The Power of DOE: How to Increase Experimental Design Success and Avoid Pitfalls

Abstract

Personal empirical experience when lecturing and consulting shows that not only students, but also experienced engineers familiar with DOE, show much more interest in the modeling of a process than to statistical inference, neglecting attention to “boundary conditions” of the process. But exactly the observation of ancillary boundary conditions of experiments, such as minimizing Beta-risk and noise, is determinant for the efficient execution of an experimental design and the effective application of DOE derived models. This essay focuses attention to the must-dos in the DOE statistics approach in order to avoid research pitfalls by presenting a fail-proof 14-step approach when applying DOE modeling.