Exploratory Research on Engineering the Transport Industries (ETI): Solving Large-Scale Logistics Problems in Real-Time: Models, Algorithms, and Information Systems

University Of Florida, Gainesville FL

Investigators

Abstract

This grant is made under the Exploratory Research on Engineering the Transportation Industries (ETI) program solicitation. This award provides funding for the development of new optimization models, algorithms, and software for solving large-scale logistics problems, in which cost effective solutions are sought by integrating transportation and inventory decisions. The network and assignment structure inherent in many logistics problems will be utilized to develop efficient algorithms to solve these problems. Due to the scale of these problems, the investigator will concentrate on efficient heuristic methods. In particular, the development of greedy heuristics and neighborhood search techniques for solving the assignment component, integrated with heuristics for solving the nonlinear network flow component. The focus on heuristics seems particularly promising for taking advantage of the larger availability of operational data, which calls for a frequent reoptimization of the flow of goods through a logistics network, based on data as it becomes available. If successful, the results of this research will lead to improvements in the design of logistics systems and the development of new heuristics for solving complex non-linear and integer network flow problems. The primary goal of this work is to produce logistics systems in which operational decisions can be made more efficiently, by providing a much better starting point for making these decisions. The factors that will be taken into account for the evaluation of the costs of a logistics network design are the, generally conflicting, costs of transportation and holding inventory, as well as customers' demands for service, such as just-in-time deliveries and single-sourcing. This will help to achieve an overall cost reduction in comparison to the traditional two-stage approach where the transportation and inventory decisions are decoupled. The proposed work will also contribute to the theory of assignment and network flow problems by providing new and efficient heuristic for such problems.