以混合遺傳演算法求解整合性區位存貨、定址以及運輸路徑之兩階層供應鏈分銷網路設計

Incorporating Inventory, Location and Routing Decisions in Two-Echelon Supply Chain Network Design: A Hybrid Genetic Approach

本研究探討兩階層供應鏈設計模式，其供應鏈層級包括單一的供應商、數個物流中心及顧客群，同時依顧客需求量的多寡，區分為大顧客群及小顧客群，分別以單點運送及多點運送的方式運輸貨物。本研究的目的係在成本最低的情形下，求取物流中心所在的位址、數量以及運輸的途徑。因為本研究模式包含定址－－存貨整合問題以及車輛路徑問題，兩者均屬困難性的問題，故本研究運用三階段遺傳演算法求解，演算的結果證明本研究所提出的演算法可有效率的達到近似最適解，同時顯現在最低成本的要求下，在目標式中相關成本間相互抵換關係下，決定最適物流中心開設的數目、位址及車輛路徑等決策。

This study presents a two-echelon supply chain network design (2E-SCND)model consisting of a vendor, a number of potential distribution centers (DCs), andend customers. According to the demand size, we classify the end customers into twosets of clients, big and small, which are replenished by direct shipping and routingpolicies, respectively. Facility location is a strategic decision that requires vastcapital investment. Failure to account for shipping costs for potential clients canlead to a sub-optimal facility location model. Therefore, this study aims not only todetermine the number and location of DCs, but also to consider the distribution planfor big and small clients. The problem belongs to the class of NP-hard problemssince it contains the location-inventory problem (LIP) and the multi-depot vehiclerouting problem (VRP), both of which are NP-hard. The study develops a geneticalgorithms-based three-phase heuristic approach to resolve this problem. Theexperimental results indicate that the proposed algorithms can efficiently yieldnear-optimal solutions and demonstrate the trade-off among the related costs.