| Summary: | 碩士 === 元智大學 === 工業工程與管理學系 === 104 === Since the transportation technology were highly developed, the global trade booming and more popular. The goods are transferred in and out the warehouses and between companies frequently. Companies need a logistics team to plan the transportation and to locate the right warehouse to store. For cost considerations, many companies outsource transportation and warehousing activities to the third-part logistics providers. Those third-party logistics usually provide use their own warehouse. The traditional warehouse has four major functions: receiving, storage, order picking and shipping, while storage and picking are costly activities. Cross-docking which is a distribution strategy in supply chain becomes more popular in recent years. It can be controlled efficiently to reduce the lead time. In cross-docking, freight is unloaded from inbound vehicles and directly loaded into outbound vehicles, with little or no storage in between.
In this paper, we considered a cross-docking system which combines the vehicle routing problem with cross-docking (VRPCD) for both inbound and outbound operations and multi-door dock assignment (DA) problem at the cross docking terminal. The objective of the problem is to minimize the sum of transportation cost, vehicle fixed cost and warehouse operation cost. Since the combine problem is a NP-hard problem, it is hard to use exact solution approach to solve it. We proposed an ant colony optimization (ACO) algorithm to solve the problem, and used the Gurobi optimization software to validate the mathematical model and tested for small size instances.
The proposed ACO is tested with benchmark problem sets of instances from the literature for the VRPCD and DA. The computational results and comparisons with best heuristics in the literature show that the ACO can obtain the same or better average solutions in 86 out of 90 instances for the VRPCD. In the dock assignment problem, the ACO can find the optimal solutions or better results for those instances that Gurobi cannot solve within the time limitation. In the combined problem, we proposed eight ACOs to solve this problem. We found the better results were provided by solving the integrated problem iteratively instead of solve the VRPCD and DA sequentially. In such an iterated approach, the solution diversity can be increased. In the future, we could consider other approaches and feedback mechanism between ant colonies to improve the ACO.
|
|---|
