Statistical tolerancing using designed experiments in a noisy environment

Koksal G., Fathi Y.

COMPUTERS & INDUSTRIAL ENGINEERING, vol.44, pp.515-526, 2003 (SCI-Expanded) identifier identifier


We consider the method of designed experiments for statistical tolerance analysis, and study the impact of experimental error on its results.-It is observed that presence of random error in the experiment environment (e.g. laboratory) could introduce bias in the moment estimators and increases their-respective variances. We propose adjustments to the method that would reduce the bias as well as the variance of these estimators. A numerical example is presented.