This paper addresses a rich variant of the vehicle routing problem (VRP) that involves pickup and delivery activities, customer time windows, heterogeneous fleet, multiple products and the possibility of splitting a customer demand among several routes. This variant generalizes traditional VRP variants by incorporating features that are commonly found in practice. We present two mixed-integer programming models and propose a metaheuristic based on Adaptive Large Neighborhood Search for the addressed problem. Additionally, to facilitate the use of the proposed approaches in real-world decision-making, we develop an open-source, publicly available web interface that allows one to set and solve VRP variants with the mentioned features. Computational experiments using benchmark instances with up to 150 customers show that the approaches can be used to obtain good-quality solutions in a reasonable time frame in practice.
Technical Report, Operations Research Group, Production Engineering Department, Federal University of Sao Carlos, Brazil. December, 2021.