A three-stage matheuristic for the blood stochastic inventory routing problem

This research introduces a blood distribution system under vendor-managed inventory that considers uncertain supply and demand. We present it as the Blood Stochastic Inventory Routing Problem, formulating it as a two-stage stochastic programming model. To solve this problem, this study proposes a three-stage matheuristic that combines a perturbation heuristic, Adaptive Large Neighborhood Search, and an exact approach. From historical data of Surabaya Blood Center in Indonesia, six sets of new instances are generated under different settings. Computational results show that our proposed three-stage matheuristic outperforms CPLEX and a two-stage matheuristic by gaining optimal or better solutions within a significantly shorter computational time. Moreover, it is robust for solving large problems, as evidenced by its ability to find high-quality solutions within a reasonable time. Finally, managerial insights are derived by evaluating performance matrices under different uncertainty levels and scenarios. According to these insights, some practical strategies are suggested with respect to the decision-maker's risk preferences and demand characteristics.