Stochastic optimization methods for vehicle routing and network design problems

点击翻译按钮获取中文摘要

Whenever they appear in practical settings, combinatorial optimization problems oftentimes involve a level of uncertainty (stochastic environment) that needs to be factored in when developing appropriate models to solve them. The reasons for this being that solutions that are optimal in stochastic settings are usually structurally different from their deterministic counterparts. In an uncertain environment, when companies or organizations face such problems, they need to find solutions that need to be both robust (i.e., remain efficient under different contextual scenarios) and resilient (i.e., able to quickly rebound when unexpected events occur). Communicational and informational technologies are now more readily available and applied in various ways to obtain more information on the parameters of the problems and get a better understanding of how uncertainty appears and affects them. However, adapted stochastic optimization models are needed to properly formulate the decisional dynamics of problems that involve uncertainty and to efficiently utilize the available information pertaining to the context. The present research program is dedicated to the study of this important issue in the context of solving both transportation (vehicle routing) and logistics (network design) problems. Three general areas of research are defined in the present program. The first two focus on the problems considered: Area 1 Dynamic and stochastic vehicle routing models and Area 2 Stochastic network design models. In these areas, general projects will be undertaken: in Area 1, Consistency in stochastic vehicle routing problems (project 1.1) and Dynamic urban vehicle routing problems (project 1.2); in Area 2, Designing robust and resilient networks (project 2.1) and Capacity planning under uncertainty (project 2.2). As for the third area, it is dedicated to the development of general solution strategies for stochastic problems (Area 3 Methodological development), which will be done in three projects: Decomposition strategies (project 3.1), Hybrid solution methods (project 3.2) and Parallel methods (project 3.3).