Distinguishing Cpk and Ppk in Minitab Process Capability Analysis

The Core Problem: Potential vs. Actual Performance

When analyzing a manufacturing process, a common mistake is relying on a single capability index to determine if a process is "capable." This leads to a critical blind spot: a process may have a tight distribution (high precision) but be centered far from the target (low accuracy). To solve this, quality engineers must distinguish between Cpk (Process Capability Index) and Ppk (Process Performance Index).

The takeaway is simple: Use Cpk to understand what your process is capable of doing if it remains stable and centered. Use Ppk to understand how the process actually performed over a specific period, including all shifts, drifts, and instabilities.

How Minitab Calculates Capability

Minitab calculates these indices by comparing the distance between the process mean and the nearest specification limit (Upper Specification Limit or USL, and Lower Specification Limit or LSL) against the process spread (sigma).

• Cpk: Uses "within-subgroup" variation. It estimates the short-term capability by ignoring shifts between subgroups.
• Ppk: Uses the "overall" standard deviation of all data points. It captures the total variation, including long-term drift.

Worked Configuration Example

Consider a scenario where you are monitoring the diameter of a machined pin. The target is 10.00mm, with a USL of 10.05mm and an LSL of 9.95mm.

1. Data Setup: Enter your measurements in a single column (e.g., C1) in the Minitab worksheet. Ensure data is collected in subgroups (e.g., 5 samples every hour).
2. Execution: Navigate to Stat > Quality Tools > Capability Analysis > Normal.
3. Configuration:
   • Single Column: Select C1.
   • Subgroup size: Enter 5.
   • Lower spec: 9.95.
   • Upper spec: 10.05.

Interpreting the Results

Upon running the analysis, Minitab generates a capability histogram. Compare the resulting indices to diagnose the process state:

Scenario	Cpk Value	Ppk Value	Diagnostic Conclusion
Cpk ≈ Ppk > 1.33	1.45	1.42	Process is stable, centered, and capable.
Cpk > Ppk	1.50	0.80	Process is precise but unstable; significant drift or shifts are occurring.
Cpk < 1.33, Ppk < 1.33	0.70	0.65	Process is either too wide (high variation) or poorly centered.

Critical Limitations and Common Mistakes

Capability indices are mathematical summaries; they can be dangerously misleading if the underlying assumptions are ignored.

The Normality Requirement

Minitab's Normal capability analysis assumes the data follows a Gaussian (bell curve) distribution. If the data is skewed, the Cpk/Ppk values will be mathematically incorrect. Verification: Before running capability analysis, run Stat > Basic Statistics > Normality Test. If the p-value is < 0.05, you must use the Non-normal capability analysis option or transform the data.

Stability vs. Capability

A common error is calculating Cpk for a process that is not in statistical control. If a control chart (like an X-bar R chart) shows points outside the control limits, the process is unstable. In this state, Cpk is a theoretical value that does not predict future performance. Always verify stability via a control chart before reporting capability indices.

Confusing Cp with Cpk

While not the primary focus, Minitab also provides Cp. Cp only looks at the spread of the data relative to the specs, ignoring where the mean is located. A process can have a perfect Cp of 2.0 but a Cpk of 0 if the mean is exactly on the specification limit. Never report Cp alone as a measure of quality.

Practical Verification

To verify your Minitab output, check the "Overall Sigma" versus the "Within Sigma" in the session window. If the Overall Sigma is significantly larger than the Within Sigma, your Ppk will be lower than your Cpk, signaling that your process is shifting over time and requires investigation into the root cause of the instability.