Time-dependent vehicle routing problem of perishable product delivery considering the differences among paths on the congested road

Many companies that conduct a perishable product delivery face a practical problem when an increasingly congested road setting exists. An inappropriate routing scheme not only leads to higher delivery costs but also results in customers' dissatisfaction. The previous literature paid limited attention to the time-dependent difference among paths (between distribution centres and customers or between customers and customers). We developed a time-dependent vehicle routing model considering the differences among paths on the congested road (TDVRP-DP) for perishable product delivery. Given the availability of road setting-related data, we proposed a method to obtain the time-dependent travel time based on the historical traffic index. We built a TDVRP-DP model to minimize the total cost and minimize the dissatisfaction of customers. Solution algorithms with a dichotomy strategy for DP and new evolutionary operators based on several multiobjective evolutionary algorithms (MOEA) were proposed. The computational results of different sizes of problems show that the dichotomy strategy for DP saves about 50% of the computation time and the new evolutionary operators improve the performance of the algorithm slightly in most cases; the NSGA-III-DN performs well for small size (with 20 customers) problems, and the RVEA-DN exhibited better performance for larger (with 50, and 100 customers) ones.