The Impacts of Errors in Factor Levels on Robust Parameter Design Optimization

In robust parameter design (RPD), the ultimate goal is to identify the settings of control factors, which lead to an optimal mean with minimum process variation. In order to achieve this goal, usually two objective functions corresponding to the mean and variance of the desired quality characteristic are considered. Next, settings for the control variables (factors) are determined such that the values achieved for the two objective functions are as close to their ideal values as possible. This article highlights the impact of the miss‐specification of noise variables as fixed factors in RPDs. The miss‐specification or error in factor levels causes inappropriate estimates of the response model, which consequently affects the optimal settings of the control variables. The results are illustrated through an experimental example. Moreover, three different formulations are applied to determine the optimal settings for the case of Larger The Better (LTB). The performance of the formulations is also evaluated.