ABSTRACT
G. Geoffrey Vining Virginia Polytechnic Institute and State University, Blacksburg, Virginia
Diane A. Schaub University of Florida, Gainesville, Florida
Carl Modigh Arkwright Enterprises Ltd., Paris, France
1. INTRODUCTION
An important approach for optimizing an industrial process seeks to find operating conditions that achieve some target condition for the expected value for a quality characteristic (the response) and minimize the process variability. Vining and Myers (1990) suggest that the response and the process variance form a dual response system. They use the dual response methodology proposed by Myers and Carter (1973) to find appropriate operating conditions. This dual response approach allows the analyst to see where the process can achieve the target condition and where the process variability is acceptable. As a result, the engineer can make explicit com promises. Del Castillo and Montgomery (1993) extend this method by show ing how to use the generalized reduced gradient, which is available in some spreadsheet programs such as Microsoft Excel, to find the appropriate oper ating conditions. Lin and Tu (1995) suggest a mean squared error approach within this context. Copeland and Nelson (1996) suggest a direct function minimiation of the mean squared error with a bound on how far the esti mated response can deviate from the desired target value.
