A Dynamic Programming-Based Heuristic for Solving the Two-Echelon Dynamic Lot-Sizing Problem with Backorder

• Main Authors: Szu-JuLin, 林思如
• Other Authors: Shine-Der Lee
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/zcxzyf

碩士 === 國立成功大學 === 工業與資訊管理學系 === 103 === A two-echelon dynamic lot-sizing problem with backorder is addressed in this thesis. The supply chain consists of an upstream supplier and a downstream retailer. The supplier orders or produces items to meet retailer's order, and shortage is not allowed. The retailer procures stock from the supplier to satisfy customer’s demand. When customer’s demand can’t be satisfied with the on-hand stock from the retailer, the shortage is backordered and filled as soon as new replenishment is available. For a given stream of demand data in finite periods, the objective is to determine the period and the associated procurement quantity to replenish the stocks for both supplier and retailer, to minimize the total relevant cost. It includes: ordering cost, procurement cost, inventory carrying cost in the supply chain, and shortage cost of the retailer. Four dominance properties of the dynamic lot sizing problem are presented. Based on these properties, an efficient heuristic that runs in polynomial time Ο(n^4) is developed, where n is the number of planning periods. The computational experiment and result have shown that the dynamic programming-based heuristic is very efficient and effective. It takes less than 1 second to solve a lot-sizing problem with n =70 periods. In comparison with the integer programming model, the average deviation of total cost is less than 0.2%.