Design of experiments considering multiple engineering characteristics

In an industrial process, other than optimizing the mean of the response, the engineer is often also interested in minimizing the variance of the response. The problem is augmented when it is required to simultaneously optimize more than one response. This paper first highlights some technical and managerial issues faced in the multivariate problem, and briefly reviews the desirability and expected loss approaches. A simple and intuitive approach, which utilizes the criterion of nondominated designs, is then proposed and illustrated with an example. To offer some guidance to the engineer tasked with optimizing a multi-characteristic problem, a useful comparative summary of the discussed approaches is also presented.