Process Capable and Centered

The process is capable and centered — how do I lock the gains, detect early erosion, and safely reduce inspection cost without losing the customer signal?

1. Freeze the process recipe: parameters, tools, fixtures, program version.
2. Re-baseline SPC control limits from the capable window; retire the pre-improvement limits.
3. Choose a drift-sensitive rule (EWMA or CUSUM) for slow-moving characteristics.
4. Formalize the reaction plan against the new limits, including escalation triggers.
5. Model the impact of reduced sampling — do not simply halve frequency.
6. Get customer agreement for any sampling change on Key Characteristics.
7. Update Control Plan, PFMEA detection rating, and inspection standards.
8. Schedule the next capability confirmation at the next major PM or tool-life boundary.

• Control limits inherited from a wider, older process — alarms lose meaning.
• Sampling frequency halved without a formal risk assessment.
• Preventive maintenance interval extended informally after "good results".
• Operator complacency; reaction plan no longer executed when limits are crossed.
• Silent parameter drift after a program edit "cleanup".

• Capable-and-centered is a fragile state, not a permanent one. The first sign of erosion is usually a slow drift in the mean, not a spread increase.
• Reducing sampling too aggressively removes the very signal you need to detect erosion — the "reward" undoes the gain.
• EWMA and CUSUM detect small persistent shifts far earlier than Shewhart rules. Use them once the process is stable.
• Do not carry the pre-improvement control limits forward. They no longer describe the process and will trigger nuisance alarms.

• Do not extrapolate the capable state to sister lines or sister machines without their own confirmation study.
• Do not treat capability as permission to skip layered process audits.
• Do not report the capable Cpk from initial PPAP as ongoing production evidence — refresh it.

• Cpk vs Ppk: Convergence confirms long-term stability; divergence over months is the early warning.
• Cpk vs Cpm: On target-critical features (fits, balance, alignment), Cpm should also be tracked.
• Cpk vs SPC: Capability is the summary; SPC is the live signal that protects it.

1. Lock in the setup parameters in the Control Plan.
2. Use Xbar-R or I-MR charts to catch drift early.
3. Reduce sampling frequency cautiously after stability is proven.
4. Use the free Process Capability Calculator to compute Cp, Cpk, Pp, Ppk, sigma level, and predicted PPM from your measurement data — no sign-up required.

• Never widen control limits to reduce reaction-plan noise; find the drift cause instead.
• Never rely solely on capability indices — pair them with a live SPC rule.
• Never reduce sampling on a customer Key Characteristic without written customer agreement.

• ✓Cpk ≥ 1.33 and Cp − Cpk ≤ 0.10 sustained across ≥ 25 subgroups
• ✓SPC limits re-baselined from the capable window
• ✓Drift-sensitive rule (EWMA or CUSUM) in place
• ✓Recipe frozen and Control Plan updated
• ✓Customer agreement obtained for any sampling change
• ✓Next confirmation study scheduled at tool life / PM boundary

• Process condition: statistical stability is required.
• Distribution assumption: use a distribution model justified for the data.
• Confirm process stability and measurement-system adequacy before interpreting a capability index.
• Use a justified distribution model or non-normal method when the normal model is unsuitable.