| Summary: | 碩士 === 國立東華大學 === 運籌管理研究所 === 100 === This thesis considers the pre-marshalling problem of export containers. Pre-marshalling not only decreases the turnaround times of container vessels but also the operation costs of container terminals and marine corporations. The thesis only considers a single container bay, and assumes that the stowage plan of a vessel dictating the retrieving sequence of containers is set before the container vessel arrival. This retrieval sequence is represented by individual identifications of containers. The container with a smaller identification has higher priority to move onto the container vessel.
Because there are many ways to pre-marshal a container bay, the ultimate goal in this thesis is to seek a feasible solution in each instance such that the number of reshuffles is small. We develop three algorithms to pre-marshal export containers. We code the algorithms in Java, and generate 1000 problem instances for each problem setting to check the performance of the algorithms.
After pre-marshalling a problem setting, we calculate the number of successful instances and the average number of reshuffles in the three algorithms. We have the following conclusions: For the (7, 5)-, (6, 5)-, (5, 5)- and (4, 5)-bays, we recommend using the EFSS-SS algorithm to pre-marshal. For the (3, 5)-bay, we prefer to use the SS algorithm. For the quality of the algorithms, the number of reshuffles does not exceed two times the number of containers in the container bay, and the algorithms can complete the pre-marshalling process with very short computational time.
|
|---|
