Approximate models for a fixed-life perishable product in multi-echelon inventory systems | |
| Author | Kanchana Kanchanasuntorn |
| Call Number | AIT Diss. no.ISE-04-03 |
| Subject(s) | Inventory control Perishable goods |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Engineering, School of Advanced Technologies |
| Publisher | Asian Institute of Technology |
| Series Statement | Dissertation ; no. ISE-04-03 |
| Abstract | Motivated by the agro-food industry, a major industrial sector in many Asian countries including Thailand, this study addresses multi-echelon inventory-distribution systems for a fixed-life perishable product under a periodic review policy. The demand at each retail outlet is assumed stationary and normally distributed. Products are supplied by outside suppliers to the central warehouse, with zero product age, immediately before leaving the suppliers. Transportation lead-times from the outside suppliers to the central warehouse and between two adjacent echelons are assumed constant, with different positive values. Stockouts at all echelons are considered as backorders, based on the firstcome- first-served allocation policy, while the unsatisfied demand at retailers is considered lost, based on the observed nature of the agro- food industry in Thailand. The purpose of this study is to develop appropriate inventory models for the described multi-echelon distribution systems. The approach comprises mainly four phases. The first phase verifies necessity of the model development as set forth, by investigating the effect of product perishability and stockout policy at retailers on certain key performance measures. The knowledge and understanding of such effects subsequently lead to the derivation of an approximate cost model in the second phase. Simulation experiments show that, in most situations, perishability causes negative effects on all performance measures. In the second phase, a simple modification of Matta and Sinha's (1995), an existing periodic-review inventory model of a two-echelon inventory system, is made to incorporate perishability and lost sales stockout policy, resulting in an approximate inventory model that can improve system performances. Each retail outlet i in the lower echelon applies (R,S;) inventory policy with an identical review interval R and different maximum inventory levels S;. The central warehouse in the higher echelon, on the other hand, applies (R,s,S) policy, where R is the same review interval as that of retail outlets; s is its reorder point; and S is its desired maximum inventory level. Simulation experiments are carried out to evaluate effectiveness of the proposed model when compared with the original model and to understand the model performances with respect to certain parameter values. The experimental results show that, with simple modifications of the proposed model, a significant improvement in the system total cost over the original model can be consistently achieved. Additionally, the proposed model is modified to instead maximize the net profit function. Due to the absence of comparable models, its effectiveness is subsequently evaluated and compared with that of OptQuest, a commercial simulationoptimization package. The experiment confirms that the proposed model can provide relatively good solution quality in a relatively small computational time. The third phase is to extend the modification methodology proposed in the second phase and to apply it to the three-echelon distribution systems. Our approach is to firstly develop relevant models for non-perishable products. Specifically, based on the extension of Matta and Sinha's (1995) model, three-echelon cost models are developed for two different inventory policies at distribution centers, namely (R,~) and (R,sj,8.J) policies. These models are subsequently modified further to address perishability and lost sales in the total cost and net profit functions. Their simulation results, as compared with those obtained from OptQuest and the extended non-perishable models, show that all the proposed models can provide reasonably good solution quality within a short computational time, and that the perishable models can yield significant improvements in net profits and total costs over the nonperishable models. Finally, extensions of the proposed models to the case of general n-echelon and slow-moving Poisson demand pattern are proposed. Our extension procedure to the system of n-echelon with specific inventory policy at each echelon is relatively simple. The same remark can also be made for the model whose demand distributions are Poisson. The outcomes from simulation experiment measuring the effectiveness of Poisson model show its superiority, in that it can give significant improvements in net profits, especially when the demand is sufficiently low and the usable lifetime is relatively short. |
| Year | 2004 |
| Corresponding Series Added Entry | Asian Institute of Technology.|tDissertation ; no. ISE-04-03 |
| Type | Dissertation |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Industrial Systems Engineering (DISE) |
| Academic Program/FoS | Industrial Systems Engineering (ISE) |
| Chairperson(s) | Anulark Techanitisawad; |
| Examination Committee(s) | Bohez, Erik L.J.; Huynh Trung Luong; Do Ba Khang;Minner, Stefan; |
| Scholarship Donor(s) | The Royal Thai Government; |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 2004 |