The thesis focuses on the proposed optimization methods (metaheuristics and mathematical models) and their coupling with a simulation model to solve problems of collection including delivery scheduling drivers. The originality of this work focuses on the diversity of resources (vehicle, driver, dock loading, unloading, container production line, cleaning area) and constraints (incompatibility vehicle / container, start date earlier desired end date, planning ...) to take into account. The objective is to provide an organization to achieve all transport while minimizing delays and overtime. The first part focuses on the transport of a single type of product. The problem is modeled as a demand profile RCPSP with variable resources. Empty transports are modeled as time-dependent assembly sequence. Two integer linear programs are proposed. The second part concerns the transport of several types of product. The problem has a double complexity that is determined by the coupling of an iterated local search with a simulation model. The simulation model allows to meet the structural and functional complexity, mainly because of the diversity of resources. The third part includes the definition of working hours of drivers. An iterative approach including a simulation model, an integer linear program and previously presented coupling is proposed. This problem is treated in a hospital setting for transporting containers clean or dirty (food, clothes, medicine) between sites of consumption and production sites. Each party is subject to an experiment with real data.
