Optimization and simulation of a cross-docking terminal. (Optimisation et simulation d'une plate-forme gérée en cross-dock)

Cross docking is an efficient and dynamic logistic process that directly transfers goods from a supplier to a customer. This thesis addresses the door assignment and resource management problem with truck time windows constraints for a specific cross dock. The problems are formulated as mixed integer programming (MIP) models. The objective is to minimize the weighted sum of the total travel distance and cost which includes labor cost and penalty cost. A heuristic based on genetic algorithm (GA) is developed to solve the problems. Two gene repair methods are proposed to repair infeasible solutions. The computational results show that genetic algorithms outperforms the solution of MIP model with CPLEX in a given CPU time, for medium and large size instances, and that the second gene repair method outperforms the first one. In order to describe the behavior and gather information on the cross dock, a model based on Petri net is built to study the cross docks and simulations are carried out with Tina. The simulation results for different resource number lead us to obtain the relevant times to improve the original time windows, the makespan at each work station, the free time interval in the cross dock and the free resource number at each time period, which provide relevant information on the resource management. Besides, according to the simulation results, the original MIP formulations are improved. Then we propose a new MIP formulation, which determine not only door assignment, but also resources at each time period at each work station. Computational results reveal that the new MIP model control resources in cross dock more efficiently and outperforms the first model.