Application Of Univariate And Multivariate Process Control Procedures In Industry

Traditional statistical process control charts used to monitor key process variables are based on the assumption that measurements are independent and identically distributed about a target value. In practice they are not and often are actually correlated. Reliance on univariate charts can lead to misleading conclusions. This paper addresses the methods for improving the quality of industrial products using T Multivariate Quality Control charts. In practice, one of the main problems in implementing the T 2 multivariate process control chart is that it only identifies the sample that causes the out-of-control situation. However, T 2 control chart is unable to identify the quality characteristic(s) that caused the out-of-control signal which is regarded as a major disadvantage in the implementation of this control chart. In this paper, Murphy’s method is implemented which not only identifies the sample, but also selects and prioritise the out-of-control variables using the T 2 control procedures. The results are compared with the performance of the individual charts. The practical example using the data provided by ESSAR Steel Limited, clearly shows the superiority of multivariate control charts over the univariate charts.