A Novel Optimal Policy Structure for Managing Inventory in Assemble-to-Order Systems
In the first part, we address the problem of inventory replenishment and allocation for a manufacturer who sells an assembled product as well as individual spare parts. We model the problem as a Markov decision process, with state space consisting of component inventory levels. By partitioning the state space into multiple disjoint lattices based on products’ component requirements, we establish the optimality of base-stock production and stock rationing policies on each lattice. Our computational results reveal the practicality of such lattice-dependent base-stock and rationing policies for ATO systems with general product structures.
In the second part, we present an approximate dynamic programming method to the general ATO problem, under Markovian assumptions on production and demand. We approximate the optimal cost function by reducing the state space of the original problem via a novel aggregation technique, which may significantly alleviate the computational burden. We show the optimality of lattice-dependent base-stock and rationing policies for the aggregate problem. We also derive error bounds for this approximation and provide computational results. We conclude with a discussion of future research directions in this field.
Parts of this research are joint work with Alp Akcay from Bilkent University, Mustafa Akan and Alan Scheller-Wolf from the Tepper School of Business, Carnegie Mellon University.
Short Bio: Emre Nadar is an assistant professor in the Department of Industrial Engineering at Bilkent University. He received his B.S. degree in Industrial Engineering from Bilkent University, and his M.S. and Ph.D. degrees in Operations Management from the Tepper School of Business, Carnegie Mellon University. His research interests include stochastic dynamic programming, queueing theory, supply chain management, and sustainable operations.
