Vehicle routing and the value of postponement

Abstract

We study capacitated vehicle routing problems (CVRPs) in which the client demands occur over time and the collector or distributor performing the service can decide when to visit the clients. It has been reported in the literature that postponement of collection (or delivery) services may decrease the overall routing cost or distance. We investigate this issue in greater detail and report on experiments in different routing settings to uncover the impact of postponement on routing efficiency. We compare the results by a CVRP local search procedure with those obtained from an analytical, Continuous Approximation model. While the two approaches tend to generate similar results, the differences grow larger when the CVRPs’ bin packing characteristics become more intricate.

Keywords:

• capacitated vehicle routing problem
• postponement
• local search
• continuous approximation theory