Chapter 8
Process Experimentation with Two or More Factors
Many important phenomena depend, not on the operation of a single factor, but on the bringing together of two or sometimes more factors at the appropriate levels.
(Box, 1990, p. 365)
Overview
The design of experiments (DOE) is a vast topic – designed experiments are major tools for use in the improve phase of many Six Sigma projects. In the previous chapter experiments involving only a single factor of interest were considered. In this chapter experiments involving two or more factors of interest will be introduced. Much process improvement experimentation is done on a one-factor-at-a-time basis. Although such experimentation can lead to improvement, there is no doubt that multifactor experiments, in which factor levels are varied systematically according to a recognized design, can be much more informative. In particular, multifactor experiments can reveal the presence of important interactions between factors. Harnessing interaction effects can lead to dramatic process improvements.
Minitab can be of assistance both with the actual design of the experiment and with the display and analysis of the resulting data. Following a general introduction, in which the concept of interaction will be introduced, experiments in the 2k series, with k factors, each with two levels, will be considered. Screening experiments and fractional factorial experiments in the 2k−p series will be introduced, together with the concept of design ...