Door assignment and sequencing problems in crossdocks and container terminals.

Abstract

This dissertation is concerned with improving the efficiency of two types of terminals commonly found as key facilities in transportation and supply chain networks: crossdocks and container terminals. Chapter 2 is concerned with a crossdock, which serves as a junction box, and is used by many transportation service providers as a consolidation or redistribution point. Assuming that the assignment of inbound trailers to doors is optimized on a nightly basis, and that the outbound trailers have permanent door assignments over a user-specified time period, Chapter 2 presents a simulated annealing-based heuristic procedure (SA) to determine both inbound and outbound trailer-to-door assignments in order to minimize the overall material handling effort. The proposed heuristic addresses congestion concerns as well as the robustness of the solutions in a worst case sense. Two heuristics and optimal solutions are presented to evaluate the SA on internally generated data sets for rectangular crossdocks with 12, 16, 20, and 32 doors and various shapes defined by flatness. Lastly, a new linear MIP formulation for the general Rectilinear-QAP is presented. In Chapter 3, under certain assumptions, a mathematical derivation for the optimal shape (square versus flat) of a rectangular crossdock and the best door locations are presented. Also, we present our insights on the structural properties of the optimal outbound door assignment for trailers with various flow compositions. In Chapter 4, the SA is used to obtain an outbound door assignment for a 118-door crossdock operated by Con-way. The outbound door assignment obtained with SA outperformed the current door assignment by approximately 35% on several data sets provided by Con-way. Also, the SA procedure is modified to solve a dynamic door assignment problem. In Chapter 5 we shift our focus to container terminals and address the problem of minimizing the total travel and setup time needed for a Rail-Mounted Gantry Crane (RMGC) to retrieve a pre-specified sequence of containers from the storage area in the process of loading vessels. A previously-published heuristic for routing general yard-side equipment is modified for routing RMGCs. Numerical experiments indicate that the modified heuristic outperforms the existing heuristic by an average of approximately 8% in solution quality while reducing the runtime by approximately 80%. Two initialization procedures for the heuristic are proposed, further improving the solution quality of the modified heuristic by approximately 5% while reducing the runtime by more than 452%.