An improved optimization algorithm for a multi-depot vehicle routing problem considering carbon emissions.

In a multi-depot vehicle routing problem (MDVRP) of same-city delivery, driving distance and actual loading can greatly influence the amount of carbon emissions generated. This paper considers fuel and carbon emission costs as part of total costs, proposes a MDVRP with minimized logistics costs and driven distance, and then establishes a mixed integer programming model. An improved chemical reaction optimization algorithm is also designed by considering this problem's characteristics (i.e., a greedy search strategy is presented to generate an initial population), and two coding approaches (i.e., two-part coding and matrix coding) are applied prior to designing four chemical reaction operators. The simulation experiment is carried out using a set of a random instances and the experimental results demonstrate that one can reduce carbon emissions by driving extra lesser distances, providing a methodological guide for MDVRPs with logistics costs and carbon emissions.