00427 a5a3b4846efe962a975cb55bde8e8b

431

Process Capability: Engineering and Statistical Issues

sound data collection scheme is essential to the estimation and interpretation of any process capability index. Capability assessments can be expensive and time consuming. Hence, simpler evaluations may be in order prior to performing a process capability or process performance assessment. Furthermore, if a process is in control, but does not meet reąuirements, an improvement program that goes beyond process capability assessments may be appropriate.

Mentch (1980) addresses these issues in his classic paper, ‘Manufacturing Process Quality Optimization Studies,” by elaborating four different types of process assessments. In increasing order of complexity, these are process performance checks, process performance evaluations, process capability studies and process improvement programs.

Process performance checks provide a snapshot of the process over a limited time frame. Only a limited amount of data, usually from historical records, are analyzed through a histogram or a normal probability plot. Typically insufficient data is collected for the fruitful use of control charts, and hence the State of control of the process is not evaluated. Process assessments are limited to establishing whether the process is on target and to provide a rough assessment of the process variability present. Interestingly, Mentch notes that this information is likely to be far morę useful than a foreman's guess or 'engineering judgement'. ”

The process performance evaluation requires a longer period of evaluation, preferably a month or longer, and like the process performance check is usually retrospective. Because of its duration it provides a morę complete assessment of the process, but does not provide an accurate indication of the potential capability the process can achieve.

A process capability study differs from the two studies previously discussed in that it is an active study of current production and a control chart is employed during the collection of the data. Mentch States that this study is conducted over a long period of time, which he interprets to mean at least 20 points on a control chart. Notę that total sample size from this study would then be the number of plotted points times the rational sample size chosen for the process. Since this study is conducted in an active modę, the variation measured over the sub-samples using an R- or S-chart permits assessment of the potential achievable process capability.

A process improvement program is directed at improving a process that is in control, but whose capability is judged inadeąuate. It is typically a team study that employs techniąues, such as Ishikawa diagrams, Pareto analysis and designed experiments, for evaluating possible improvements. While process capability can be guessed from a process improvement program, it is