Abstract: Process Capability Indices (PCI) has been widely used as a means of summarizing process performance relative to set of specification limits. The proper use of process capability indices are based on some assumptions which may not be true always. Therefore, sometime whether the process capability indices can truly reflect the performance of a process is questionable. Most of PCIs, including Cp, Cpk, Cpm and Cpmk, neglect the changes in the shape of the distribution, which is an important indicator of problems in skewness-prone processes. Wright proposed a process capability index 'Cs' to detect shape changes in a process due to skewness by incorporating a penalty for skewness. In this paper, the effect of skewness on assessment of accuracy of Wright's capability index Cs is studied and comparison is made with Cp, Cpk, Cpm and Cpmk indices when the distribution of the quality characteristic (spring force) considered is skewed slightly. This paper also discusses how modelling the non normal data using statistical software and results were compared with other methods.
Keywords: Non-normal distribution, Process capability index, Skewness, Modeling non-normal data, normality
Recieved: 12.08.2014 Accepted: 26.11.2014 UDC: 54.061