Home

[Lecture, Oct 21] Approximations for Single-Warehouse Multi-Port Inventory Systems

time: 2019-10-20

Title: Approximations for Single-Warehouse Multi-Port Inventory Systems

Speaker: Prof. SHU Jia, School of Economics & Management, Southeast University

Time: 3:30-5:00 pm, October 21, 2019

Venue: Room 214, Building 37, Wushan Campus


Introduction to the Speaker:

Prof. SHU Jia is a professor and doctoral supervisor at School of Economics & Management, Southeast University. He received a PhD in High Performance Computation for Engineered Systems from Singapore-MIT Alliance Program of National University of Singapore. His research fields include Logistics and Supply Chain Management, Traffic Management, and Medical Management.


Abstract:

This paper develops an inventory management model to understand how free storage and free release times provided by ports affect a firm's inventory decision. Adopting the classical echelon inventory modelling approach, we synthesize the effects of free storage and free release times in a single-warehouse multi-port system and construct a model for integer-ratio policies. The structural properties of the continuous relaxation of the integer-ratio policy model are analyzed to facilitate the development of an $O(N\log N)$ algorithm for the relaxed model, where $N$ stands for the number of ports. Although the optimal cost of the relaxed model is not a lower bound on the average cost of any feasible inventory policy (we construct a counterexample to demonstrate this), we show that it is a lower bound on the average cost of any feasible zero-inventory-ordering (ZIO) policy. The solution to the relaxed model can be rounded to a power-of-two policy that is close to the true ZIO optimum. In many practical scenarios, e.g., when the length of time for free storage and free release at each port does not exceed its EOQ reorder interval, such a policy is at least 93\%-effective. Based on the model, we also derive several interesting insights on the pertinent issues in inventory management. For instance, we show that the optimal solution to the relaxed model can well synchronize the two-echelon inventory replenishment activities asymptotically.