A decision support system for the dynamic hazardous materials vehicle routing problem

The problem of delivering hazardous materials to a set of customers under a dynamic environment is both relevant and challenging. The objective is to find the best routes that minimize both the transportation cost and the travel risk in order to meet the customers' demands or needs, within predefined time windows. Aside from the difficulties involved in the modeling of the problem, the solution should take into consideration the demands revealed overtime. To deal with this problem, a solution approach is required to continuously adapt the planned routes in order to respond the customers' demands. In this paper, the dynamic variant of the Hazardous Materials Vehicle Routing Problem with Time Windows (DHVRP) is introduced. Besides, a decision support system is developed for the DHVRP in order to generate the best routes, based on two new meta-heuristics: a bi-population genetic algorithm and a hybrid approach combining the genetic algorithm and the variable neighborhood search. An experimental investigation is conducted to evaluate the proposed algorithms, using Solomon's 56 benchmarks instances and through several performance measures. We show through computational experiments, that the new approaches are highly competitive with regards to two state-of-the-art algorithms.