Adaptive large neighborhood search for multi-trip vehicle routing with time windows

We consider a multi-trip vehicle routing problem with time windows (MTVRPTW), in which each vehicle can perform several trips during its working shift. This problem is especially relevant in the context of city logistics. Heuristic solution methods for multi-trip vehicle routing problems often separate routing and assignment phases in order to create trips and then assign them to the available vehicles. We show that this approach is outperformed by an integrated solution method in the presence of time windows. We use an automatic configuration tool to obtain efficient and contextualized implementations of our solution methods. We provide suitable instances for the MTVRPTW as well as an instance generator. Also, we discuss the relevance of two objective functions: the total duration and the total travel time. When minimizing the travel time, large increases of waiting time are incurred, which is not realistic in practice.