The simulated trading heuristic for solving vehicle routing problems

We present an improvement heuristic for vehicle routing problems. The heuristic finds complex customer interchanges to improve an initial solution. Our approach is modular, thus it is easily adjusted to different side constraints such as time windows, backhauls and a heterogeneous vehicle fleet. The algorithm is well suited for parallelization. We report on a parallel implementation of the Simulated Trading heuristic on a cluster of workstations using PVM. The computational results were obtained using two sets of vehicle routing problems which differ in the presence of time windows. Our results show that Simulated Trading is better suited for problems with time windows.