Ship-pack replenishment optimization in a two-echelon distribution system with lost sales and seasonal product obsolescence

Abstract

As a retailer attempts to leverage a two-echelon distribution system to forward-deploy inventory, a number of cost elements must be considered when deciding the quantity of a given SKU to replenish to the forward-deployed node. These costs include processing costs, transportation costs, and obsolescence costs, among others. An important consideration that is often overlooked is the elevated processing costs related to sending single units, or “eaches”, instead of units packed in full case quantities. Furthermore, for seasonal retailers, there are often times large obsolescence costs associated with products that reach the end of their life-cycle and are still stocked at the forwarddeployed node. The goal of this work is to develop a cost-optimal replenishment strategy with high inventory productivity, taking into consideration the end-to-end supply chain costs that are faced when forward-deploying seasonal products in a two-echelon distribution system.

The thesis focuses on the decision of replenishing eaches versus full cases, and in the case of full case replenishment, the optimal rounding logic to use when deciding whether to replenish a full case of a stock-keeping-unit (SKU) or to send zero units. In the most simplistic case of full case replenishment, a retailer will decide to replenish a full case if the order quantity is non-zero. However, a more nuanced approach that replenishes a full case at specific thresholds (i.e. when the order quantity is at least 50% of the full case quantity) can lead to lower end-to-end supply chain costs. The thesis creates a simulation model that uses a base stock model to estimate the resulting inventory productivity of the system and the costs associated with each replenishment policy.

The model simulates over 8,000 SKUs for a seasonal retailer and finds that the proposed replenishment system can significantly improve inventory productivity (i.e. inventory turns) relative to the retailer’s current replenishment system. We pilot the inventory policy in the retailer’s forward deployed node and validate that the system’s inventory performance in practice is comparable to the modeled performance. The validation provides further confidence in our cost optimization model, which finds that an optimal replenishment policy can lead to between 2.7% - 4.1% savings relative to a baseline replenishment policy that replenishes eaches. A sensitivity analysis is also conducted on cost inputs to show the impact of optimal replenishment decisions if the retailer’s cost structure is impacted. The cost input change is formulated as an internal carbon tax that would significantly impact transportation costs. The sensitivity analysis concludes that the optimal replenishment policy consistently yields 0.7 - 1.8% savings as costs are varied.