Solving the battery swap station location-routing problem with capacitated electric vehicles using an AVNS algorithm for vehicle-routing problems with intermediate stops

- We show how to extend an AVNS for the VRPIS to address the BSS-EV-LRP.
- The extended AVNS is able to significantly improve previous results from the literature.
- The AVNS is robust showing small differences between average and best solution quality on the established instances.
- The AVNS strongly reduces the number of BSSs used in a solution compared to previous results.
- Therefore, we introduce additional benchmark instances which prove to be more meaningful with regard to the necessity of using BSSs.

In this paper, we show how to extend solution methods for vehicle-routing problems with intermediate stops (using the example of an Adaptive Variable Neighborhood Search (AVNS) algorithm) to solve the recently introduced battery swap station location-routing problem with capacitated electric vehicles. The problem calls for the simultaneous determination of (i) the battery swap stations (BSSs) to be constructed out of a set of candidate locations, and (ii) the electric vehicle routes to serve a set of customers with the goal of minimizing the sum of construction and routing cost. On the benchmark instances from the literature, the extended AVNS is able to significantly improve the previously known best solutions for the large majority of instances while using only a small fraction of the run-times reported for the comparison methods of Yang and Sun (2015). Moreover, the AVNS proves robust with regard to its average solution quality and is able to strongly reduce the number of constructed BSSs in the solutions compared to the results from the literature. Therefore, we generate additional benchmark instances which prove to be more meaningful with respect to the necessity of using BSSs and that are suitable to analyze the impact of varying construction cost on the location decision.