Ant colony system with characterization-based heuristics for a bottled-products distribution logistics system

The aim of this paper is to show the solution of the Vehicle Routing Problem with Time Windows (VRPTW) as a key factor to solve a logistics system for the distribution of bottled products. We made a hybridization between an Ant Colony System algorithm (ACS) and a set of heuristics focused on instance characterization and performance learning. We mainly propose a method to make a constrained list of candidate customers called Extended Constrained List (ECL) heuristics. Such a list is built based on the characterization of the time-window and the geographical distribution of customers. This list gives priority to the nearest customers with a smaller time window. The ECL heuristics is complemented by the Learning Levels (LL) heuristics, that allows the ants to use the pheromone matrix in two phases: local and global. In order to validate the benefits of each heuristics, a series of computational experiments were conducted using the standard Solomon's benchmark. The experimental results show that, when the ECL heuristics is incorporated in the basic ACS algorithm, the number of required vehicles is reduced by 28.16%. When the LL heuristics is incorporated, this reduction increases to 36.83%. The experimentation reveals that, by a suitable characterization, preexisting conditions in the instances are identified in order to take advantage of both of the ECL and LL.