Agere Systems Inc.
David Wu, Chris A. Armbruster, Berrin Aytac, Rosemary T. Berger
Lehigh University
09/01/03
08/31/04
Development of methods for characterizing the demands for short life-cycle technology products using a leading-indicator engine that identifies products that provide advanced warning of demand changes for a group of products.
*managing the capacity in its increasingly complex, global supply chain
In this research, we explore the development of new methods for characterizing demand and the applicability of these methods within the context of capacity planning. Specifically, we focus on developing a new approach for characterizing the demand of short-lifecycle technology products with the specific purpose of facilitating capacity management in a contract manufacturing environment. The short lifecycle and volatile demand of high-tech products poses serious challenges to the time series forecasting methods fundamental to commercial demand planning systems. As an alternative, we propose a demand characterization method based on the concept that one or more products could provide advanced warning of demand behavior for a cluster of products. To validate our approach, we conduct an extensive analysis of a data set from Agere that covers more than 3500 products over a twenty-six month period. Through several case studies, we investigate the opportunities for applying the approach to the process of planning capacity.
The analysis identified leading indicators that predicted the demand pattern of the product group one to seven months ahead of time with correlation values ranging from 0.51 to 0.95.
We have identified products that provide advanced warning of demand changes and that produce reliable demand forecasts. The following have been identified as significant results of the work:
1. Products can be grouped into clusters based on similarity measure
* Technology
* Manufacturing Resources
* Statistical Characteristics
* Lifecycle Patterns
2. One or more products can be identified whose demand pattern predicts overall demand of cluster
* Correlation of demand pattern in relation to its group
* Time lag by which demand pattern leads the group
The research of this paper has several implications for the operations of Agere:
3. Advanced warning provides a time-lagged model that can predict the demand patterns of a broader group
4. Demand of broader group is important to planning functions
* Negotiating and securing appropriate capacity within foundry partners
* Projecting revenue for financial forecasts