SBIR Phase II: Parts Forecasting for Configurable Products

This SBIR Phase II project will develop a new methodology for parts forecasting for discrete manufacturing. Emcien is developing a software suite to enable a product manager to better manage a configurable manufactured product. This suite includes a method for forecasting the demand for a configurable product at the full configuration level of detail. This means forecasting unique configurations, each with an expected volume. The method depends on extracting customer buying patterns from the sales history for the product. The mathematical algorithms for extracting and representing these patterns, and forecasting using these patterns are the main contributions of the research. The set of parts needed to build a configurable product generally depends on combinations of options, so it is not possible to plan parts requirements from an aggregate forecast. By using a configuration level forecast, it is possible to expand each unique configuration into component parts, and then use the associated volumes to produce a complete parts forecast. American manufacturers are specializing in complex, configurable, high-end products, as mass produced commodity products move offshore. Allowing customers to customize a product results in significant numbers of alternative product configurations. This variety increases costs in many ways. One important way is the increased difficulty of planning parts requirements. The current practice of basing parts planning on a few popular variants leads to excess inventory of some parts and shortages of others. Excess inventory incurs both holding and obsolescence costs. Shortages can interrupt production and cause both lost sales and quality problems. Emcien has developed a methodology that, among many other benefits, can improve the accuracy of parts planning.