Statistically Derived Static Process Models

There are a number of occasions in most manufacturing processes where product quality is measured infrequently and often offline. In these cases statistically derived steady state models have proven adequate to identify important process variables which must be better controlled to improve product quality or to increase process throughput. This paper describes some of the important factors to be considered in developing this type of model from plant operating data and defines the unique conditions which are usually present when this model are valid. Three specific examples of general processes where this approach has been found to be of value are described.