Process Capability Evaluation for a Non-normally Distributed One

Abstract

The common process capability indices (PCIs) Cp, Cpk, Cpm are widely used in practice. The use of these PCIs is based on the assumptions that process is in control and its output should be normally distributed. In practice normality is not always fulfilled. Therefore, the use of common PCIs leads to erroneous in capability evaluation. In this paper, capability evaluation for non-normally distributed process is carried out in industrial environment with two approaches. The first includes transforming data to normally distribute by Box-Cox transforming method then using the common PCIs. This method failed to transform these data. The second approach includes the use non-normal percentile method with Burr XII distribution. This paper proves that the second approach is more effective in evaluating the capability of this process. Practical case is applied in the State Company for Electrical Industries (SCFEI) particularly in (Water Pump) factory and Minitab 16 Software is used to reduce the long calculation of statistical values and to plot control charts.

Keywords

• process capability
• non
• normally distributed data
• process capability index