TABLE OF CONTENTS Response surface methods (RSM) are primarily relevant when the decision-maker desires (1) to create a relatively accurate prediction of engineered system input-output relationships and (2) to "tune" or optimize thoroughly of the system being designed. Since these methods require more runs for a given number of factors than screening using fractional factorials, they are generally reserved for cases in which the importance of all factors is assumed, perhaps because of previous experimentation.
The methods described here are called "standard response surface methods" (RSM) because they are widely used and the prediction models generated by them can yield 3D surface plots. The methods are based on three types of design of experiments (DOE) matrices. First, "central composite designs" (CCDs) are matrices corresponding to (at most) five level experimental plans from Box and Wilson (1951). Second, "Box Behnken designs" (BBDs) are matrices corresponding to three level experimental plans from Box, Behnken (1960). Third, Allen et al. (2003) proposed methods based on so-called "expected integrated mean squared error optimal" (EIMSE-optimal) designs. EIMSE-optimal designs are one type of experimental plan that results from the solution of an optimization problem.
We divide RSM into two classes: (1) "one-shot" methods conducted in one batch and (2) "sequential" methods based on central composite designs from Box and Wilson (1951). This chapter begins with "design matrices" which are used in the model fitting part of response surface methods. Next, one-shot and sequential response surface methods are defined, and examples are provided. Finally, a brief explanation...
