Process Capability Calculator
Cp, Cpk, Pp and Ppk from your own measurements, with the parts most capability calculators leave out. A Cpk from a small sample is a range, not a number, and this page shows the range. It also checks whether the bell curve behind the figures fits your data.
Cpk means nothing if the process is not stable, so look at a control chart of the same data first. When Cpk and Ppk disagree, the process is moving between subgroups, and that is the finding.
Nothing you type here leaves your browser. Every calculation runs in this page. The page itself counts visits, like every page on this site, but never what you enter.
1. Your measurements and the specification
Leave blank for an upper limit only
Leave blank for a lower limit only
1 for one part at a time, up to 10
Separate values with commas, spaces, tabs or new lines. A column pasted from a spreadsheet works. Example: a shaft turned to 25.00 mm ± 0.05 mm, invented.
Histogram, with the spec limits, the mean and the fitted normal curves
2. Your Cpk, and how sure you can be of it
Cpk, approximate 95% confidence interval
Dashed lines mark 1.00 and 1.33, the two thresholds most customers ask about.
3. Cpk against Ppk: is the process moving?
4. Parts outside the specification
5. Does the bell curve fit your data?
The arithmetic, with your numbers
Shown so you can check it, or redo it in a spreadsheet. The spreadsheet formulas assume your measurements are in A2:A41, one per row.
What this does not tell you
- It assumes the process was stable while you measured. Capability describes a process that behaves the same way from one hour to the next. If the process shifted partway through, the figures describe neither half. Plot the data on an individuals or X-bar and R control chart first, and only quote a Cpk once the chart shows no out-of-control signals.
- The order of the data matters. The within-subgroup spread comes from neighbouring values. Paste the measurements in the order they were made. If you sort them first, the moving ranges shrink and Cpk comes out far too high.
- The expected parts per million lean entirely on the bell curve. At a Cpk of 1.33 the predicted defects sit four standard deviations from the mean, far beyond anything a sample of 40 or 100 parts has seen. A small departure from normal in the tails can move that figure by a factor of ten. The normality check catches gross departures, not subtle ones.
- The interval is an approximation. Bissell's formula assumes normal data and works best above about 25 measurements. It also treats the standard deviation as if it came from all n values, which is close for the overall figure and a little optimistic for the within-subgroup one.
- It cannot see measurement error. Every measurement includes the gauge's own spread. If the gauge has not passed a repeatability and reproducibility study, part of what this page calls process variation is the gauge, and Cpk is understated.
- A 95% interval is wrong about one time in twenty. It is a statement about the method, not a guarantee about your number.
The rest of the Lean Six Sigma toolkit
- Control Chart Maker: I-MR, c and p Charts with Run Rules: I-MR, c and p charts with the four Western Electric run rules.
- Sigma Level Calculator: DPMO, yield and how sure you can be: DPMO and sigma level with the 95% interval, and whether a kaizen really cut defects.
- Takt Time and Line Balance Calculator: Takt time, line balance and the bottleneck, with stations that miss takt on slow cycles.
- OEE Calculator: Availability, Performance, Quality Losses: OEE with each loss in minutes, and the one fix worth most.