Allocating Production Capacity Among Multiple Products

• Published in: Operations research Vol. 44; no. 5; pp. 724 - 734
• Main Author: Glasserman, Paul
• Format: Journal Article
• Language: English
• Published: Linthicum, MD INFORMS 01.09.1996; Operations Research Society of America; Institute for Operations Research and the Management Sciences
• Subjects: Allocations; Applied sciences; Approximation; Asymptotic properties; asymptotically optimal allocations; Average cost; Carrying costs; Cost allocation; Cost engineering; Cost functions; Exact sciences and technology; Inventories; Inventory control, production control. Distribution; inventory/production approximation; Manufacturing; Minimization of cost; Objective functions; Operational research and scientific management; Operational research. Management science; Operations research; Optimization; probability; production allocation; Production capacity; programming; Replenishment; resource allocation; Safety stock; Safety stocks; stochastic model applications; Supply & demand; Production system; Production management; Inventory control; Replenishment; Resource allocation
• DOI: 10.1287/opre.44.5.724
• ISSN: 0030-364X

We consider the problem of allocating production capacity among multiple items, assuming that a fixed proportion of overall capacity can be dedicated exclusively to the production of each item. Given a capacity allocation, production of each item follows a base-stock policy, i.e., each demand triggers a replenishment order to restore safety stocks to target levels. We present procedures for choosing base-stock levels and capacity allocations that are asymptotically optimal. Our objective is to minimize holding and backorder costs, or to minimize holding costs subject to a service-level constraint. Asymptotic optimality refers to large backorder penalties or stringent service-level constraints. Numerical results indicate that our rules perform very well even far from the asymptotic regime. A further approximation step results in allocation rules based on heavy-traffic limits; these, too, perform well.