MODIFICATION OF CLEMENTS’ METHOD FOR ASSESSING THE CAPABILITY OF A NON-NORMAL PROCESS WITH AN APPLICATION
Abstract
In many industries process capability studies are conducted in order to determine the capability of the process to produce acceptable products and it is one of the important activities of statistical process control. In order to express the capability of a process, process capability indices are frequently used. However, they are usually computed under the assumption that the process data follow a normal distribution. Normality assumption may be violated and the use of these traditional capability indices may cause misleading interpretation about the capability of the process. One of the most widely discussed methods to handle non-normality is Clements’ method which was proposed in 1989. Clements’ method uses the Pearson family of curves for calculating capability indices for any shape of distribution. It requires the estimation of the mean, standard deviation, skewness and kurtosis and makes use of the classical estimators of skewness and kurtosis. In this study, we discussed the use of more robust estimators of skewness and kurtosis in the calculation of process capability index by Clements’ method. For this purpose, capability indices with the use of these robust estimators are computed by simulation and the mean square errors of them are reported. The comparison is done through simulating Weibull and lognormal data with several different parameter values. Finally, a real life application to oil pump manufacturing in automotive industry is presented.
Keywords
References
- Box GEP, Cox DR. An analysis of transformations. J Roy Statist Soc B 1964; 26: 211-243.
- Johnson NL, Kotz S. Discrete Distributions. New York, NY, USA: John Wiley, 1969.
- Ryan TP. Statistical Methods for Quality Improvement. New York, NY, USA: John Wiley, 1989.
- Ryan TP, Schwertman NC. Optimal limits for attributes control charts. J Qual Technol 1997; 29: 86–96.
- Niaki STA, Abbasi B. Skewness reduction approach in multi-attribute process monitoring. Commun Stat - Theory Methods 2007; 36:12, 2313-2325.
- Clements JA. Process capability indices for non-normal calculations. Qual Prog 1989; 22: 49-55.
- Liu PH, Chen FL. Process capability analysis of non-normal process data using the Burr XII distribution. Int J Adv Manuf Technol 2006; 27: 975-984.
- Wu HH, Swain JJ, Farrington PA, Messimer SH. A weighted variance capability index for general non-normal processes. Qual Reliab Engng Int 1999; 15: 397-402.
- Choobineh F, Branting D. A simple approximation for semivariance. Eur J Oper Res 1986; 27: 364-370.
- Chang YS, Choi IS, Bai DS. Process capability indices for skewed populations. Qual Reliab Engng Int 2002; 18: 383-393.