An Experiment on Stack-next Algorithm for Dynamic Container Stacking Problem at Sea Port

Sammanfattning

Container usage in seaborne transportation is increasing day by day. That's why it is also becoming more challenging to manage the containers in a container terminal. To handle more containers in a limited area, containers are stacked in a place called yard by using yard crane. The yard crane is also used to retrieve the containers when it is time for the container to move to another place. Since the handling of the containers are becoming more difficult, it is important to use the yard crane with a good plan which will reduce number of moves of the yard crane and it will save time and energy. This research focuses on export flow and import flow of the containers. Here, export flow means when the containers come to the container terminal from trucks and trains, then stacked in the yard and then loaded in a vessel. The import flow goes to the opposite direction. The proposed solution for this research is a newly developed algorithm called stack-next algorithm for stacking the containers. The idea is: before starting to stack, the container terminal management team must know which containers will come and make different groups of containers. The group will indicate which containers will be retrieved on which time. Based on the grouping, the stack-next algorithm will stack the containers in a way that the containers that need to be retrieved earlier will stay above the containers that need be retrieved later. An experiment was conducted on 1000 different scenarios and tested on two different block sizes. There were 400 containers grouped in 5 different groups and in each scenario, the order of the containers were different. The proposed solution was able to give promising results on 80% of the physical capacity. The stacking of the containers was so efficient that it requires only one move per container to stack and only one move per container to retrieve. No relocation of the containers would take place. But the result was achieved in computer generated randomized ordered scenarios. This needs more testing on how it would work on practical scenarios.