Measurement System Analysis in a Production Environment with Multiple Test Parameters

• Published in: Quality engineering 2003-01, Vol.16 (2), p.297-306
• Main Author: Larsen, Greg A.
• Format: Article
• Language: English
• Subjects: Mathematical models; Measurement; Measurement system analysis; Multivariate simulation; Quality control; Reliability engineering; Simulation; Studies; Systems analysis; Test errors; Variance components
• ISSN: 0898-2112; 1532-4222
• DOI: 10.1081/QEN-120024019

The purpose of measurement system analysis (MSA) is to separate the variation among devices being measured from the error in the measurement system. The total measurement system error can be further decomposed into variance components associated with the measurement equipment and repeatability. An analysis of variance approach based on a variance component model is used to model the variables of interest. Once estimated, the variance components are used to compute various metrics, which quantify the adequacy of the measurement system for the application in which it is used. Confidence intervals computed on the variance components and metrics indicate the amount of precision in the estimates. The MSA is typically conducted on a single measurement variable with a single measurement instance. The aim of this paper is to extend the univariate single-instance case to a common manufacturing test scenario where multiple parameters are tested on each device with a sequence of tests, which may include retest and test and repair steps. The methods presented are illustrated with examples from an industrial application.