Content area
Full Text
Response surface methodology (RSM) is a collection of statistical design and numerical optimization techniques used to optimize processes and product designs. The original work in this area dates from the 1950s and has been widely used, especially in the chemical and process industries. The last 15 years have seen the widespread application of RSM and many new developments. In this review paper we focus on RSM activities since 1989. We discuss current areas of research and mention some areas for future research.
KEY WORDS: Bayesian Designs; Computer Experiments; Generalized Linear Models; Multiple Responses; Robust Parameter Design; Split-Plot Designs; Variance Dispersion Graphs.
Introduction
Previous Review Articles
Box and Wilson (1951) laid the foundations for response surface methodology (RSM) (a list of abbreviations is provided in the Appendix). That paper is important not only because it described what became an entire field of research for the next 50 years but also because it changed dramatically the way that engineers, scientists, and statisticians approached industrial experimentation. Their paper outlined a sequential philosophy of experimentation that encompasses experiments for screening, region seeking (such as steepest ascent), process/product characterization, and process/product optimization. Clearly, RSM includes much more than second-order model fitting and analysis. Indeed, RSM, broadly understood, has become the core of industrial experimentation. Box and Liu (f999) illustrated the application of RSM to the common training example of paper helicopters. Box (1999) provided a retrospective on the origins of RSM. That paper also outlined a more general philosophy of sequential learning, of which RSM is one tool.
Over the past fifty years, there have been three extensive reviews of response surface methodology. The Hill and Hunter (1966) review paper featured an extensive bibliography and presented applications in the chemical and process industries, where the majority of RSM applications were found at that time, which was natural given that the seminal work of Box and Wilson occurred at a major chemical company. The Mead and Pike (1975) review paper in UiomeiMc.s focused more on the modeling of biological data than on a discussion of RSM as we view it.
The most recent review paper was that of Myers, Khuri, and Carter (1989). They emphasized the changes that had occurred in RSM theory and practice during the 1970s and 1980s....