Cyclic stochastic two-echelon inventory routing for an application in medical supply
针对德国诊所网络中的药品供应问题，提出两阶段随机规划模型，优化循环配送模式与再订购点，并引入无人机应急补货，案例显示可降低总成本29%。
Drug availability in clinics is essential for patient services, whose demand for medication is uncertain. Thus, clinics must have a variety of drugs available, leading to high inventory holding costs. In Germany, it is common for a larger central clinic to take over the procurement of drugs and distribute them to smaller surrounding clinics, which results in a two-echelon network structure. The clinics, however, operate according to their inventory policy as they plan independently. Additionally, the inventory policies include instant replenishment orders to avoid shortages, which can be executed by various vehicles, such as vans or aerial drones, because the orders only involve a few medications. We present a two-stage stochastic program for a multi-product two-echelon inventory routing problem with stochastic demands. We decide on the cost-optimal cyclic delivery patterns and reorder points for the clinics with instant replenishment orders as recourse decision. Further, we introduce an adaptive large neighborhood search with problem-specific operators that modify the routing, delivery periods, and reorder points. We present a case study at a large German clinic that supplies multiple surrounding clinics and plans to integrate drone instead of van deliveries for emergency resupply. Our integrated approach leads to cost savings of 57% for the surrounding clinics and 18% for the central clinic. Using drone delivery compared to van delivery, the average stock of medication at surrounding clinics can be reduced, resulting in a total cost decrease of 29% while maintaining medication availability. • Deciding on the cost-optimal delivery days, routing, and reorder points for clinic supply. • Introducing a two-stage stochastic program that includes emergency resupply as a recourse decision. • Developing an ALNS that executes an algorithm for determining reorder points in each iteration. • Presenting a novel case study that leads to multiple managerial insights.