The GRASP Metaheuristic for the Electric Vehicle Routing Problem

The Electric Vehicle Routing Problem (EVRP) is a recently formulated combination of the Capacitated Vehicle Routing Problem (CVRP) and the Green Vehicle Routing Problem (GVRP). The goal is to satisfy all customers' demands while considering the vehicles' load capacity and limited driving range. All vehicles start from one central depot and can recharge during operation at multiple charging stations. The EVRP reflects the recent introduction of electric vehicles into fleets of delivery companies and represents a general formulation of numerous more specific VRP variants. This paper presents a newly proposed approach based on Greedy Randomized Adaptive Search Procedure (GRASP) scheme addressing the EVRP and documents its performance on a recently created dataset. GRASP is a neighborhood-oriented metaheuristic performing repeated randomized construction of a valid solution, which is subsequently further improved in a local search phase. The implemented metaheuristic improves multiple best-known solutions and sets a benchmark on some previously unsolved instances.