A class of three-level experimental designs for response surface modeling

Most empirically constructed response surface models are based on polynomia ls containing terms of order 2 or less. Experimental designs involving thre e equally spaced levels of each factor are popular choices for collecting d ata to fit such models. Because complete three-level factorial plans requir e more experimental runs than can usually be accommodated in practice, smal ler designs are typically used. The families of three-level designs most of ten used in this context are the Box-Behnken plans and various forms of the central composite designs. This article introduces a different method for constructing composite designs, motivated by notions of sequential experime ntation and the minimax and maximin distance criteria used in spatial model ing. Operational and performance characteristics of some designs constructe d by the method are compared to those of competing Box-Behnken and central composite plans.