报告题目：Virtual Stockpile Pooling and Real-Time Retrieval forEfficient Emergency Response
报 告 人：刘方
刘方，中国科学院大学经济管理学院副教授。博士毕业于美国杜克大学Fuqua商学院运营管理方向，本科毕业于北京大学数学学院。主要研究方向有供应链管理，机制设计，人道主义救援，可持续发展等。现在承担百人计划B项目并曾获CSAMSE会议best paper award二等奖MSOM data research challenge特奖。论文主要发表在国际顶级期刊UTD期刊Operations Research, Manufacturing & Service Operations Management, Production Operations Management等。
The traditional approach for maintaining stockpiles in the supply chain is static stockpiles (SS) where each warehouse is assigned a static inventory threshold and only the inventory above the threshold can be used to satisfy the regular demand. This approach circulates the stockpiles via the supply chain and hence reduces waste. Alternatively, virtual stockpile pooling (VSP) dynamically adjusts the thresholds among the warehouses. However, these two approaches omit the retrieval process. In this paper, we analyze the VSP with dynamic retrieval (VSPR) and benchmark it with VSP and SS with retrieval (SSR). Under SSR, allocating all the emergency supplies to the warehouse with the least long-run average operating cost is optimal. Under VSPR, we solve a three-stage optimization problem and derive the closed-form solution at the last two stages. The solutions can be obtained from an easy-to-execute finite-step method. Specifically, the optimal retrieval policy can be obtained from a greedy procedure. The optimal reallocation policy can be characterized by ranking the inventory segments rather than ranking the warehouses under VSP. The optimal order policy is a base stock policy. We also find that the emergency supplies are only reserved at some warehouses i.e., the key warehouses and identify conditions under which a warehouse is a key warehouse. In a case study, VSPR saves up to 13.8021% operational cost and improves service levels 4.9026% at most. Other than emergency response, VSPR can also be applied to supply chain management, healthcare operations, energy and natural resource management, and military defense.