Mathematical programming heuristics for the production routing problem

Abstract

This paper develops a framework and solution methodology for optimizing the integration of production, inventory, and distribution for supplying retail demand locations from a production facility. Two mathematical programming-based heuristics are developed which use a relaxed mixed integer model to determine an initial solution. One approach uses set partitioning and another approach employs the concept of seed routes to determine an approximate solution and a vehicle routing metaheuristic is used to sequence routes for each time period in the planning horizon. A multi-iteration improvement procedure determines the final solution. The proposed multi-phase approach achieves many new best known solutions to test problems for this challenging supply chain optimization problem. Additionally, experiments are conducted to assess the inclusion of fixed vehicle costs and the impact on fleet size and solution quality.