Cpk

Cpk is a process capability index that quantifies how well a stable process can produce output within specified tolerance limits, taking into account both the spread of the data and how centered the process mean is between the specification limits.

What Cpk represents

Cpk compares the natural variation of a process (usually estimated using the process standard deviation) against the distance from the process mean to the nearest specification limit. It is typically used when both an upper specification limit (USL) and a lower specification limit (LSL) are defined.

In common form:

• Cpk = minimum of (CPU, CPL)
• CPU = (USL − mean) / (3 × standard deviation)
• CPL = (mean − LSL) / (3 × standard deviation)

A higher Cpk value indicates that, assuming a stable and approximately normal distribution, more of the process output is expected to fall within the specification limits. A Cpk close to zero suggests the process mean is near or outside at least one specification limit.

Use in manufacturing and regulated environments

In industrial and regulated manufacturing, Cpk is commonly used to:

• Assess if a process is capable of meeting customer or internal specifications before full-scale production.
• Monitor ongoing process performance as a quality and risk indicator, often alongside leading indicators like equipment condition or setup accuracy.
• Support decisions about process adjustments, equipment maintenance, or improvement projects.
• Provide quantitative evidence in PPAP, validation, or qualification activities, subject to applicable procedures.

Cpk values are often calculated from measurements captured in MES, SPC, LIMS, or quality systems and may be surfaced in dashboards as part of process performance or leading indicator panels.

Assumptions and limitations

Interpreting Cpk correctly depends on several conditions:

• Stable process: The process should be statistically stable over the period of data collection. Large shifts, trends, or seasonal effects reduce the usefulness of Cpk.
• Representative data: The sample should represent normal operating conditions, not just best-case or heavily screened data.
• Distribution shape: Cpk is most straightforward to interpret when the process data are approximately normally distributed.
• Specification clarity: Cpk is defined with respect to stated LSL and USL. It does not apply if only a target without limits is defined.

Cpk itself does not guarantee compliance and does not describe the underlying causes of variation. It is one metric in a broader control and quality management framework.

Common confusion

• Cpk vs. Cp: Cp considers only the process spread relative to the specification width and assumes the process is perfectly centered. Cpk also accounts for how far the mean is from the center, so Cpk is always less than or equal to Cp.
• Cpk vs. Ppk: Ppk is a performance index using overall (often long-term) variation, including between-lot or between-shift effects. Cpk uses within-process variation under stable conditions. Both can be reported, but they answer slightly different questions.
• Cpk vs. control limits: Cpk is based on specification limits (requirements). Control limits in SPC charts are based on process behavior. A process can be in statistical control with a low Cpk if it is stable but not capable of meeting specifications.

Relation to leading indicators

In many manufacturing systems, Cpk is treated as a lagging indicator because it is calculated from produced parts or batches. However, trending Cpk over shorter time windows or at critical process steps can be used in a more leading way, signaling emerging capability loss before quality issues appear in final inspection or customer returns.