Optimization models and algorithms for large-scale, capacity constrained pick-up and delivery problems with time windows

Abstract

Major package delivery companies employ hundreds of thousands of people, generate billions of dollars in revenues and operate very large fleets of ground vehicles ranging from custom- built package cars to large tractors and trailers. A crucial point for the profitability of these companies is, for a given level of service, to be able to run their operations at the lowest possible cost. In this thesis, we will contemplate the problem of the scheduling and routing on a regional and daily basis of the large tractor and trailer fleet of a large package delivery company. Our aim is to design a method for building the schedules associated with minimal operating costs. We consider deterministic situations in which all parameters are known exactly and we exclude possibilities of disruptions. Nonetheless even with these simplifications, the problem we consider is complex and large-scale, containing a very large number of constraints and parameters. Throughout this thesis, we examine different theoretical approaches including optimization models and algorithms. We implement some of these approaches in order to get practical results which can be implemented in practice.