Departure time optimization of a set of dependent routes

In this thesis, we look into departure time optimization in a network of dependent routes. The routes can be dependent, for example, because a pickup in one route needs to take place before the delivery in a different route. Currently, this problem is solved by ORTEC by optimizing the routes one by one. In this thesis, a labeling algorithm is proposed to solve the complete network simultaneously. Different approaches of this algorithm are given, both exact and heuristic. By means of numerical experiments the proposed algorithm is compared to the algorithm as currently used by ORTEC. From the experiments we find that small instances can be solved to optimality within acceptable computation times. However, when the instances grow bigger, the computation times of the exact approach grow exponentially. The heuristics show varying performances, both in computation time and solution quality, where the construction algorithms show the most promising results. When multiple time windows are present, the computation times of all versions of the algorithm show a large increase.