Inventory models for milti-echelon distribution systems

Abstract

The purpose of this study is to determine an optimal or near optimal inventory policy for a two-echelon distribution system, in which one central warehouse stocks goods for distribution to retailers. The criterion is to minimize the total variable system cost per year incurred at the central warehouse and retailers as a system. The main part of this study is in the following three areas: First, this thesis presents a deterministic inventory model for a two-echelon distribution system. The model is developed under the assumption that the central warehouse and retailers order periodically. Based on the characteristics of the optimal policy, an iterative algorithm is developed to find an optimal inventory policy. Solutions of the model to a large number of test problems show that the model outperforms other existing models in the literature without sacrificing the computation time. Second, this thesis presents a stochastic inventory model for a single-echelon system in which, during a stockout period, a fraction β of demand is backordered and remaining fraction 1-β is lost. An iterative algorithm is developed to find an optimal inventory policy. At the extremes β=1 and β=0, the model presented reduces to the usual backorders and lost sales case, respectively. Third, this thesis presents a stochastic inventory model for a two-echelon distribution system. Based on the interactions between the central warehouse and retailers, an iterative algorithm is developed to find an optimal or near optimal inventory policy. Numerical experience with a large number of problems shows that the algorithm converges rapidly. A numerical example is given to illustrate the savings in annual cost by the two-echelon inventory model compared to the situation where the suboptimal inventory policies are determined sequentially, from the retailer level to the central warehouse.