Comparison of policies in dynamic routing problems

Abstract

We consider a company that has to satisfy customers' pick-up requests arriving over time every day. The overall objective of the company is to serve as many requests as possible at a minimum operational cost. When organizing its business the company has to fix some features of the service that may affect both service quality and operational costs. Some of these features concern the time a request is taken into account to plan its service, the associated deadline and the way requests are managed when the system is overloaded. In this paper we analyse several policies that can be implemented by the management of a carrier company in a multi-period context. For example, a company might reject all the requests that cannot be feasibly scheduled or accept all the requests and rely on a backup service in order to serve requests that are difficult to handle. Another interesting issue considered in this paper is the impact of collaborative service where two or more carrier companies, with their own customers, decide to share customers in order to optimize the overall costs. We set up a general framework to allow comparison of alternative service policies. Extensive computational results evaluating the number of lost requests and the distance travelled provide interesting insights.