Ph.D. Industrial Engineering and Engineering Management
xv, 171 leaves : ill. ; 30 cm
Abstract
Storage space and yard cranes are two scarce resources used in container storage yard. The efficient allocation of these two resources translates directly to the efficient yard operations. In this research, we develop a systematic approach to allocate storage space to incoming containers and dynamically deploy cranes among storage blocks to ensure high throughput of containers. The performance of our approach in all the sub-problems is numerically calibrated and found to be significantly better than the real life practice....[ Read more ]
Storage space and yard cranes are two scarce resources used in container storage yard. The efficient allocation of these two resources translates directly to the efficient yard operations. In this research, we develop a systematic approach to allocate storage space to incoming containers and dynamically deploy cranes among storage blocks to ensure high throughput of containers. The performance of our approach in all the sub-problems is numerically calibrated and found to be significantly better than the real life practice.
The study begins with the space allocation problem. According to the storage characteristics of a yard with the yard-crane system, the space allocation process is divided into two stages. First, in storage block assignment, the number of containers allocated to each block is determined. Then in storage position assignment, the exact positions of containers within each block are determined. Because of the multiple objectives involved in the storage block assignment, the problem is further broken down into two levels, with level I for aggregate container assignment and level II for vessel-based container assignment. At level I, the number of inbound and the number of outbound containers assigned to each block, in aggregate level without distinguishing vessels, are found from an integer programming model with the objective of minimizing the weighted average of the imbalance in the total number of containers and of the imbalance in the number of vessel-related container moves across blocks. The results from level I are refined at level II at which the vessel identifications of the assigned containers are found by another integer programming model together with a transportation model. Two types of relationships among containers, independent (containers from different customers) and group (containers from the same customer and hence are interchangeable), are considered at level II.
The objective in the stage of storage position assignment is to minimize the total reshuffling in each block. Because of their different characteristics, the inbound and outbound containers are treated differently. For inbound containers, the storage positions in a block are determined under two departure patterns: first-come-first-leave and random pickup time; for outbound containers, the storage positions are determined under two arrival patterns: simultaneous arrivals at the beginning of a period and disperse arrivals throughout a period.
After finishing the storage space assignment, the workload in each block can be estimated and hence can be used to dynamically deploy cranes among blocks. The dynamic crane deployment problem is formulated as a mixed integer programming model. Two crane deployment policies are studied with the objective of minimizing the leftover workload at the end of a planning period. Under the first policy, each crane is allowed to move only once in a planning period. The task is to determine the optimal deployment time and the route of cranes. A Lagrangean relaxation heuristic is used to solve the problem. Two solution techniques are developed to improve the performance of Lagrangean relaxation method. Under the second policy, cranes are allowed to move more than once in a deployment period. The task is to determine the optimal deployment frequency and the route of cranes. A least cost heuristic is developed to solve the problem.
Post a Comment