Response maps for peak process performance

Executive summary This is the third article of a series on design of experiments. The first publication provided tools for process breakthroughs via two-level factorial designs. 1 The second article illustrated how to re-formulate rubbers or plastics using powerful statistical methods for mixture design and analysis. 2 The authors now bring their focus back to process improvement and show how to hit the sweet spot of high yield of in-specification products made at lowest possible cost. The key is in-depth design of experiments aimed at producing statistically validated predictive models. Response maps made from these models point the way to pinnacles of process performance. Response surface methods are powerful optimization tools in the arsenal of statistical design of experiments. Before employing response surface methods, process engineers should take full advantage of a far simpler tool for design of experiments-two-level factorials, which can be very effective for screening the vital few factors (including interactions) from the trivial many that have no significant impact. See our first article for a case study on factorial design and, for more details, the book we wrote for nonstatisticians. 3 Assuming the potential for further financial gain, follow up the screening studies by doing an in-depth investigation of the surviving factors via response surface methods. Then generate a "response surface" map and move the process to the optimum location. This article provides a brief on RSM with applications to plastics and rubber.