Auto-Carrier Distribution Problem with Highly Variable Daily Demand at Dealers

• Main Authors: Hao-Ping Jiang, 江浩平
• Other Authors: Tsung-Sheng Chang
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/20701940147110189314

碩士 === 國立東華大學 === 運籌管理研究所 === 99 === The delivery of new automobiles to dealers is one of the major tasks to an auto-carrier’s daily operations. In practice, a truck leaves its base depot and has to return to this base depot after finishing assigned tasks. In the case of highly variable daily demand at dealers, however, forcing the trucks to back to their origin depot may not be operationally effective. To address this problem, we propose an idea that considers not only the current routing problem but also the forecast demand at dealers. We begin by providing a guiding principle for dispatching the trucks according to available information about the future demand. Next, a proper network model is built and a mathematical formulation is given; we exploit the inherent structure of the formulation and develop an algorithm based on Dantzig-Wolfe Decomposition method to solve the problem. In this Dantzig-Wolfe Decomposition algorithm, one of the subproblem is the so-called “resource constrained elementary shortest path problem”, which is solved by a dynamic programming algorithm in this study; another subproblem possesses network structure and is solved by the network simplex method. Finally we propose a heuristic to obtain a feasible integer solution. We test our algorithm with several levels of variance at dealers’ future demand. Experiment results show that the proposed algorithm is able to solve the auto-carrier distribution problem with up to 8 depots, 7 distribution center, and 112 dealers in a weekly planning problem.