| Summary: | Thesis (M.Eng.)--Massachusetts Institute of Technology, Dept. of Civil and Environmental Engineering, 1999. === Includes bibliographical references (leaves 59-60). === Over the last twenty years the transportation industry has undergone a dramatic shift into container operations. The advantages of this mode of transportation are numerous, especially for the ocean carriers. The use of containers adds a high degree of versatility to their ships and increases the utilization of the vessels by means of a remarkable decrease in the loading and unloading operations time. However, the introduction of the containers adds, as well, a considerable investment cost to an industry that was already very capital intensive. The pressure of the high cost investment in equipment in addition to a remarkable competition in the sector forces every player in the industry to try to obtain the maximum efficiency in the utilization of its assets. Global trade is not in general balanced, and so the demand for containers at the different ports of the world varies greatly. As a result of this unbalanced situation, empty containers must be reallocated from mainly importing areas to those at which the overall outflow of freight is larger than the inflow. Managing the container inventory and the container reallocation, subject to the particular requirements of the industry and the present and future demand is known as the Container Allocation Problem. The purpose of this thesis is the development of a model for this problem so as to maximize the profit to be obtained from the management of a shipping line container inventory. The container avocation problem is modeled by the user of a large-scale, multi-stage stochastic network formulation that incorporates the uncertainty factor in the demand side of the problem. This network formulation captures the space-time dynamics of the reallocation process while using an objective function that minimizes the cost of the container operations in the long run. A continuous rolling horizon to limit the number of nodes in the network is used in the modeling of this system so as to make this problem tractable. Finally, a solution algorithm for this problem is proposed. The algorithm decomposes the initial non-linear network formulation into an iteration of successive linear approximations that can be solved via a classical linear programming method. === by Ricardo Balzola. === M.Eng.
|
|---|
