Naval Wholesale Inventory Optimization

Abstract

The U.S. Naval Supply Systems Command (NAVSUP), Weapon Systems Support, manages an inventory of approximately 400,000 maritime and aviation line items valued at over $20 billion. This work describes NAVSUP’s Wholesale Inventory Optimization Model (WIOM), which helps NAVSUP’s planners establish inventory levels. Under certain assumptions, WIOM determines optimal reorder points (ROPs) to minimize expected shortfalls from fill rate targets and deviations from legacy solutions. Each item’s demand is modeled probabilistically, and negative expected deviations from target fill rates are penalized with nonlinear terms (conveniently approximated by piecewise linear functions). WIOM’s solution obeys a budget constraint. The optimal ROPs and related expected safety stock levels are used by NAVSUP’s Enterprise Resource Planning system to trigger requisitions for procurement and/or repair of items based on forecasted demand. WIOM solves cases with up to 20,000 simultaneous items using both a direct method and Lagrangian relaxation. In particular, this proves to be more efficient in certain cases that would otherwise take many hours to produce a solution.