed a decision-making model to guide the process of examining the feasibility of implementing reverse logistics in thirdparty providers such as transportation companies. Jayaraman et al. [5] discussed the reverse distribution network of an electronic equipment remanufacturing company in the US and developed a decision model to minimize reverse distribution costs. Nagurney and Toyasaki [7] developed a multi-tiered e- cycling network model consisting of four tiers of nodes (sources of the electronic waste, recyclers of the electronic waste, processors of the electronic waste and demand market) for the reverse supply chain of an electronic waste recycling operation. Schultmann et al. [8] modeled the reverse logistics aspects of the end-of-life vehicle (ELV) treatment in Germany using vehicle routing planning. Lieckens and Vandaele [9] extended traditional facility location-allocation models (formulated as mixed integer linear programs and determining which facilities to open that minimize the cost while supply, demand and capacity constraints are satisfied) by introducing queuing relationships into the network in order to incorporate a product s cycle time and inventory holding costs, in addition to dealing with the higher degree of uncertainty and congestion, typical characteristics of such networks.

In the area of remanufacturing, Guide and Srivastava [0] discussed the scheduling policies for remanufacturing shops based on the information from turbine jet engine remanufacturing. In the context of a remanufacturing environment, they examined the location of inventory buffers and their impact on other managerial operating decisions. Mahadevan, et al. [] investigated an inventory system with manufacturing and remanufacturing using a push policy. They examined the operation of the system as a function of return rates, backorder costs, and manufacturing and remanufacturing lead times. Smith and Keoleian [2] developed a life-cycle assessment model (LCA) to investigate the savings in energy use and raw material consumption, and pollution prevention that are achieved through remanufacturing an engine compared to an OEM manufacturing a new one. 3.2 Closed-loop supply chain Spengler [3] presented the design and implementation of a decision support system for electronic scrap recycling companies in Germany. Beamon and Fernandes [4] presented a closed-loop supply chain model of a manufacturing operation that produces new and remanufactured products. The model consists of four echelons including manufacturers, warehouses, customer zones and collection centers. Sheu, et al. [5] proposed a linear multi-objective programming model to deal with integrated logistics operational problems of Green-supply chain management (G-SCM). The model was developed to optimize the operations of both the integrated logistics and the corresponding used-product reverse logistics in a given chain of five layers based on a real world case study for a computer manufacturer. Zhang and Lashkari [6] developed a closed-loop supply chain model of a lead-acid battery manufacturing and recycling process. The model is used to make decisions regarding raw material procurement, production, recycling and inventory levels, and the transportation modes between the echelons. Vlachos, et al. [7] presented a system dynamics model for dynamic capacity planning of remanufacturing in closed-loop supply chains dealing with a single product, where the forward chain has two echelons: producer and distributor, and the reuse activity is remanufacturing. Ko and Evans [8] presented a mixed integer nonlinear programming model for dynamic supply chain management by third party logistics providers (3PLs). The objective is to minimize the total costs incurred in the forward and reverses flows. Listeş [9] presented a generic two-stage (plant and market) single-period, stochastic programming model for the design of closed loop networks and used a decomposition-based approach to solve the problem. The model explained a number of alternative scenarios which may be constructed based on critical levels of design parameters such as demand and returns. Lu and Bostel [20] develop an algorithm to solve the twolevel location model with three types of facility which are producers, remanufacturing centers and intermediate to be sited, considering forward and reverse flows that cover remanufacturing activities. The objective is to minimize the costs of setting up facilities, shipping and receiving products. 4. The Model Structure In view of the above literature survey, we propose to develop an integrated multi-product, multi-stage, multiperiod general integer linear programming model in this study. The model will consider both the forward flow and the reverse flow in the supply chain, concentrating on the end-of-life material recovery. The model considers the following members in the chain: ) Suppliers of raw materials 2) Aluminum casting plants 3) manufacturing plants 4) Automobile assembly plants 5) Dealerships/customer centers 6) Remanufacturing centers 7) Recycling centers 4. Mathematical formulation The objective of the model is to minimize the total costs related to the production, transportation, and the assembly of aluminum engines, subject to a number of constraints. The model is too large to be presented here; however, we provide the following general information about the structure of the model. The objective function consists of the following cost components: - Purchasing cost of raw material - Transportation costs - Inventory costs - In-transit inventory costs - Labour costs - Handling costs

- Minus any revenues realized at the engine plants from the sale of engine parts to the remanufacturing centers The following classes of constraints are in effect:. Production capacity at the aluminum casting plant, engine plants, auto assembly plants, collection centers, recycling centers and remanufacturing centers 2. Storage capacity at the aluminum casting plant, engine plants, auto assembly plants and remanufacturing centers 3. Production labor hour at the aluminum casting plant, engine plants and remanufacturing centers 4. Transport carriers capacity in terms of weight and volume 5. Inventory capacity a. Inventory balance at the aluminum casting plant, engine plants, auto assembly plants, and remanufacturing centers b. In-transit inventory balance: - from the aluminum casting plant to engine plants; - from engine plants to auto assembly plants; - from auto assembly plants to dealerships; - from collection centers to recycling centers c. Safety stock at the aluminum casting plant, engine plants, auto assembly plants and remanufacturing centers In the formulation of the model the following assumptions are considered:. All in-transit inventory transportation costs are accounted for at the source. The in-transit shipments from engine plants to auto assembly plants, for example, are charged to engine plants they originated from. The intransit inventory in three of these stages from engine plants to remanufacturing centers, from collection centers to remanufacturing centers and from remanufacturing centers to dealerships is small enough in number to be insignificant to our study. 2. All transportation costs are also accounted for at the source. The transportation costs from engine plants to auto assembly plants, for example, are charged to engine plants. 3. The number of returned end-of-life vehicles is calculated based on the number of vehicles that are retired every year multiplied by the Company s market share. 4. The rate of material loss during the manufacturing operations at the aluminum casting plant is about 0%. 5. The remanufacturing centers do not keep inventories of engine parts for rebuilding engines. 6. No inventories are held at the collection center, since these centers act as decision points in the chain. 7. At the engine plants, the labor requirements are met through regular-time, overtime, additional labor hiring and layoffs. At the aluminum casting plant and at the remanufacturing centers the labor requirements are met through regular time and overtime. A worker operates 8 hours a day, 5 days a week. 8. The labor costs accounted for in the model are related to engine production, which includes the labor costs at the aluminum casting plants, engine plants, and the remanufacturing centers. The relevant labor costs at the auto assembly plants and the collection centers are included in the handling costs. 9. New engines sent from the engine plants to dealerships to replace warranty returns, engine warranty returns sent from customers to dealerships, and engine warranty replacements sent from dealerships to collection centers are small enough in numbers to be insignificant in our study. Their values are therefore assumed to be equal to zero. 5. Some Numerical Results The proposed model is solved using Lingo 9.0 [2]. Given the size and the complexity of the problem, it is necessary to use a relaxed version of the model that considers the large number of general integer variables as continuous variables. However, the results of the relaxed model are still a fairly good approximation of the integer solution. The model has 925 variables and 804 constraints, and the typical solution time is between 5 to seconds. The model generates a great deal of information regarding the production levels, inventories, safety stocks, and the transportation activities at various member locations in the chain. The presentation of the Table 2 shows the detailed results with respect to the operational costs of the chain over a period of a month. The two largest cost components are the handling costs and the labor costs, accounting for about 72% and 8% of the total cost, respectively. The next two largest cost components are the purchasing cost of raw material (at about 3.4% of the total), and the transportation costs (at about 3.%). Table 3 shows the amount of aluminum ingots purchased from suppliers, and the amount of recycled aluminum purchased from recycling centers, by the aluminum casting plant during a month. The Company maintains a ratio of 85% recycled aluminum to 5% virgin raw material.

Table 3: Amount of Raw Material Purchased Decision variables S T Input Amount, in lbs P XC st aluminum ingots 2,06,966 P2 s = t = S T s= t = XC st recycled aluminum,939,474 Table 4 shows the number of engines produced XM nqmt and the average daily inventory of engines XIV5 nqmt held at each engine plant. For example, the number of engine s of type n= in product family q= produced at engine plant m= is,220 per month and the average inventory of engines is,22 units per day. Table 4: Production and Average Daily Inventory of s at Plants Product Production Inventory family, T type, plant, XIV5 q n m XM nqmt nqmt (units/day) t =,220,22 2 6,500 650 3 2 3,780 378 4 2 42,600 4,260 4 3 4,540,454 5 4,240,24 6 5 2,540 254 6 2 5 49,300 4,930 7 2 5 2,540 254 Table 2: Cost Components (dollars/month) at Optimal Solution TOTAL COST 368,87,947 CAPITAL COST(INVENTORY COSTS AND PURCHASING COSTS) 20,757,297 EXPENDITURES(LABOR, HANDLING, TRANSPORTATION, AND IN-TRANSIT COSTS) 350,608,960 Distribution of the Total Costs PURCHASING COSTS aluminum ingots purchased 2,423,07 recycled aluminum purchased 0,48,553 TOTAL PURCHASING COSTS 2,57,570 TRANSPORTATION COSTS from the aluminum casting plant to engine plants 96,956 from engine plants to auto assembly plants,94,420 from auto assembly plants to dealerships 7,809,845 from collection centers to recycling centers 903,440 from engine plants to remanufacturing centers 6,39 from collection centers to remanufacturing centers 26,283 from remanufacturing centers to dealerships 556,204 TOTAL TRANSPORTATION COSTS,33,467 INVENTORY COSTS aluminum ingots held at the aluminum casting plant 2,67 recycled aluminum held at the aluminum casting plant 0,960 engine parts held at the aluminum casting plant 90,249 engine parts held at engine plants 90,768 assembled engines held at engine plants 839,00 assembled engines held at auto assembly plants 854,654 vehicles held at auto assembly plants 6,292,573 rebuildable engines held at remanufacturing centers 0 rebuilt engines held at remanufacturing centers 4,896 TOTAL INVENTORY COSTS 8,85,727 IN-TRANSIT INVENTORY COSTS from the aluminum casting plant to engine plants 37,66 from engine plants to auto assembly plants 23,069 from auto assembly plants to dealerships 6,099,07 from collection centers to recycling centers 64,977 TOTAL IN-TRANSIT INVENTORY COSTS 6,432,733

LABOR COSTS regular time labor cost at the aluminum casting plant 27,745 overtime labor cost at the aluminum casting plant 0 regular time labor cost at engine plants 43,424,60 overtime labor cost at engine plants 3,796,25 hiring cost at engine plants 223,680 layoff cost at engine plants 0 regular time labor cost at remanufacturing centers 9,245,570 overtime labor cost at remanufacturing centers 0 TOTAL LABOR COSTS 66,87,820 HANDLING COSTS at auto assembly plants 204,59,900 at collection centers 6,885,040 TOTAL HANDLING COSTS 266,044,940 TOTAL REVENUE 3,78,30 Table 5 shows the number of vehicles produced XUS3 nqat and the average daily inventory of vehicles XIV7 nqat held at each auto assembly plant. For example, the number of vehicles with engine type n=4 in product family q= produced at auto assembly plant a= is 3,924 per month and the average inventory is 392 units per day. Table 6 shows the number of flattened hulks sent from collection centers (decision points) to recycling centers (XPS4 crkt ), and the number of rebuildable engines sent from collection centers to remanufacturing centers (XPS6 cukt ). Table 5: Production of Vehicles with Different s and Average Daily Inventory of vehicles at Auto Assembly Pants type, n Product family, q Auto assembly plant, a Production T t = XUS3 nqat Inventory XIV7 nqat (units/day) 4 3,924 392 6 2,540 254 7 2 2,540 254 4 2 5,267,454 6 2 3 4,392 488 4 4 0,22 964 6 2 4 2,730 260 6 2 5 5,45 490 6 2 6 2,220 2,22 6 2 7,994,68 2 8 6,500 650 4 9,25,246 3 0 3,780 378 4 0 6,580,658 5,240,24 2,220,22 6 2 3 3,89 402 Table 6: Number of Flattened Hulks and Rebuildable s Sent from Collection Centers Number of flattened hulks C R K T XPS4 c= r= k = t= (units/month) 6. Conclusion crkt Number of rebuildable engines C U K T XPS6 c= u= k = t= (units/month) 99,80 35,800 In recent years, the proliferation of the legal obligations to recycle used products as well as profit incentives aimed at recovering value from returned products have created a need for the efficient design of closed-loop supply chains. In the field of auto industry, establishing a manufacturerwide, closed-loop supply chain would support the treatment of future end-of-life vehicles. This paper has proposed a multi-stage, multi-period, multi-product model of a closed loop supply chain that includes purchasing, production, and end-of-life products recycling and remanufacturing. The mixed integer programming model can be used as an effective business decision making tool for any auto manufacturers, and is easily adapted to other scenarios by adding or removing cukt

any relevant constraints. The model was verified in the context of large auto manufacturer using data extracted from industry sources. From a purely computational point of view, the proposed model solves for 925 variables and 804 constraints. The typical solution time using the LINGO solver is between 5 to seconds. The main contribution of the paper is the development of a model for planning, optimization and integration of the forward and reverse distribution networks in the context of an aluminum engine manufacturing and recycling. The model encompasses the engine manufacturing process including the purchasing, production, and end-of-life product recycling and remanufacturing. Acknowledgements This study was partially funded by the Natural Sciences and ering Research Council, Canada.. References [] A. Gunasekaran, Supply chain management: theory and applications, European Journal of Operational Research, 59(2), 2004, 265-268. [2] D. Simchi-Levi, P. Kaminsky, & E. Simchi-Levi, Designing and Managing the Supply Chain: Concepts, Strategies, and Case Studies (New York, NY: McGraw- Hill, 2000). [3] R. Ganeshan, & T.P. Harrison, An introduction to supply chain management, available at: http://lcm.csa.iisc.ernet.in/scm/supply_chain_intro.html. 995. [4] S. Chopra, & P. Meindl, Supply Chain Management: Strategy, Planning, and Operation (Englewood Cliffs, NJ: Pearson Prentice Hall, 2003). [5] V. Jayaraman, R. Patterson, & E. Rolland, The design of reverse distribution networks: models and solution procedures, European Journal of Operational Research, 50(), 2003, 28-49. [6] D. Krumwiede, & C. Sheu, A model for reverse logistics entry by third-party providers. The International Journal of Management Science (Omega), 30(5), 2002, 325 333. [7] A. Nagurney, & F. Toyasaki, Reverse supply chain management and electronic waste recycling: a multitiered network equilibrium framework for e-cycling. Transportation Research, Part E, 4(), 2005, -28. [8] F. Schultmann, M. Zumkeller, & O. Rentz, Modeling reverse logistic tasks within closed-loop supply chains: an example from the automotive industry, European Journal of Operational Research, 7(3), 2006, 033-050. [9] K. Lieckens, & N. Vandaele, Reverse logistics network design with stochastic lead times, Computer & Operations Research, 34(2), 2007, 395-46. [0] V.D.R. Guide Jr., & R. Srivastava, Inventory buffers in recoverable manufacturing, Journal of Operations Management, 6(5), 998, 55-568. [] B. Mahadevan, D.F. Pyke, & M. Fleischmann, Periodic review, push inventory policies for remanufacturing, European Journal of Operational Research, 5(3), 2003, 536-55. [2] V.M. Smith, & G.A. Keoleian, The value of remanufactured engines: life-cycle environmental and economic perspectives, Journal of Industrial Ecology, 8(-2), 2004, 93-22. [3] T. Spengler, Management of material flows in closed-loop supply chains: decision support system for electronic scrap recycling companies, Proc. 36th International Conference on System Sciences, Waikoloa, Hawaii, 2003. [4] B.M. Beamon, & C. Fernandes, Supply-chain network configuration for product recovery, Production Planning & Control, 5(3), 2004, 270-28. [5] J. Sheu, Y. Chou, & C. Hu, An integrated logistics operational model for green-supply chain management, Transportation Research, Part E, 4(4), 2005, 287-33. [6] H. Zhang, & R.S. Lashkari, 2005, Design and analysis of a closed-loop supply chain network for leadacid battery recycling, Proc. International Conference on Operations and Supply Chain Management, Bali, Indonesia, 2005, L-L8. [7] D. Vlachos, P. Georgiadis, & E. Iakovou, 2007, A system dynamics model for dynamic capacity planning of remanufacturing in closed-loop supply chains, Computers & Operations Research. 34(2), 2007, 367-394. [8] H.J. Ko, & G.W. Evans, A genetic algorithm-based heuristic for the dynamic integrated forward/reverse logistics network for 3PLs, Computers & Operations Research. 34(2), 2007, 346-366. [9] O. Listeş, 2007, A generic stochastic model for supply-and-return network design, Computers & Operations Research, 34(2), 2007, 47-442. [20] Z. Lu, & N. Bostel, 2007, A facility location model for logistics systems including reverse flows: The case of remanufacturing activities, Computers & Operations Research, 34(2), 2007, 299-323. [2] LINDO Systems, Inc., Lingo User s Guide. (Chicago, IL: LINDO Systems, 2004).