The Continuous Time Service Network Design Problem

Abstract

The service network design problem (SNDP) addresses the planning of operations for freight transportation carriers. Given a set of requests to transport commodities from specific origins to specific destinations, SNDP determines a continuous movement of vehicles to service demand. Demand becomes available for pick up at its origin by a given availability time and has to be dropped off at its destination by a given delivery deadline. The transportation plan considers matters of vehicle routing, consolidation, service schedule, empty vehicle repositioning, assignment of freight to operating vehicles, and vehicle stops and waiting times. The literature studies a periodic time approach to SNDP. This thesis generalizes the periodic time approach to SNDP by introducing a continuous time network and model. Several network and model reduction techniques are introduced, and a multi-cut Benders decomposition is developed to solve the continuous time model. To improve convergence of Benders decomposition, we strengthen the algorithm with a family of valid inequalities for SNDP. Numerical results show the benefits of the continuous time approach. Substantial reductions in computational effort and improved lower bounds are achieved by the multi-cut Benders decomposition algorithm.