| Summary |
Last-mile delivery is one of the critical logistic processes for third-party logistics operators to make efficient distribution plans. One of the research directions is to introduce operational flexibility such as agents to the traditional Vehicle Routing Problems (VRPs) to achieve a balance between cost and efficiency. Although the agents' cost and the penalty cost associated with compromised service quality are thus involved, the distribution network can be greatly simplified.
With the spread of the Coronavirus (COVID-19), how to arrange vaccine distribution efficiently while reducing the operational cost has become an important issue. Compared with traditional VRPs, vaccine distribution has several major differences, including the requirement of strict temperature control and extended process time due to various procedures at each stop. With limited driver working hours, vaccine distribution is more complex than traditional VRP-type distribution. It is required to consider the possible excessive needs for vehicles and drivers.
This study focuses on one specific version of the VRP considering distribution flexibility and node service time to support the planning processes of third-party logistics companies for vaccine distribution. Customers are categorized into agent customers, flexible customers, and non-flexible customers. Then, the vehicle routes and agent assignments are optimized according to the characteristics of each customer. Different from other VRPs with operational flexibility, in this study, the service arrangement of flexible customers is non-predetermined, to be served by a vehicle or assigned to an agent. This study formulates a mixed integer programming model to define the associated decision problem. A solution algorithm based on the Tabu Search method is designed to derive good quality approximate solutions within an acceptable computation time. Based on the numerical experiment, incorporating agents into consideration can reduce the routing cost and the number of needed vehicles. Also, the algorithm shows good-quality solutions and computational efficiency in each numerical example. Therefore, the mathematical model and the heuristic algorithm developed in this study should be able to serve as appropriate decision-making support for vaccine distribution operators. Nonetheless, the penalty cost of using agent assignments and the parameter combinations related to route time limit had a certain degree of influence on solution quality. |
