149x Filetype PDF File size 2.63 MB Source: ft.unand.ac.id
FUZZY MULTI-OBJECTIVE PERIODIC REVIEW INVENTORY PROBLEM IN A DYADIC SUPPLY CHAIN SYSTEM Dicky Fatrias and Yoshiaki Shimizu Department of Mechanical Engineering, Toyohashi University of Technology Toyohashi 441-8580, Japan Email: dicky@ise.me.tut.ac.jp, shimizu@me.tut.ac.jp ABSTRACT This paper presents a fiizzy periodic review inventory model in a dyadic supply chain by incorporating some uncertain parameters. To cope with such uncertainty, a fuzzy multi-objective approach is introduced. A solution procedure is proposed such that several fuzzy goals are satisfied and, in the process, the optimal ordering policy and inventory level are determined.Through providmg hypothetically constructed case problem, the usefulness of our proposed model is demonstrated. KEY WORDS: fiizzy multi-objective problem, inventory, dyadic supply chain 1. INTRODUCTION satisfied and, in the process, the optimal The issue of considering uncertainty in ordering policy and safety stock of manufacturer inventory problem has received a great deal of and the optimal target stock level of retailer are attention in the field of production/inventory determined. Through providing hypothetically management. In the context of periodic review constructed case problem, we provide the system several researches have studies such acceptable solutions. issue using stochastic approach under different concems.Recent researches carried out in this direction include Song & Lau (2004), Bijvank& 2. SYSTEM DESCRIPTION Johansen (2012), Prasertwattana & Shimizu. In what follows, the proposed inventory model (2007) and Fatrias& Shimizu (2010).With the will be described briefly. In all cases, we put the development of fiizzy set theory (FST), the following assumptions. fiizzy approach is also employed for the 1. The manufacturer uses the periodic review modeling of uncertain parameters inventory with lot sizing policy andsafety stock to problems. The FST copes with the uncertainty control its inventory. related to unavailability and incompleteness of 2. The retailer uses the periodic review with data as well as and imprecise nature of goals of target stock levelto control its inventory. which the use conventional probability 3. Only a single product is considered in the distribution impossible in this case. model. Without loss of the generality, the In this regard,this research proposes a fiizzy manufacturer uses one unit of raw material multi-objective periodic review inventory model to produce one unit of product. in a typical Supply chain (SC) system in 4. For both manufacturer and retailer, only one whichsingie-manufacturer, single-retailer is order is allowed to place at any period. considered. Specifically, we attempt to develop 5. Production rate of the manufacturer is a fiizzy periodic review inventory model in a assumed fixed and higher than the mean mixed imprecise and/or uncertain environment demands. by incorporating the fiizziness of demand, lead 6. Unfulfilled demand at manufacturer is time and cost parameters. considered as backorder while unfulfilled To cope with such problem, solution procedure demand atretailer is considered as shortages. is proposed such that several fiizzy goals are Managing Assets and Infrastructure in the Chaotic Global Economic Competitiveness 197 The system model is described based on the manufacturer. foUowingnotation listed for major parameters. p = Unit purchasing cost of retailer. c = Unit holding cost of finished product of Index retailer. Number of planning horizon. T = J = Unit shortage cost of finished product t Period(/ = 1,2,...,7). of retailer. Number of days in each period. TCm = Total Cost of manufacturer. h = Parameters of Manufacturer TCr = Total cost of retailer. A = Forecast demand of manufacturerat Decision Variabies period/. LS = Lot sizing policy of manufacturer. Q, = Order quantity of manufacturer at ss = Safety stock level of manufacturer. period/. S = Target stock level of retailer. PR = Production rate of Manufacturer. Im, = lead time of raw material delivery ffomsupplier at period /. Supplier j-j * Manufacturer Retailer —f-»-|^ustomers Qpn = Production quantity produced at period/. Es, = Ending stock of raw materials of Figure 1. System Configuration at period /. Ess, = Ending safety stock of at period /. The members in this chain consist of one Qm, = Ordering quantity at period /. supplier, one manufacturer, one retailer, and end Qb, = Backorder quantity at period /. customers as shown in Figure 1. However, this Qsl, = Sales volume at period /. study focuses on a dyadic relationship in the EI, = Ending inventory at period /. chain between the manufacturer and the BR = Backorder rate of manufacturer. retailer(the supplier and endcustomers are considered as external members in the Parameters of Retailer chain).We assume that these two members are d, = End customer demand at period /. owned and controlled by a central company. A = Lead time of product delivery from Inventoryof each memberis controlled by a periodic review in make-to-stock environment, manufacturerat period /. in whichdemand and leadtime, and cost In = Ending inventory of finished product parameters are considered as a fuzzy at period /. number.The inventory level of manufacturer and Qsr, = Shortage quantity after receiving retailer are reviewed at every period time /, over replenishment at period /. totally T periods (plaiming horizon). Each Qor, = Order quantity at period /. period consists of interval of time tp days. Qre, = Replenishment quantity received at period/. 2.1 Manufacturer LR = Loss rate of retailer. Manufacturer receives raw materials from Cost Parameters outside supplier which has unlimited capacity, transforms it to finished product and then 0, = Order cost of manufacturer at period /. distributes the products to retailer. However, the r = Unit purchasing cost of manufacturer. supplier may delay the supply of raw materials m = Unit production cost of manufacturer. to the manufacmrer. Therefore, the manufacturer h = Unit holding cost of raw material of has to select the appropriate material ordering policy and hold safety stock of product to cope manufacturer. with the uncertainty in demand and delivery / = Unit holding cost of product of lead-time. manufacturer. The ordering quantity of manufacturer is b = Unit backorder cost ofmanufacturer. directly influenced by lot sizing policy {LS) T = Unit transportation cost of which is adopted for ordering raw material. 198 Managing Assets and Infrastructure in the Chaotic Global Economic Competitiveness After the best pattern of LS is selected, the Qb, manufacturer will check the amount of Min 5R, =E {Qorj) (3) inventory on hand at the beginning of the period. If the amount on hand is less than the sum of the demand and the amount to fill back Min LRj='Z (4) the safety stock, then the manufacturer will place the order to supplier. Otherwise no order will be issued. The manufacturer can start the production at the 3. SOLUTION METHODOLGY beginning of each period if raw material on hand The proposed fuzzy periodic review inventory exists; otherwise the manufacturer has to wait model is actually a multi-objective mfaced imtil arrival of raw material fi-om the supplier by integer programming model (MOMIP). To solve timer+/ffj, As a consequence, the production the model, a solution procedure is proposed, quantity imder the combined order condition First, the equivalent crisp MOMIP model is may become higher than "lot-for-lot" case and converted into a single-objective MIP model. results in higher capability to supply retailer's Then, one evolutionary optimization search demand (lower shortage cost) at the expense of method named Differential Evolution (DE) is higlier holding cost. applied to find an optimal solution. 2.2 Retailer 3.1 The Auxiliary Crisp MOMIP Model The retailer makes a regular order to the Transforming a fuzzy MOMIP model into an manufacturer periodically to raise up the auxiliary crisp MOMIP model require an inventory to the target stock level. The order appropriate method. For this purpose, Jimenez quantity(gor,) is determined by comparing the methodis applied because it is computationally ending stock level (/r,) at the review time t with efficient to solve a fuzzy problem (See Jimenez the desired target stock level (5), which is equal et al., 2007).According toJimenez method, the to (S-Ir,). This target stock level is not only to auxiliary eq./ (l)-(4) can be formulated as cover the end customer's demand but also to follows: cover the effect of its fluctuation as well as the Min TCm = late delivery and unfulfilled quantity of products fi-om the manufacturer. t + Z Q, 2.3 Objective Functions ( bi^' + 2b'"" +b'"''\ This study considers four objective functions to i. Es. Qsl. evaluate the system performance. The first (=1 objective minimizes total cost of T {Ess. + EI,) manufacturer(rCm); the second objective minimizes total cost of retailer(TCr); the third objective function minimizesbackorder rate of ^ti" +2t'"" +t°<"^ manufacturer (BR); and thefourth objective Qsl, (5) function minimizesloss rate of retailer(I/?). Min TCr = Min TCm= j;^d + f^SxQ,+f^hxEs,+ pP"+2p""+p rc'*'-l-2c"'"-tc^'l Qsl,+t Ir, tcx{Ess,+EI,)+tb^Qb,+ -^2s""+s*"'' Qsr, (6) (=1 (=1 t^xQsl, (1) (=1 Qb, Min BR,='Z (7) Min TCr = tp>'Q^l,+idxIr,+ZsxQsr, Qorj) (=1 (=1 (=1 (2) Managing Assets and Infrastructure in the Chaotic Global Economic Competitiveness 199 4. COMPUTATINAL EXPERIMENT Qsrj To illustrate the usefulness of the fuzzy MOMIP in LRj=t model using the proposed solution procedure, a Min numerical experiment is provided and the result (=1 is reported in this section utilizing input parameters shown in Table I. FomuUlc fuzzy MOMINLP 3.2 The Proposed Solution approach pciiodic invelofy model The steps of the proposed solution procedures Detennine membership functioir GcDcrale initul population are summarized as follows (Figure 2): for fiizzy puumeters ind of Inrget vector. O - 0 Stepl: Formulate the fiizzy MOMIP objeelive fiincriom Comptfte and evaluate the (MOMINLP) periodic review inventory model I Convert the MOMINLP into in fitness of each target vector as described in section 2. 1 equivnlent crisp model i Apply muution, crossover Find the range of each of objective and selection operator to Step 2: Determine the appropriate generate new target vector function by cntcubling their membership function for fuzzy parameters and minimum and maximum vahie \ objective functions. In this formulated problem, target vector fiizzy parameters and objective functions are Cooveil Ac equivalent crup MOMINLP model into a single- represented by linear membership fiinction. objective MINLP using Zimmerman no method Termination Step 3: Convert the fiizzy MOMIP into an I auxiliary crisp MOMIP model. To this end, all DifTcrential Evolution Algoridun the imprecise cost parameters in the objective I functions as well as the demand and lead time Gain tfie acceptable solution parameters are converted into the crisp ones using Jimenez method. Step 4: Determine the rage of each objective Figure 2. Solution methodology function by calculating the minimum and maximum value of each of them. To calculate 4.1 Setting the Lower and Upper Bound the minimum and maximum value of each For LS, the manufacturer has to decide whether objective function, the auxiliary multi-objective to make the order at the beginning of every crisp model should be solved each time only one period or combine the order in a big batch. objective. Therefore, the binary coding is applied to Step 5: Convert the auxiliary crisp MOMIP represent the value of LS. Thejjand .S are model into a single-objective MlPbased on considered as the amount of products (units) at Zimmermann'saggregation function (Zimmermann, I993).The formulation of the manufacturer and the retailer, respectively. Zimmermann'saggregation function is as So the integer coding is used to represent these follows: values. Lower bound of LS is 0,which means"not place the order" in current period but Max 2 (9) combine it to the previous period's order. subject to: Upper bound of LS is 1 means "place the 2,
no reviews yet
Please Login to review.