A hybrid simulated annealing and variable neighborhood search algorithm for the close-open electric vehicle routing problem

Abstract

Electric Vehicles (EVs) are the future of transportation, but due to their battery and charging technology they cannot yet directly replace traditional vehicles. Nonetheless, EVs are a great option for city-logistics, due to the small distances and their zero local emissions. In this paper, a novel variant of the Electric Vehicle Routing Problem (EVRP), called Close-Open EVRP (COEVRP), is presented. It considers ending EV trips at Charging Stations, as opposed to other EVRP variants that only allow for en-route charging. This new variant follows a traditional routing scheme, allowing EVs to recharge only at the end of their route. The objective is to minimize energy consumption, as well as the number of vehicles. The energy consumption function takes into account the weight of the transported items. A mathematical formulation for the problem is presented and small instances were solved using a commercial solver. To solve larger instances, a hybrid metaheuristic combining Simulated Annealing and Variable Neighborhood Search algorithm was employed and thoroughly tested.

Study on Routing and Scheduling for Unmanned Electric Loaders Considering Charging

“…The constraint (11) represents the calculation method for determining the arrival time of the delivery vehicle at the customer point, which is based on its arrival time at the last customer node. Considering the inviolable time window constraints, two cases are primarily considered: one where the delivery vehicle arrives before the earliest service time of the last customer node, resulting in waiting time; and another where it arrives after this earliest service time.…”

“…This approach helps to avoid premature convergence and escape from local optimal solution traps, which is of certain guiding significance for considering electric vehicle charging in MDVRPTW. Stamadianos et al [11]proposed a hybrid metaheuristic algorithm that combines simulated annealing and variable neighborhood search to solve the Closed-Open Electric Vehicle Routing Problem (COEVRP), significantly reducing energy consumption and vehicle numbers on large instances. Xu et al [12]introduced a Genetic Algorithm (GA) incorporating Large Neighborhood Search (GA-LNS), which outperformed standalone GA, Simulated Annealing (SA), and Tabu Search (TS) algorithms in benchmark tests for MDVRPTW simulation.…”