Inventory Optimization in a
Multi-Echelon Production System with
Unpredictable Demand and Level-Loading
Constraints
James A. Rappold, Ph.D.
University of Wisconsin-Madison
Motivated
by a joint strategic and tactical planning problem in a global biotech firm, we
consider a two-echelon production-inventory system that produces several
hundred items in a plant with finite, perhaps random, capacity. The fundamental
problem is that item-level demand is highly uncertain and is unpredictable in
statistical terms for the majority of items. In addition, there are
level-loading constraints (and metrics) that compel production in circumstances
in which additional inventory is not desirable. In this research, we develop a
computationally efficient optimization model. We examine a coordinated
production and inventory strategy that balances customer service goals,
production capabilities, operating policies, and inventory levels in a multi-echelon
network. We discuss implementation issues and present a simple framework to
guide necessary organizational changes.