Exclusive Content & Downloads from ASQ
Optimal Three-Level Designs for Response Surfaces in Spherical Experimental Regions
Summary: [This abstract is based on the author's abstract.] Because of their simplicity and efficiency, Box-Behnken designs have been widely used to estimate a second-order model in three or four factors, but as the number of factors increases, the run size of these designs increases, making them less useful. It is recommended in this article that D-optimal and I-optimal three-level designs be used for spherical design regions involving three or more factors. An optimal design algorithm offers flexibility of run size and permits construction of second-order designs for more complex response surface applications. The restriction to three-level designs offers little loss in efficiency, provided the design is not nearly saturated.
- Topics: Design of Experiments
- Keywords: Box plots, Computer plotting, Optimal design, D-optimality, Mixture design, Second-order designs, Split-plot design
- Author: Mee, Robert W.
- Journal: Journal of Quality Technology