An Efficient Near-Exact Algorithm for Large-Scale Vehicle Routing with Time Windows

In this paper, we present an eff icient near-exact algorithm for the vehicle routing problem with time windows (VRPTW). The LP relaxation of the set partitioning model of the VRPTW is solved by generating columns on a suitably reduced network. A heuristic bounding strategy and a randomized algorithm are then applied to find the integral solution. This algorithm is capable of obtaining nearly optimal solution even for 100customer problem with wide time windows. Particularly, we are able to improve the best published results for several Solomon′s benchmark problems (Solomon 1987) which cannot hitherto be solved optimally. Hence, our approach can potentially deliver quality solutions to VRPTW in real-time.