Optimal Solution of the Local Delivery Problem Through Minimum K-Trees

Abstract

The local delivery problem has been extensively studied and is a classic in the field of vehicle routing and scheduling. This talk will describe a recently developed optimization algorithm for this problem which exploits a connection between the local delivery problem and minimum K-trees. I will first set the stage for this discussion by reviewing the enormous activity in the vehicle routing area that has occured during the last decade in both academia and industry. I will focus in particular on projects in which I have been involved at Air Products and Chemicals, DuPont and Exxon that produced implemented routing models and algorithms with significant economic benefits. Altough all of these algorithms are heuristics (as has been the focus of almost all past research in this area), I will indicate why the time is now ripe to consider optimization approaches.