What is the difference between Cpk and Ppk?

Both use the distance from the mean to the nearer specification limit divided by three sigma. Cpk uses the within-subgroup sigma (R-bar / d2), the short-term variation; Ppk uses the overall standard deviation of every reading, which includes drift between subgroups. With a mean 0.4 above a lower limit of 10 and a within sigma of 0.09, Cpk is 0.4 / 0.27 = 1.48; if the overall sigma is 0.12, Ppk is 1.11.

Which Cpk or Ppk value do I need?

The one your customer specifies. Many automotive customers follow the AIAG PPAP manual, which asks for an index of 1.67 on initial process studies and 1.33 on ongoing production, but customer-specific requirements vary. The template does not assume a value: type yours on Setup and the tiles show On target, Near target or Below target against it.

How many readings does a capability study need?

A common practice is 25 subgroups of 4 or 5 consecutive parts, about 100 to 125 readings, from a stable process. Fewer readings give unstable estimates: with 125 readings an observed Cpk of 1.33 has a 95 percent lower confidence bound near 1.15, and with 30 readings the bound is much lower.

Why are observed ppm and expected ppm different?

Observed ppm counts the readings that actually fell outside the specification. Expected ppm assumes the readings follow a normal distribution with the overall mean and sigma and integrates the tails beyond each limit. With few or no readings out of spec, observed ppm is often 0 while expected ppm is small but positive; for a non-normal characteristic the expected figure can be badly wrong, so it is labelled as an estimate.

Can I use it for a one-sided specification such as runout or flatness?

Only partly. The workbook expects both limits. For a one-sided upper limit, read the upper-side indices (CPU and PPU) and ignore Cp, Pp and the lower side. Characteristics bounded at zero are usually skewed, so the normal-model indices and expected ppm are approximate.

Do I need a control chart as well?

Yes. Capability indices only predict future output when the process is stable. Plot the same subgroups on an X-bar and R chart first; if points fall outside the control limits, find and remove the special causes before you quote Cpk or Ppk.

Quality Systems and Compliance

Process Capability Cp Cpk Pp Ppk Excel Template

Enter a subgrouped capability study and get Cp, Cpk, Pp and Ppk against your customer's requirement, a histogram against the specification and ppm out of spec.

Overview

This process capability template is a working Excel workbook for quality engineers, supplier quality engineers and machine shops that need Cp, Cpk, Pp and Ppk from real study data. It ships with an example study of 25 subgroups of five readings for a fictional ground shaft, so the tiles, histogram and charts are populated when you open it; replace the example rows with your own and everything recalculates.

The workbook estimates two sigmas from the same readings. The within sigma is the average subgroup range divided by d2, the short-term variation between consecutive parts, and gives Cp and Cpk. The overall sigma is the sample standard deviation of every reading, which also contains drift between subgroups, and gives Pp and Ppk. Each index is shown with both sides (upper and lower), and the workbook counts the readings outside the specification as observed ppm next to an expected ppm from a normal model.

In the example, the shaft runs about 0.0045 mm above mid-tolerance and wanders between wheel dressings: Cpk meets the 1.33 requirement but Ppk misses the 1.67 asked for an initial study, and the histogram shows why. Centring the process would raise both. The Method sheet states the assumptions (a stable, roughly normal process) and what the workbook does not do, such as confidence intervals and non-normal capability.

What this template includes

Measurements table: subgroup number, date and up to five readings per subgroup, with mean, range and readings out of spec per row
Within sigma from R-bar / d2 and overall sigma from every reading, so capability (Cp, Cpk) and performance (Pp, Ppk) sit side by side
Each side of Cpk and Ppk (CPU, CPL, PPU, PPL), observed ppm out of spec and expected ppm from a normal model, labelled as an estimate
Dashboard with Cp, Cpk, Pp and Ppk tiles against the required indices you set, a histogram with out-of-spec bins in a second colour and the fitted normal curve, and an indices chart
Check column that names problems in words: reading count not n, duplicate subgroup, reading outside spec
Method sheet with every formula and its source, and a Verify sheet that re-checks the formulas against an independent reference (Self-check: PASS)
CSV with live formulas for the measurements table; the dashboard, charts, dropdowns and self-check are in the .xlsx

Suggested use case

Use it for a PPAP or first-production capability study, to answer a customer's request for Cpk evidence, or to check whether a process can hold a tighter tolerance before you quote it.

How to use it

On Setup, name the characteristic and unit, enter the lower and upper specification limits, the subgroup size and your customer's required Cpk and Ppk.
On Measurements, replace the example rows with your study: one row per subgroup of consecutive parts, in production order.
Fix anything the Check column names, and confirm the process is stable on a control chart before you quote the indices.
Read the Dashboard: the tiles give Cp, Cpk, Pp and Ppk with a status word against the requirement, and the histogram shows where the parts sit in the tolerance.
Compare Cpk with Ppk: a large gap means the process drifts between subgroups; a gap between Cp and Cpk means it runs off-centre.

Frequently Asked Questions

What is the difference between Cpk and Ppk?: Both use the distance from the mean to the nearer specification limit divided by three sigma. Cpk uses the within-subgroup sigma (R-bar / d2), the short-term variation; Ppk uses the overall standard deviation of every reading, which includes drift between subgroups. With a mean 0.4 above a lower limit of 10 and a within sigma of 0.09, Cpk is 0.4 / 0.27 = 1.48; if the overall sigma is 0.12, Ppk is 1.11.
Which Cpk or Ppk value do I need?: The one your customer specifies. Many automotive customers follow the AIAG PPAP manual, which asks for an index of 1.67 on initial process studies and 1.33 on ongoing production, but customer-specific requirements vary. The template does not assume a value: type yours on Setup and the tiles show On target, Near target or Below target against it.
How many readings does a capability study need?: A common practice is 25 subgroups of 4 or 5 consecutive parts, about 100 to 125 readings, from a stable process. Fewer readings give unstable estimates: with 125 readings an observed Cpk of 1.33 has a 95 percent lower confidence bound near 1.15, and with 30 readings the bound is much lower.
Why are observed ppm and expected ppm different?: Observed ppm counts the readings that actually fell outside the specification. Expected ppm assumes the readings follow a normal distribution with the overall mean and sigma and integrates the tails beyond each limit. With few or no readings out of spec, observed ppm is often 0 while expected ppm is small but positive; for a non-normal characteristic the expected figure can be badly wrong, so it is labelled as an estimate.
Can I use it for a one-sided specification such as runout or flatness?: Only partly. The workbook expects both limits. For a one-sided upper limit, read the upper-side indices (CPU and PPU) and ignore Cp, Pp and the lower side. Characteristics bounded at zero are usually skewed, so the normal-model indices and expected ppm are approximate.
Do I need a control chart as well?: Yes. Capability indices only predict future output when the process is stable. Plot the same subgroups on an X-bar and R chart first; if points fall outside the control limits, find and remove the special causes before you quote Cpk or Ppk.