ABSTRACT
During an ongoing industrial process, quality control is typically performed with target values derived from a standard “reference” sample whose units have been determined to be of acceptable quality on all the analyzed variables. However, at the initial stage of a new or changed process, a thorough investigation of its characteristics and capabilities is required. In a process capability study, there typically are no preexisting data on the process characteristics. In particular, no target values based on prior information from the production of the component are available, and the target values therefore have to be calculated internally. It is important to distinguish between product specification limits which are derived from customer needs, and process quality characteristics which depend upon the processes involved in the production and delivery of the product. Univariate process characteristics are typically given in terms of an average, estimating the process mean, and a standard deviation, estimating the process variability. In the multivariate case the vector of means replaces the univariate process mean anp the correlations between variables are added to the variables’ standard deviations or variances.
