Rich vehicle routing problem: model and experimental design analysis
Résumé
Over the last years the Vehicle Routing Problem (VRP) has been enriched by different complicated constraints encountered in the real-life. In this paper, we tackle a rich VRP namely the Multi Compartment Multi Commodity Heterogeneous Fixed Fleet Vehicle Routing Problem with hard Time Windows (MCMCHFFVRPTW). The objective of the problem is to determine an assignment of the non-intermixable products into vehicle compartments of different sizes and a set of vehicle routes of total minimum distance and minimum waiting time such that the customer demands are satisfied. The main contribution of this paper is to identify the most relevant instance factor for this problem. Thus, we develop a generic parameterized mathematical model and a data set generator able to produce a great variety of instances based on various service policies and fleet characteristics. New sets of instances were generated by systematically varying design parameters. First, an exact resolution for four models using CPLEX 10.0 is provided. Second, we carry out a D-optimal factorial design using the Design Expert 7.0 software to select from five factors, the factors which lead to significant variations of the objective value (OV). The statistical study shows that the fitted model is adequate to analyze the effect of the retained parameters on the OV