A Variable Neighborhood Search Integrated in the POPMUSIC Framework for Solving Large Scale Vehicle Routing Problems

This paper presents a heuristic approach based on the POPMUSIC framework for solving large scale Multi Depot Vehicle Routing Problems with Time Windows derived from real world data. A Variable Neighborhood Search is used as the optimizer in the POPMUSIC framework. POPMUSIC is a new decomposition approach for large scale problems. We compare our method with a pure VNS approach and a Memetic Algorithm integrated in a POPMUSIC framework. The computational results show that the integration of VNS in the POPMUSIC framework outperforms the other existing methods. Furthermore different distance metrics for the decomposition strategies are presented and the results are reported and analyzed.