|Thursday, February 20|
|PS1 Poster Session I & Opening Mixer||
Thu, Feb 20, 5:15 PM - 6:45 PM
1. Application of Binary Search Algorithm to Improve Response Surface Design (302769)William Doerr, FM Global Research
Sachin Mujawar, FM Global Research
*Liming Xu, FM Global Research
Keywords: Experiment design, response surface, binary search
This presentation describes a procedure to combine a binary search algorithm with methods of response surface design to reduce the number of required experiments. In many applications of response surface design, reduction in the number of experiments becomes imperative, which inspires this study to develop an approach to incorporate binary search with traditional response surface design. Traditional methods generally consist of a sequential factorial or fractional factorial designs in which initial experiments are used to approximate the response surface. The approximated response surface is then used to find the direction of change for the factors, mostly along the gradient of the approximated response surface or the path of steepest ascent. Therefore, the efficiency of traditional response surface design depends on the gradient of the response surface at the location of the initial design. Low gradient region of response surface may require a larger number of experiments than the limit restricted by resources. A binary search, which relies on subject-matter knowledge to determine initial range of search, can reduce the number of experiments required for many applications.