What is Cpc:
Cpc stands for Control Process Capability. It identifies whether a control process is capable of measuring a characteristic correctly, i.e. whether the process dispersion related to repeatability and reproducibility is acceptable in relation to the tolerance interval of the characteristic. Cpc is measured by:
Cpc=\frac{\text{Tolerance interval}}{6*\text{dispersion}}
The Cpc is considered as:
- Good if Cpc < 3
- Limit if 1 < Cpc < 4
- Unacceptable if Cpc < 4
The Cpc is the foundation of any improvement process. It's hard to work with measurements if they don't mean much! But when using the R&R method, it's not uncommon to come across an unacceptable Cpc. So how should you react? Here are 4 steps to follow to solve a measurement capability problem:
Validate how the test was performed
The first step is to validate the way the test was carried out. You need to check that there are no outliers, that the operators were qualified, that no form faults have been included in the repeatability...
Search for sources of variability around the 5Ms
If the main source of dispersion is repeatability, we need to use a cause tree to identify all the sources of this variability around the 5Ms of the measurement process (Measuring Device, Medium, Method, Means, Manpower). We then seek to freeze as many parameters as possible, such as the place of measurement on the product, environmental conditions, etc., even if it means no longer being representative of the measurement in production.
If freezing these parameters restores an acceptable Cpc, then it's possible to investigate further to find the real cause of the dispersion, and eventually to restore a process capable of production.
Repeat measurements several times
In the event of a high repeatability error, even when the maximum parameter has been fixed, the situation can be improved by repeating the measurements several times and taking the average of several measurements. By averaging n measurements, Cpc is multiplied by \sqrt{n}.
⚠️Attention However, in this case, the same measurement procedure must be used in production as in the R&R test.
Observe handling differences between operators
In the event of high reproducibility error, it is generally necessary to observe the differences in handling between the different operators and to formalize the control procedure unequivocally. In such cases, operator training is essential.
