Spatiotemporal-Dependent Vehicle Routing Problem Considering Carbon Emissions

Aiming to improve the timeliness of logistics distribution and render the optimized route scheme effective under the real traffic network, we study the green vehicle routing problem with dynamic travel speed from both dimensions of time and space. A discrete formulation is proposed to calculate the travel time based on periods and arcs, which allows a vehicle to travel across an arc in multiple periods. Then, we establish a mixed-integer nonlinear programming model with minimum distribution costs including transportation costs, carbon emissions costs, and penalty costs on earliness and tardiness. A hybrid adaptive genetic algorithm with elite neighborhood search is developed to solve the problem. In the algorithm, a neighborhood search operator is employed to optimize elite individuals so that the algorithm can stimulate the intensification and avoid falling into a local optimum. Experimental instances are constructed based on benchmark instances of vehicle routing problem. The numerical results indicate that the proposed algorithm is rather effective in global convergence. Compared with the routing schemes in which travel speed merely varies with time periods or locations, the vehicle route optimized on spatiotemporal-varying speed outperforms them in terms of carbon emissions and timeliness. The research can provide a scientific and reasonable method for logistics enterprises to plan the vehicle schedule focusing on spatiotemporal-dependent speed of the road network.