(REFEREED RESEARCH)

SUPPLY CHAIN IN APPAREL INDUSTRY: A TRANSSHIPMENT PROBLEM WITH TIME CONSTRAINTS

HAZIR GİYİM ENDÜSTRİSİNDE TEDARİK ZİNCİRİ YÖNETİMİ: ZAMAN KISITLI BİR AKTARIM PROBLEMİ

Deniz Türsel ELIIYI Emine Zehra YURTKULU Dicle YURDAKUL ŞAHİN Izmir University of Izmir University of Economics Izmir University of Economics Department of Industrial Systems Engineering Department of Logistics Management Department of e-mail: [email protected] Administration

ABSTRACT is one of the main sources of competitive advantage for companies. As a useful tool for inventory and transportation management in supply chains, transshipment points provide an effective mechanism for correcting discrepancies between demand and available inventory. This study aims to model the transshipment problem of a company in the apparel industry with multiple subcontractors and customers, and a transshipment depot in between. Unlike a typical transshipment problem that considers only the total cost of transportation, our model also considers the supplier lead times and the customer due dates in the system and it can be used for both supplier selection and timely distribution planning. The proposed model can also be adapted easily by other companies in the industry. Keywords: Transshipment problem, Lead time constraints, Apparel industry.

ÖZET Tedarik zinciri yönetimi şirketler için rekabet avantajı sağlamadaki temel kaynaklardan birini oluşturmaktadır. Tedarik zincirlerinde envanter ve ulaştırma yönetimi konusunda faydalı kullanımları olan aktarım noktaları özellikle talep ve envanter miktarı arasındaki farkı kapatma konusunda etkin bir mekanizma sunmaktadır. Bu çalışmanın amacı hazır giyim sektöründe birden çok fason imalatçı ve müşteri ile faaliyet gösteren ve aktarım deposuna sahip bir şirketin aktarım probleminin modellenmesidir. Yalnızca toplam maliyeti minimize etmeyi amaçlayan tipik aktarım modellerinden farklı olarak modelimiz tedarik sürelerini ve müşterilerin termin zamanlarını da dikkate almakta, bu sayede hem tedarikçi seçiminde hem de zamanında dağıtım planlamasında kullanılabilmektedir. Önerilen model sektördeki diğer firmalar için de kolaylıkla adapte edilebilir. Anahtar Kelimeler: Aktarım problemi, Tedarik süresi kısıtları, Hazır giyim sektörü.

Received: 12.07.2010 Accepted: 31.03.2011

1. INTRODUCTION terms of both the customers and the reduce its costs and improve the level on the other (2, 3). of service without increasing the stock Supply chain management is one of level and bearing additional costs (5). the main sources of competitive Supply chain management entails advantage for companies and its effective replenishment and inventory Transshipment problems are variations importance is increasing due to its policies. The use of transshipment of transportation problems in which effects on solving problems faced by points in supply chains renders goods and services are distributed many companies in terms of monitored movement of stocks to between sources, storage points and mismatches between customer intermediary storage locations between destinations. Like in many cases, the demand and supply, which will lead to two echelon levels. Transshipments objective function of transshipment low levels of customer satisfaction and are effective policies for correcting problems is to minimize cost. eventually decreasing sales and discrepancies between demand and Transshipments problems were first market share (1). Logistics is a critical inventory available at specific defined by Orden (6) as an extension part of supply chain management, and locations, serving as a tool for effective of transportation problems with is used to control the flow of materials, management of stocks that are already transshipment points between the services and information taking into procured and delivered into the system sources and the destinations (7). account the cost of these activities on (4). By using transshipments as a tool Transshipment models can be used to one side and the value created in for utilizing stocks, a company can enhance cost efficient movement of

176 TEKSTİL ve KONFEKSİYON 2/2011 goods and improve the level of most of the production is via contract production right after receipt of the customer satisfaction. Dynamic manufacturing. required materials. Upon completion of transshipment problems involving production at the workshop, the order Competition is dense in the textile and inventory decisions have several is shipped via the transportation apparel industry. The main competitors locations, and in each of these company. of the company are its own locations companies have to bear fixed subcontractors, who are willing to sell The company has a depot used as a replenishment, per unit replenishment, their products directly to the transshipment point. The products and per unit holding costs. customers. The company works under manufactured at various Transshipping inventories between the pressure of strict due dates. High subcontracting facilities are supposed locations also includes fixed and per penalty fees apply in case the to be gathered at this depot for final unit costs. The objective in these kind deliveries are untimely. The relationships arrangements and packaging. Different of problems is to determine the with both suppliers and customers play items of personnel uniforms such as replenishment quantities and how key roles in the personnel garments hats, shirts and trousers may be much to transfer in each demand industry, and trust is a key factor in produced by different subcontractors. process to ensure cost minimization. In these relationships. Therefore, conformity The final packaging and labeling the studies by Herer and Tzur (8, 9), to quality standards and deadlines is processes bring these items together. the authors develop optimal and crucial. There are mainly five key However, due to tight due dates of the heuristic algorithms for dynamic players in the transshipment process: customers, and as a result of working transshipment problems incorporating with many subcontractors, the fixed replenishment and transshipment 1. The customer: The source of the company usually has to ship from the costs. A dynamic and deterministic request or order. supplier directly to the customer at the environment is considered, where the last minute. This result stems from transshipments are not used as a tool 2. The company: The decision maker. ineffective planning of the whole for emergency source of supply, but This represents our company, transshipment process. The instead they can be used to offset which receives the order from the transportation cost is considerably high different replenishment and holding customer and outsources the for these last-minute shipments as the costs at different stock locations. production via different suppliers and ateliers. usual modes of transportation cannot In this study, we use integer be used. The decision makers in the programming to solve the transshipment 3. Suppliers/Subcontractors: company wish to use their depot and problem of an apparel company. One Companies that provide raw mate- provide timely deliveries at the same of the core aims of this study is to rials, semi-finished and finished time. ensure that the offered model and the products used in the production The problem is analyzed within the corresponding solution can be used process. context of a transshipment scheme. practically by the company, leading to 4. The workshop: The atelier in which Basic elements of the transshipment more efficient management of the the production activity is outsourced/ problem are the capacity of supply supply chain. In Section 2, we define performed. points, requirements of the demand the problem in detail. Next, we develop points, and the costs associated with the mathematical model in Section 3. 5. The transportation company: The shipping of one unit of product from Numerical results are presented in firm that is responsible for the each supply point to each Section 4. Discussion and conclusion transportation of finished products transshipment point, from each is presented in Section 5. to customers. transshipment point to each demand The process flow of the order point, and from each supply point to 2. PROBLEM DEFINITION fulfillment process is provided in Figure each demand point directly. The We consider a real-life problem faced 1. The process begins with the receipt objective is to minimize the total by an apparel company located in of a customer order. Then, the company transportation cost. We model the Izmir, Turkey. The company currently communicates with its suppliers to problem as a transshipment problem operates in the segment of ready-to- discuss if required materials are with due date restrictions. Unlike wear garments. The company has two already available at stocks or if they typical transshipment models, the due basic branches of operations; namely, should be manufactured. Hence the date of an order is also considered in the casual wear branch and personnel lead time for each order is determined our model since the company faces garments branch. Personnel garments and the order is scheduled for high penalty costs. We present our are manufactured partly in-house, but production. The workshop starts model in the following section.

Figure 1. The order fulfillment process.

TEKSTİL ve KONFEKSİYON 2/2011 177

3. THE MATHEMATICAL MODEL We define the indices and parameters used in the model as below: s: supplier index t: transshipment point index c: customer index p: product type index

Dcp: Demand quantity depending on customer and product type.

DDcp: Deadline of order depending on customer and product type. lsp: Lead time of production depending on supplier and product type. cst: Cost of shipping one unit from supplier point to transshipment point. csc: Cost of shipping one unit from supplier point to the customer. ctc: Cost of shipping one unit from transshipment point to the customer. tst: Shipping time from supplier point to transshipment point. tsc: Shipping time from supplier point to the customer. ttc: Shipping time from transshipment point to the customer. prp: The percentage of volume a specific product p occupies in a package. ⎧1, if supplier sp is eligible for providing product Eps = ⎨ ⎩0, otherwise

We define our decision variables are given below. Binary variables: ⎪⎧1 if product p, which is produced by supplier s, is assigned to customer c on the route S→ C⎫ X scp = ⎨ ⎬ ⎩⎪0 otherwise ⎭ ⎪⎧1 if product p, which is produced by supplier s, is assigned to customer c on the route S →T ⎫ Ycpst = ⎨ ⎬ ⎩⎪0 otherwise ⎭ ⎪⎧1 if product p, which is produced by supplier s, is assigned to customer c on the route T→ C⎫ Zcpt = ⎨ ⎬ ⎩⎪0 otherwise ⎭ Integer variables:

Xpcs: Number of packages shipped from supplier s to customer c.

Ypst: Number of packages shipped from supplier s to transshipment point t.

Zptc: Number of packages shipped from transshipment point t to customer c.

Our objective function in Equation 1 minimizes the sum of the batch-shipment-costs.

nn nn nn Min (1) ∑∑Xpcscsc***++ ∑∑ Ypc stst ∑∑ Zpc tctc sc==11 st == 11 tc == 11

Batch-shipment-cost is represented by three different cost components: The total cost of sending packages directly, the total cost of sending packages from suppliers to transshipment points and the total cost of sending packages from transshipment points to customers.

Every customer must be served with each product either directly from a supply point or through the transshipment point. This constraint is reflected by the following equation:

178 TEKSTİL ve KONFEKSİYON 2/2011

nn ∑∑XZscp+= cpt 1, ∀cp, (2) st==11 Constraint set 3 below ensures time limits for direct shipments. The company works with strict deadlines for each order. These time constraints make our model different from the typical transshipment problems. Sum of the lead time and the shipping time from supplier to customer cannot exceed the deadline of an order. Note that, instead of including due dates and related penalty costs in the objective function, strict deadlines are included as constraints in the model. This is due to the fact that there are very high penalty costs for late orders, and the company decision makers never prefer this situation regarding future relationships with customers. So it can be said that the company actually has deadlines rather than due dates.

ltDDMsp+≤ sc cp +(*(1 − X scp )),∀cps,, (3)

Constraint set 4 ensures time limits for shipments via transshipment points. Sum of the lead time, the shipping time from supplier to transshipment point and the shipping time from transshipment point to the customer cannot exceed the deadline of an order.

lttDDMsp++≤ st tc cp +(*(1)), − Y cpst ∀cpts,,, (4)

The next constraint set assures that, at every transshipment point, if a product p for customer c is sent from one of the possible suppliers to the transshipment point, then it must be sent out from the transshipment point to customer c. Hence, these constraints ensure the conservation of flow for the transshipment points.

n ∑YZcpst= cpt , ∀cpt,, (5) s=1

Xpcs denotes the number of packages shipped from the supplier to the customer directly. The next constraint set settles the total number of packages to be shipped, depending on the number of products to be shipped and their capacity usage in a package.

n Xpsc≥ ∑ X scp**, D cp Pr p ∀sc, (6) p=1

Ypst denotes the number of packages shipped from the supplier to the transshipment point, and Zptc denotes the number of packages shipped from the transshipment point to the customer. Constraint sets 7 and 8 compute the total number of packages to be shipped in a similar way similar as constraint set 6.

nn Ypstcpstcpp≥ ∑∑ Y**, D Pr ∀st, (7) cp==11

n ZpZDPrtc≥ ∑ cpt**, cp p ∀tc, (8) p=1

Not all suppliers can provide all product types. Product p can only be sent from a supplier s if that supplier is eligible to provide that product. There may be more than one eligible supplier for a product, and our model makes optimal supplier selections based on transportation cost, lead times and transshipment times. The following constraint sets ensure that only eligible suppliers are selected for each product in both direct shipments and shipments through the transshipment points.

X scp≤ E ps , ∀cps,, (9)

YEcpst≤ ps , ∀cpst,,, (10) Finally, types of variables in the model are defined.

Xpcs, Ypst, Zptc, ≥ 0 and integer, X , Y , Z binary, ∀cpst,,, scp cpst cpt (11)

4. PROBLEM DATA In this problem instance with a 20-day customers changes between 25 and planning horizon, there are 13 31 monetary units and the shipping The data used is obtained from the potential suppliers, 11 customers and takes 2 to 3 days. The unit shipment company by taking a recent snapshot 5 product types. The unit direct costs from the suppliers to the of the system involving many orders. shipment cost from the suppliers to the transshipment point is between 3 to 7

TEKSTİL ve KONFEKSİYON 2/2011 179 monetary units depending on the through 1 transshipment node. The shipped the products from the location of the supplier. The unit supplier nodes are candidate supplier suppliers directly to the customers. In shipment cost from the transshipment nodes, which provide different lead addition, the company had to make the points to each customer changes times for the incoming orders. The consolidation of the packages in the between 6 and 12 monetary units. The model therefore also performs supplier suppliers’ workshops, and the packages shipment from the supplier to the selection regarding the due dates of were transferred from one supplier to transshipment point, and from the the orders at minimum total cost of another for rearranging and transshipment point to the customer transportation. reconsolidation, which lead to conflicts takes approximately 1 day each. Table and higher labor and time costs. According to the optimal solution 1 shows the lead times of each Labeling processes were delayed, obtained from GAMS 22.5, only 4 potential supplier based on product interrupted or even cancelled, decreasing suppliers are selected to serve the 11 type. The pr values, i.e. the percentage the quality of shipments. This is partly p customers. The optimal solution is of volume that a specific product type p due to the fact that the company did depicted in Figure 2 in a network occupies in a package are estimated not prefer the suppliers giving the best representation. The numbers on the as 2% for product types 1, 2 and 3, lead times, although the manufacturing arcs of the network represent the and 8.3% for product types 4 and 5. prices were very similar among them. number of shipped packages between The demand of each customer from The total number of shipments in the any two nodes. The reasons for each product is given in Table 2. The optimal solution of our model is 15 for selecting only 4 suppliers out of 13 due date of each customer is 20 days this instance while the company had to supplier candidates are either the high for this instance. We provide the results make more than double this amount, costs of transportation from the other of the computational study in the next i.e. 30 direct shipments plus the inter- suppliers or the due date restrictions. section. supplier travels. The number of packages The optimal solution suggests that all shipped in the optimal solution is 76, packages should be shipped through Table 1. Lead times of potential suppliers while the company used 89 packages the transshipment point, which is an for each product type (in days). for shipments due to inefficient expected result of high direct shipment planning. The company’s intuitive p1 p2 p3 p4 p5 costs. The transshipment point also solution is depicted in Figure 3. s1134666 facilitates the distribution and s257666 transportation of the packages. For s388766 instance, 72 packages come from s455566 suppliers in our data set. Since these s5146795 packages are consolidated in the s6127856 transshipment point and then shipped to the customers, complete delivery to s7108977 the customer is possible in one shot. s849566 s985577 s1096676 s1187786 s1286766 s1377888

Table 2. Customer demand data (in unit of product).

p1 p2 p3 p4 p5 c1 0 0 46 0 138 c24800016 c3 1136 852 35 0 101 Figure 3. Solution of the company. c4 54 36 18 0 0 c5 42 32 15 0 0 The solution time of the model in c6 0 4 2 0 0 GAMS for this size of an instance with c7 21 14 8 0 0 334 variables is ignorable due to the c8 6 4 1 0 0 simplistic network-type formulation of c9 21 8 14 0 0

c10181230 0 the problem. The optimization model c11 0 0 0 5 0 Figure 2. The optimal solution. provides a cost advantage to the company as well as improving customer Next, we explain how the company satisfaction related with the timely 5. COMPUTATIONAL RESULTS actually proceeded in this specific shipments. When the actual cost of the The problem is modeled and solved problem instance and compare their company is compared to the results using GAMS 22.5 optimization software. intuitive solution with the optimal obtained from the optimization model, As it was stated in the previous solution. Because of the tight due it can be observed that there is a section, the gathered real life data dates and since one of their most reduction in the transportation costs includes a planning horizon where the important accounts is among these 11 almost by 35%. This means that using orders from 11 customers should be customers, the company actually did the transshipment point provides cost fulfilled using 13 supplier nodes not use transshipment point and advantage and operational efficiency.

180 TEKSTİL ve KONFEKSİYON 2/2011

6. CONCLUSION optimal suppliers in terms of cost and systems, or they are not capable of time constraints. The simplicity and using the proposed methods easily In this study, we consider a real-life easy implementation are the due to complex computational aspects. problem of an apparel company advantages of the developed model. The simple optimization model in this regarding the transportation of goods. GAMS is a commercial optimization study is proves useful in improving the We propose a transshipment model for software, and may not be readily shipment operations in both time and the problem that can be adapted available in many companies. However, cost terms. Hence, we believe that our practicably by other companies in the the proposed compact model can study contributes to the literature and industry. According to the findings of easily be handled in widely available industry by closing the gap between the study, use of the transshipment spreadsheet software such as the theoreticians and practitioners. point increases the overall efficiency MSExcel and solved by built-in and decreases the total transportation solvers. Introducing such optimization ACKNOWLEDGEMENTS cost. models into the everyday operations of We would like to acknowledge the Our transshipment model differs from small to medium size companies in the support of the company managers for previously used models in the textile and apparel industry will greatly providing the necessary data for our literature due to the addition of time increase their efficiency of operations. computations. constraints. The model also provides However, many of such companies are decision support to the supplier either not aware that some optimization selection problem by selecting the methods can be adapted to their

REFERENCES 1. Reyes P.M., 2006, “A game theory approach for solving the transshipment problem: a supply chain management strategy teaching tool”, Supply Chain Management: An International Journal, 11, 288–293.

2. Wu Y., 2008, “A mixed-integer programming model for global logistics transportation problems”, International Journal of Systems Science, 39, 217–228.

3. Harrison T.P., 2003, “Principles for the strategic design of supply chains”, in: The Practice of Supply Chain Management: Where Theory and Application Converge, T.P. Harrison, H.L. Lee and J.J. Neale (Eds), USA: Kluwer Academic Publishers, 3-12.

4. Herer Y.T., Rashit A., 1999, “Lateral Stock Transshipments in a Two-Location Inventory System with Fixed and Joint Replenishment Costs”, Naval Research Logistics, 46, 525-547.

5. Ozdemir D., Yucesan E., Herer Y.T., 2006, “Multi-location transshipment problem with capacitated transportation”, European Journal of Operational Research, 175, 602-621.

6. Orden A., 1956, “The transshipment problem”, Management Science, 2, 276-286.

7. Hemaida R.S., Kwak N.K., 1994, “A Linear Goal Programming Model for Trans-Shipment Problems with Flexible Supply and Demand Constraints”, Journal of the Operational Research Society, 45, 215-224.

8. Herer Y.T., Tzur M., 2001, “The Dynamic Transshipment Problem”, Naval Research Logistics, 48, 386-408.

9. Herer Y.T., Tzur M., 2003, “Optimal and heuristic algorithms for the multi-location dynamic transshipment problem with fixed transshipment costs”, IIE Transactions, 35, 419-432.

Bu araştırma, Bilim Kurulumuz tarafından incelendikten sonra, oylama ile saptanan iki hakemin görüşüne sunulmuştur. Her iki hakem yaptıkları incelemeler sonucunda araştırmanın bilimselliği ve sunumu olarak “Hakem Onaylı Araştırma” vasfıyla yayımlanabileceğine

karar vermişlerdir.

2011 YILINA ABONE OLMAYI UNUTMAYINIZ

TEKSTİL ve KONFEKSİYON 2/2011 181