Mr. James Myers September 6, 1996
Process Builder, Inc.
4508 S.W. Massachusetts St.
Seattle, WA 98116
Dear Jim,
I am pleased to send you this letter describing my experience with Design of Experiments (DOE) techniques. I took Dave Doehlert's training course in January 1996. Nearly every lab associate at Desotech has now had this training. Since my background is in chemistry, I was initially apprehensive about a four day course in statistics. However, Dave's teaching style, which included numerous sample problems with an emphasis on application of the techniques was excellent.
I ordered my first design from Dave in late January almost immediately after I completed his training. This was a six component mixture design. This problem was a "hybrid photopolymer system" which relies on two simultaneous curing mechanisms. Because of the nature of this system, it was impossible to predict the behavior based on experience alone. It took us about a month to complete the 24 runs which included measuring seven responses. The data was put into Dave's Strategy software which ran great on the first try. The resulting contour plots immediately helped us understand the complex nature of this system. The Gridsearch feature showed three distinct sweet spots in the design. Two of the sweet spots were selected and the compositions were prepared. The agreement between the predicted and measured properties was excellent. Also, the test results were clearly better than the competitive technology.
We evaluated our sweet spot compositions with our customer in Europe in late March. During the evaluation we found that the target for one of the responses needed to be changed. I was able to pull up the appropriate contour plot on my lap-top PC and see what factor needed to be changed. I was also able to confirm that the other responses would not be adversely affected by the change. I called back to our lab in the US and they confirmed that the change had exactly the predicted effect. The modified sample was sent by air freight in two days.
The compositions developed in this study were unique, and a US Patent application was filed in early May. The product is now in a scale-up testing phase. We expect significant sales of this product in the next 12 months. This is one of the fastest development cycles I have experienced, and it would have not been possible without DOE.
Several additional DOE projects are now in progress at Desotech including one 10 factor (63 run) problem with 13 responses. The study will result in the optimization of a very important Desotech product. I am convinced that DOE is an essential tool in our business.
Sincerely,
Stephen C. Lapin, Ph.D.
Technical Team Leader
Process Builder DOE Support Home Page
Suggestions and complaints should be sent to abcorwin@processbuilder.com