Outliers effect in measurement data for T-peel adhesion test using Robust parameter design

Journal article, 2014

As many researches focused on application of robust design engineering in practical case study, very less concerned on the criticality to data measurement system in parameter design. This paper will emphasize on the importance to be critical to data obtained during experiment. The existence of outliers is often ignored and the impact overlooked, thus endanger the results by producing false alarm and giving completely wrong parameter setting. The optimum condition from the data that contains outliers is compared with the corrected data measurement. The finding presents the indication procedure on how to confirm whether the data is reliable or not for evaluation. The data is unreliable when two main indicators are detected. Firstly, the measurement data plot detects outlier through linear regression analysis as it does not belong on the linear line. Secondly, poor reproducibility presented by estimation and confirmation of signal-to-noise ratio. This failure affects the experimental design and lead to wrong optimum condition. T-peel adhesion test using orthogonal array L9 is done as a case study to elucidate the detection of outlier and outlier effect on optimum condition.