Abstract
Given the high level of complexity and wide application of the vehicle routing problem (VRP), numerous solution algorithms have been developed for the past several decades, including many recent meta-heuristic algorithms with great success and popularity. In order to balance computational load and solution quality and to address the issue of flexibility and simplicity, this study developed a heuristic algorithm based on several classical mathematical programming techniques. The VRP is first formulated in the form of the set covering problem (SCP), and the Lagrangian relaxation is used as the backbone in designing the iterative algorithm. In addition, a concept similar to column generation is used to maintain a partial set of potential routes to reduce computational load. Based on the numerical experiment, the solution quality of the heuristic algorithm is stable. The result suggests that the solution algorithm should be able to deal with the operational problems arising from a highly dynamic environment.
