McLean, D.,Felfli, Douha.2009-03-232009-03-2319931993Source: Masters Abstracts International, Volume: 33-02, page: 0584.9780315896727http://hdl.handle.net/10393/6799http://dx.doi.org/10.20381/ruor-15020Most standard control charting schemes for Statistical Process Control are based on the assumption that measurements of the product quality variable are independent within and between subgroups. When this assumption is violated, the performance of the control procedures deteriorates. While most studies in the recent literature have concentrated on resolving the problems associated with between-subgroup correlation, this thesis explores the effects of within-subgroup correlation. Such a situation occurs when multiple measurements of the same variable are taken from a single item and treated as a rational subgroup. Past attempts to use standard control charts resulted in numerous frustrating false alarms. In this study, various methods for determining the existence and magnitude of within-subgroup correlation were evaluated. Using computer simulations to generate run length distributions, it was found that the performance of the traditional X, Cumulative Sum (CUSUM) and Exponentially Weighted Moving Average (EWMA) charts is very poor in the presence of within-subgroup correlation. Three alternate approaches were proposed and evaluated: the use of modified univariate charts, principal components charts and multivariate versions of the X, CUSUM and EWMA charts. Based on the comparison of the average and median run length performance of each chart, the use of a modified EWMA chart is recommended. The effectiveness of the proposed methods was demonstrated with applications to chemical processing steps in the semiconductor manufacturing industry.171 p.Engineering, Chemical.Thesis