Cumulative sum techniques: an industrial quality control tool

Abstract

Cusum control techniques have proved to be some of the most useful tools in statistical quality control. The simplicity of their method of application and their efficiency in detecting changes in quality level of products of continuous industrial processes show some advantages over their predecessor, the Shewhart control chart. This thesis is concerned with their theoretical as well as their practical aspects. On the theoretical aspect, detail of their development and the calculation of Average Run Length and the Run Length distribution with varying decision intervals and reference values are discussed. A Markov chain approach is regarded as a most suitable method for the calculation of Average Run Length and the Run Length distribution. As a practical application, a project from the petroleum industry regarding the variation of grease viscosity has been successfully carried out using Cusum techniques.