00422 Kd9d19b15fc678a197c7290d30e0b4a

426

Pignatiello & Ramberg

A “process capability study ” is a scientific systematic procedurę for determining the capability of a process by means of control charts, and if necessary changing the process to obtain a better capability. This procedurę is continued as long as may be necessary until the problem which prompted the study is solved ... In any case the term “process capability study” implies the solution of problems.

We see, therefore, that the process capability study includes using control charts, process improvement efforts via design and analysis of experiments to improve capability as well as other statistical tools for problem solving. The Handbook (page 61) later wams about the “wrong ways of estimating the process capability:”

Do not attempt to use a distribution without a control chart. A distribution will not give a reliable estimate of the process capability unless the data making up the distribution came from a controlled process.

and

Do not attempt to use ... past data without a control chart. Even a long and intimate knowledge of the job cannot show whether ... past data is anything like the real process capability.

Thus, before one considers which process capability index to use, a substantial effort must first be invested in studying and leaming about the process.

We now list sonie of the issues that must be addressed in a process capability study:

•    Plan the study, including team formation, the selection of a representative

time frame for data collection, the concept of what constitutes a rational sample and other logistical issues. The sampling process must be studied as well to assure that the sample selected from the process is representative of the population as a whole.

•    Collect the data, using a control chart for monitoring the data collection

process. The process needs to be shown to be stable. Special causes of variability must be identified.

•    Eliminate special causes of variability, and evaluate the stability of the

process. Sources of variability may include measurement variability,