What is the Process Capability Index?
The process capability index measures how well a manufacturing or business process can produce output within its specification limits. Cp describes the potential capability — how the spread of the process compares to the allowed tolerance — while Cpk describes the actual capability, accounting for whether the process is centered between the limits. These indices are core tools in Six Sigma and statistical process control (SPC).
How to use this calculator
Enter the Upper Specification Limit (USL), Lower Specification Limit (LSL), the measured process mean (\(\mu\)), and the process standard deviation (\(\sigma\)). The calculator returns Cp, Cpk, and the one-sided indices CPU and CPL. A higher value means a more capable process; values of 1.33 or above are commonly required, and 1.67+ is typical for critical characteristics.
The formula explained
Cp = (USL − LSL) / (6σ). The denominator 6σ represents the natural process spread (±3σ around the mean covers about 99.73% of output). Cpk takes the smaller of two one-sided ratios — CPU = (USL − μ)/(3σ) and CPL = (μ − LSL)/(3σ) — so an off-center process is penalized. When the process is perfectly centered, Cp equals Cpk.
$$\begin{gathered} C_p = \frac{\text{USL} - \text{LSL}}{6\,\sigma} \qquad C_{pk} = \min\!\left(C_{pu},\, C_{pl}\right) \\[1.5em] \text{where}\quad \left\{ \begin{aligned} C_{pu} &= \frac{\text{USL} - \mu}{3\,\sigma} \\[0.4em] C_{pl} &= \frac{\mu - \text{LSL}}{3\,\sigma} \end{aligned} \right. \end{gathered}$$
Worked example
Suppose USL = 10, LSL = 2, \(\mu = 6\), \(\sigma = 1\). Then $$C_p = \frac{10 - 2}{6 \times 1} = \frac{8}{6} = 1.333.$$ \(C_{pu} = \frac{10 - 6}{3} = 1.333\) and \(C_{pl} = \frac{6 - 2}{3} = 1.333\), so \(C_{pk} = \min(1.333,\, 1.333) = 1.333\). Because the process is centered, Cp and Cpk match.
Interpreting Your Cp and Cpk Results
Cp compares the width of the specification (USL − LSL) with the natural spread of the process (6σ). It tells you whether the process could fit inside the limits if it were perfectly centered. Cpk also accounts for where the mean sits: it is the smaller of CPU and CPL, so it measures how close the mean is to the nearest limit in units of 3σ.
The table below maps Cpk to an approximate sigma level (3 × Cpk) and the expected total defect rate in parts per million (PPM) for a centered, normally distributed process with two-sided specification limits.
| Cpk | Approx. sigma level (3 × Cpk) | Expected defects (PPM, two-sided, centered) | General reading |
|---|---|---|---|
| < 1.00 | < 3 | > 2,700 | Not capable |
| 1.00 | 3 | ≈ 2,700 | Marginal; no safety margin |
| 1.33 | 4 | ≈ 63 | Commonly cited minimum for an acceptable process |
| 1.50 | 4.5 | ≈ 6.8 | Good |
| 1.67 | 5 | ≈ 0.57 | Very good; often used for critical characteristics |
| 2.00 | 6 | ≈ 0.002 | Six Sigma level capability |
Acceptance targets vary by industry and customer. Values such as 1.33 or 1.67 are common conventions, not universal laws, so check your own customer or standard requirements.
Reading the Cp–Cpk gap
- Cp = Cpk: the process mean is at the center of the specification.
- Cp > Cpk: the process is off-center. The larger the gap, the larger the shift toward one limit. Recentering the mean can raise Cpk up to the value of Cp.
- Cpk < 1.00: the process is not capable of reliably meeting the specification.
- Cpk = 0: the mean sits exactly on a specification limit, so about half of the output falls outside it.
- Negative Cpk: the mean lies outside the specification limits, so more than half of the output is nonconforming.
- Low Cp: centering alone will not help. The variation (σ) must be reduced or the tolerance widened.
Assumptions behind the numbers
- The data are approximately normally distributed. For skewed or bounded data the PPM figures in the table can be badly wrong.
- The process is stable and in statistical control. Check with a control chart before trusting Cp/Cpk.
- Cp and Cpk use a short-term (within-subgroup) σ. Using the overall, long-term σ gives the performance indices Pp and Ppk, which are usually lower.
- The measurement system must be adequate, because gauge error inflates the observed σ.
Key Terms and Symbols
- USL (Upper Specification Limit)
- The largest value of a characteristic that the customer or design allows.
- LSL (Lower Specification Limit)
- The smallest allowed value.
- Process mean (μ)
- The average of the process output, estimated from sample data.
- Standard deviation (σ)
- A measure of process spread. Within-subgroup (short-term) σ reflects only common-cause variation inside rational subgroups and is used for Cp and Cpk. Overall (long-term) σ includes shifts and drift between subgroups and is used for Pp and Ppk.
- Cp
- Potential capability: \(C_p = \frac{USL - LSL}{6\sigma}\). It ignores centering.
- CPU
- Upper capability: \(C_{pu} = \frac{USL - \mu}{3\sigma}\).
- CPL
- Lower capability: \(C_{pl} = \frac{\mu - LSL}{3\sigma}\).
- Cpk
- Actual capability: \(C_{pk} = \min(C_{pu},\, C_{pl})\). It equals Cp only when the process is centered.
- Pp / Ppk
- Performance indices that use the same formulas as Cp and Cpk but with the overall long-term σ. They describe what the customer actually experiences over time.
- Specification limits vs control limits
- Specification limits come from customer or design requirements and apply to individual units. Control limits are calculated from process data (typically mean ± 3σ of the plotted statistic) and show whether the process is behaving consistently. They are unrelated and must not be mixed up.
- Statistical control
- A state in which only common-cause variation is present, so the process is predictable. Capability indices are meaningful only for a process in control.
- PPM / DPMO
- Parts per million or defects per million opportunities, the expected rate of nonconforming output.
- Sigma level
- The distance from the mean to the nearest specification limit in units of σ. It equals 3 × Cpk when no long-term shift is assumed. Some Six Sigma conventions add a 1.5σ long-term shift, which is why "6 sigma" is often quoted as 3.4 DPMO.
FAQ
What is a good Cpk value? Generally Cpk ≥ 1.33 is considered capable, and ≥ 1.67 is preferred for safety-critical features.
Why is Cpk lower than Cp? Cpk drops below Cp whenever the process mean drifts away from the center of the spec range. The further off-center, the larger the gap.
Can Cpk be negative? Yes — if the mean falls outside a specification limit, the corresponding one-sided index becomes negative, indicating the process produces out-of-spec parts on average.