This article details the advantages of design of experiments (DOE) over the OFAT (changing only one factor at a time) approach to experimentation. By varying factors at two levels each, but simultaneously rather than one at a time, experimenters can uncover important interactions.
This article provides an introduction to design of experiments (DOE) for improvement of coatings processes and formulations. It includes a case study on a spin-coater.
The first article, "Trimming the FAT out of Experimental Methods," presented arguments against one-factor-at-a-time (OFAT) techniques in favor of multifactor DOE. This follow-up article offers a case study that illustrates how a two-level factorial DOE can reveal a breakthrough interaction. A similar article appeared in OE (Optical Engineering) magazine.
This introductory article provides compelling reasons to abandon traditional scientific methods that deploy only one factor at a time (OFAT) in favor of multifactor testing techniques known as design of experiments (DOE). Only via DOE can experimenters detect interactions, which often prove to be the key to success.
This article introduces a new, more efficient type of fractional two-level factorial design of experiments (DOE) tailored for the screening of process factors. These designs are referred to as Min Res IV.
A basic primer, taken from "DOE Simplified" text, on underlying statistics and simple forms of DOE.
For many central composite designs (CCDs), particularly large ones, the usual alphas put the axial points outside the region of operability. A CCD with an alpha of one, known as a "face centered design" (FCD), avoids this problem by drawing the axial point back onto the face of the hyper cube. However, as the number of FCD factors increase, the correlation among the squared terms in the quadratic in the face-centered cube also increases. For k>5 this causes the variance inflation factors (VIFs) associated with the squared terms to become quite high. As a compromise between FCD and standard CCD, this white paper provides the case for a "practical" alpha of the fourth root of the number of factors (k). For k of 5 or more, this practical alpha balances statistical properties with operational necessities.
This article deals with thorny issues that confront every experimenter how to handle individual results that do not appear to fit with the rest of the data. (A somewhat modified version of this article was published in Quality Engineering. April 2007.)
Failures in the processing and use of products can often be prevented by applying a form of DOE called ruggedness testing. Check out this article to see how it's done for machine-made bread.