A Candidate List Extension for Ant Colony System Solution of the VRP with Time Windows

In this paper the solution of the Vehicle Routing Problem with Time Windows (VRPTW) using the Ant Colony System (ACS) is approached. A state-of-the-art implementation of ACS uses a candidate list to incorporate geographic information of the customers. We propose a new candidate list extension to include information of the length of the time windows, for identifying those clients with a small length of time window in order to give them a larger visit priority when the ants search fora new solution. To validate the benefit of this strategy, a series of computational experiments were conducted with the use of the standard Solomon’s Benchmark.The results show that the proposed strategy is able to reduce by a 2.82\% the average number of vehicles used for the Solomon non clustered instances.