A variable neighborhood descent with ant colony optimization to solve a bilevel problem with station location and vehicle routing

Electric vehicles are becoming popular in transport systems due to subsidies provided by the governments and the reduction of environmental issues. A bilevel optimization problem rises when the interests of governments (minimizing the infrastructure costs) and transportation companies (minimizing the routing costs) are considered. Also, both electric vehicles and internal combustion vehicles can be used, increasing the complexity of the problem. This work proposes a Variable Neighborhood Descent combined with an Ant Colony Optimization with local search and a Route Selection Procedure for solving a bilevel optimization problem. Variable Neighborhood Descent is applied at the upper level in the Station Allocation Problem while Ant Colony Optimization with local search and Route Selection Procedure are applied to the lower level in the Vehicle Routing Problem. Computational experiments were performed using two different sets of instances and the results obtained indicate that the proposal achieved good results at both levels when compared with other approaches from the literature, with low construction and routing cost and always keeping the proportion of electric vehicles higher than requested.