Test bed for comparing response surface methodologies

A response surface is defined as the function corresponding to the true response of a dependent variable over given ranges of independent variables. Various experimental design methodologies are used to determine the characteristics of an unknown response surface, but there is little empirical evidence indicating what strategies are most effective when dealing with specific types of surfaces or when particular characteristics are studied. We propose a test bed of response surfaces, whose general form and characteristics are controlled by the user, for testing experimental design methodologies. We discuss how such a test bed might be used to determine the best strategy for solving an industrial problem that arose in the formulation of automatic transmission fluid. Implementation of the test bed into computer code is left for future research.