SPC · Statistical process control
Separate common-cause variation from signals that require investigation.
What to prepare
Keep measurements in time order and define rational subgroups.
Method and steps
I-MR is used for individual observations; Xbar-R for subgroups. A P chart tracks proportions when the inspected count is known. Limits are estimated from a justified reference period.
- Define the decision and the criterion you are tracking.
- Check the data and the method assumptions.
- Calculate and review the chart together with its context.
- Document the limitation of the result and the next action.
Practical example
An observation beyond a control limit triggers a check of the context: setup, material and measurement. It is not automatically equivalent to a product outside specification.
Mistakes to avoid
A maximum of 15 observations is an exercise. It does not certify stability and does not replace the history required to estimate limits.
Applying it in QAPP
Start from the existing form and define how the data is validated. Link the result to the person responsible for checking it. During the pilot, compare data quality and reporting time with the initial situation; do not assume automation validates the analysis by itself.