Comparing fast VRP algorithms for collaborative urban freight transport systems: a solution probleming analysis

This paper proposes a comparison between two fast heuristic algorithms to solve a multi-carrier 2E-VRP in city logistics, under realistic conditions. We propose a cluster-first route second algorithm to compare the performance of two route construction and post-optimization algorithms on real-size test cases. The clustering phase is made by a seep algorithm, which defines the number of used vehicles and assigns a set of customers to it. Then, for each cluster, which represents a vehicle, we build a min-cost route by the two following methods. The first is a semi-greedy algorithm. The second is a genetic algorithm that includes post-optimization at the level of each route. In this work we make the route construction and post-optimization without any possible exchange of the routes to guaranty a pertinent comparison between both algorithms. After presenting both approaches, we apply them, first to classical 2E-CVRP instances to state on the algorithm capabilities, then on real-size instances to compare them. Computational results are presented and discussed. Finally, practical implications are addressed.