Quality Engineering vol. 24 issue 1 - January 2012
Abstract: When an experiment involves both mixture and process variables, it can be necessary to restrict the randomization of experimental runs to reduce the size of the experiment. This results in a split-plot structure. If some process variables are noise variables, it is important to find a parameter setting that is robust to the variability of those factors. A real example of a mixture-process experiment within a split-plot structure is presented. The example is used to show that a dual optimization using both a mean model and a variance model can find the settings that are robust to the noise variables.
Keywords: Design of experiments (DOE); Mixture design; Noise; Optimal design; Robust design; Split-plot design
(0) Member Reviews