SHEWHART-TYPE CHARTS IN NONSTANDARD SITUATIONS

Standard textbooks on statistical process control generally offer simple procedures to construct control charts. In many situations the resulting charts are of limited use because the variance components of the process are not adequately taken into account. In this article, an example of this is discussed. The use of an analysis of variance model in constructing more effective control charts is described. Control charts for the mean and linear contrasts are developed, using different estimators of variability. Guidance with respect to construction and evaluation of the charts is provided. The charts are illustrated with the example, and the efficiency is evaluated.