Cp & Cpk – Interactive Process Capability

Understand process capability visually, not just mathematically. Explore how process spread and centring affect Cp and Cpk, then use the calculator to relate the visual model to real specification limits and process data.

Cp = potential capability Cpk = actual centred capability Visual + numerical learning

1.33

Cp · spread capability

1.33

Cpk · actual capability

Centred

Process centring

Capable

Illustrative interpretation

The simplest way to think about Cp and Cpk

Specification window vs process behaviour

Cp asks: “Is the natural process spread narrow enough to fit inside the specification?”

It measures potential capability. It does not care whether the process is centred.

Cpk asks: “Given where the process is actually running, how safely does it fit inside the specification?”

It reflects both spread and centring.

Cp = Cpk

The process is centred. Capability is being limited mainly by process variation.

Cpk < Cp

The process has shifted away from the specification centre. The bigger the gap, the greater the centring problem.

Cp < 1

The natural process spread is wider than the specification window. Recentring alone cannot solve it.

High Cp, low Cpk

The process could be very capable, but it is running too close to one specification limit.

Train & tunnel analogy

Think of the specification as the clear opening of a railway tunnel and the process as a train approaching it head-on. The train width represents the natural process variation (6σ), while the train's horizontal position represents the process mean. A narrow train entering through the centre has good Cp and Cpk. A narrow train approaching one side still has good Cp, but Cpk falls because the clearance to the nearest tunnel wall is reduced.

Typical reference values

Index	Typical interpretation
< 1.00	Process variation/centring does not fit comfortably within specification.
1.00	Natural 6σ spread is just equal to the tolerance width if centred; effectively no statistical margin.
1.33	Common minimum capability target in many industrial applications.
1.67	Often used for tighter or more critical capability expectations.
2.00	Very capable process, assuming stability and a suitable distribution model.

Important before trusting Cp/Cpk

The process should be statistically stable; capability does not replace control-chart thinking.
The measurement system must be adequate for the tolerance being studied.
The distribution assumption should be appropriate; strongly non-normal data may need different analysis.
Use specification limits, not control limits.
Capability should be calculated on a meaningful, representative data set.

Key point: a impressive Cpk number is not proof that the process is controlled, causal mechanisms are understood, or future output will remain capable.

Interactive Cp / Cpk Simulator

Move the sliders

Cp · process spread capability1.33

Higher Cp = narrower natural process spread relative to the tolerance.

Cpk · actual capability1.33

Cpk cannot exceed Cp for the same two-sided capability study. Lower Cpk shifts the process toward one specification limit.

Shift direction

Live visualisation

Capable

Specification limitsSpecification centreProcess meanNatural process spread

Capability Calculator

Use real process inputs

Lower specification limit (LSL)

Upper specification limit (USL)

Process mean

Process standard deviation (σ)

Calculated result

1.67

Cp · potential capability

1.33

Cpk · actual capability

20%

Mean offset from specification centre as a share of half-tolerance

What is happening mathematically?

You do not need the formulas to understand the visual, but the calculator uses the standard two-sided capability relationships. Cp compares total specification width with the natural 6σ process width. Cpk then looks at the smaller safety margin between the process mean and either specification limit. That is why Cpk falls when the process shifts off-centre.

When Cp is good but Cpk is poor

Focus on centring: setup, target setting, tool offsets, control strategy, drift, material effects or other factors moving the process mean.

When Cp and Cpk are both poor

Reducing variation is usually the priority. Recentring may help Cpk, but it cannot fix an inherently wide process.

When Cp ≈ Cpk

The process is reasonably centred. The capability index is then mainly telling you about process variation relative to tolerance.

Cp/Cpk vs Pp/Ppk

Cp/Cpk are commonly used to describe potential and centred capability based on within-process variation. Pp/Ppk use overall observed variation and are often used as performance indices. The exact interpretation depends on your statistical method, software and organisational/customer requirements.

Practical rule: do not compare capability numbers unless you understand how the standard deviation was estimated, the data subgrouping, the time period, and whether the process was stable.

Common mistakes

Using Cp/Cpk before confirming process stability.
Confusing specification limits with control limits.
Using too little data or a non-representative short run.
Ignoring measurement-system variation.
Assuming a high Cpk proves defect-free output.
Applying normal-distribution capability blindly to non-normal data.
Quoting only Cpk and hiding a much higher Cp, which can conceal a centring problem.

Suggested capability review workflow

Confirm specification and measurement system

Check stability and data suitability

Review Cp and Cpk together

Act on spread, centring, or both

Important engineering note

Capability thresholds such as 1.33 or 1.67 are common conventions, not universal acceptance criteria. Always apply the controlling drawing, contract, customer requirement, industry standard, control plan or organisational procedure.