Cp and Cpk in Simple Terms
Cp asks: would the process fit inside the tolerance if it were perfectly centred? Cpk asks: does it fit, given where it actually runs? If Cp is good but Cpk is low, the process is capable but off-centre, which is usually easier to fix than too much variation.
Both use within-subgroup (short-term) variation and assume the process is stable, which is shown on a control chart.
How It Works
- Cp = (USL minus LSL) divided by 6 sigma.
- Cpk = the smaller of (USL minus mean) and (mean minus LSL), divided by 3 sigma.
| Cpk | Typical interpretation |
|---|---|
| Below 1.0 | Not capable, parts outside tolerance are likely |
| 1.0 to 1.33 | Marginal, needs close control |
| 1.33 or more | Common target for ongoing production |
| 1.67 or more | Common target for special characteristics and launch |
Targets come from the customer. AIAG PPAP uses 1.67 for initial studies of special characteristics.
Example
A shaft has a tolerance of 20.00 plus or minus 0.06 mm. The process sigma is 0.01 mm, so Cp is 0.12 divided by 0.06, which is 2.0. The mean is 20.03, so Cpk is (20.06 minus 20.03) divided by 0.03, which is 1.0. Re-centring the tool offset to 20.00 raises Cpk to 2.0 without any other change.
Common Mistakes
- Calculating Cpk on an unstable process.
- Ignoring non-normal data.
- Using too few parts; 100 or more readings in 25 subgroups is a common minimum.
- Confusing Cpk with Ppk, which uses overall variation.
The OperaHive SPC software runs Cp/Cpk studies from shop-floor data. The SPC and Cpk guide covers the method in depth.