Solving a Real-Life VRP with Inter-Route and Intra-Route Challenges

We address a rich vehicle routing problem (VRP) motivated by practice. The problem includes three challenges that have received at most little attention in literature: inter-route restrictions, unmatched pickups and deliveries, and priority orders. These characteristics arise in a VRP application where a large international herd improvement company needs to plan milk sampling and measuring tools distribution activities. The latter require expensive equipment that is of limited availability, resulting in inter-route restrictions due to the use of a central inventory. The equipment can also be moved directly from one farm to another, and thus can be modeled as unmatched pickup and deliveries. We solve the problem using a large neighborhood search heuristic, where we also employ local search procedures to intensify the search on promising solutions. We show that we are able to generate plans with more tasks planned and at the same time with 5% less distance traveled as compared to current practice.