What Is Process Capability Analysis?
Process capability analysis is a set of statistical techniques used to determine whether a manufacturing process is capable of consistently producing output within specification limits. In other words, it answers a fundamental question: can this process make good parts, and how confident can we be that it will keep making good parts over time?
Capability analysis is the bridge between process control (keeping the process stable) and process improvement (making the process better). A process that is in statistical control (stable and predictable) can still be incapable of meeting tight specifications. Conversely, a very capable process might still produce defects if it drifts off-target.
For quality engineers working with Chinese electronics manufacturers, capability metrics are essential for: qualifying new processes, setting realistic inspection AQL levels, comparing suppliers, and predicting long-term defect rates.
The Difference Between Cp and Cpk (and Pp/Ppk)
The four most common capability indices are often confused. Here is the key distinction:
- Cp (Process Capability) — measures the potential capability of a process if it were perfectly centered. It compares the spread of the process (6σ) to the width of the specification window (USL − LSL).
- Cpk (Process Capability Index) — measures the actual capability of the process as it is currently centered. It takes into account both spread and centering by using the minimum distance to either specification limit.
- Pp (Process Performance) — similar to Cp but uses overall (long-term) standard deviation instead of within-subgroup (short-term) standard deviation.
- Ppk (Process Performance Index) — similar to Cpk but uses overall standard deviation. This is the most realistic measure of actual long-term performance.
The "C" indices (Cp/Cpk) represent what the process could do if kept in perfect control (short-term, within-subgroup variation only). The "P" indices (Pp/Ppk) represent what the process actually does over time, including all sources of variation (shift-to-shift, tool wear, material lot changes, etc.).
How to Calculate Cp and Cpk
Cp Formula
Cp = (USL − LSL) / (6 × σ_within)
Where USL is the Upper Specification Limit, LSL is the Lower Specification Limit, and σ_within is the within-subgroup (short-term) standard deviation, typically estimated from an X-bar R chart using R-bar / d2.
Cpk Formula
Cpk = min(Cpu, Cpl)
Cpu = (USL − X-bar) / (3 × σ_within)
Cpl = (X-bar − LSL) / (3 × σ_within)
Where X-bar is the process mean. Cpu is the upper capability index (distance from mean to USL), and Cpl is the lower capability index (distance from mean to LSL). Cpk takes the smaller of the two, because the closer spec limit determines the true capability.
One-Sided Specifications
For specifications with only one limit (e.g., "strength must be at least X" or "defect rate must be below Y"), use only Cpu or Cpl as appropriate. Cp is not meaningful for one-sided specs.
Interpretation Benchmarks
Capability index values have widely accepted interpretations. These are industry-standard benchmarks from automotive and electronics quality programs:
- Cpk < 1.0 — Incapable: The process cannot consistently meet specifications. Expect significant defect rates. Requires immediate process improvement.
- Cpk = 1.0 — Barely capable: Approximately 0.27% out of spec (about 2700 ppm) if the process stays perfectly centered. Realistically, with normal drift, expect higher defect rates. Acceptable only for non-critical features.
- Cpk = 1.33 — Capable (common target): The "1.33" target comes from the Six Sigma 4σ level. A Cpk of 1.33 means the process spread uses only 75% of the specification width, leaving margin for drift. About 63 ppm out of spec at center — low enough for most applications.
- Cpk = 1.67 — Highly capable: Common target for critical features, safety components, and high-volume automated processes. Approximately 0.6 ppm out of spec at center. Provides excellent margin against drift.
- Cpk = 2.0 — Six Sigma level: The theoretical Six Sigma target. Even with a 1.5σ shift, defect rate remains below 3.4 ppm. Realistic for very stable automated processes with tight control.
For China manufacturing contexts, a Cpk of 1.33 is a reasonable minimum target for most consumer products. Critical or safety-related components should aim for 1.67 or higher.
X-bar R Control Charts and How They Relate to Capability
Control charts and capability analysis are two sides of the same coin. Before you can meaningfully calculate Cp/Cpk, the process must be in statistical control — meaning only common cause variation is present, with no special causes (out-of-control points, trends, shifts, or patterns).
Here is how the two tools work together:
- Collect data in rational subgroups (typically 3–5 consecutive parts per subgroup).
- Plot the subgroup averages on an X-bar chart and the subgroup ranges on an R chart.
- Calculate control limits and evaluate stability. If the process is not stable, find and eliminate special causes first.
- Once stable, use R-bar / d2 to estimate σ_within (the within-subgroup standard deviation).
- Use that σ_within to calculate Cp and Cpk.
Calculating capability on an unstable process gives misleading results — the indices will look better or worse than reality, and you cannot trust the prediction of future performance.
When to Use Pp/Ppk vs Cp/Cpk
Use Cp/Cpk when:
- The process is in statistical control
- You want to estimate the process's inherent or potential capability
- You are comparing machine capability (short-term)
- You have a recent X-bar R chart demonstrating stability
Use Pp/Ppk when:
- You do not know if the process is in control
- You want to measure actual real-world performance over a longer period
- You are doing a process capability study during a production run
- You need to report actual defect performance to customers
AIAG guidelines recommend reporting both sets of indices in a formal capability study. The gap between Cpk and Ppk tells you how much additional variation is coming from longer-term sources (tool wear, material changes, shift differences).
Practical Example from Electronics Manufacturing
Suppose you are evaluating a surface mount technology (SMT) line that places 0402-size resistors. The target placement position is 1.000 mm from the pad edge, with a tolerance of ±0.050 mm (so LSL = 0.950 mm, USL = 1.050 mm).
You collect 25 subgroups of 5 parts each. The overall average (X-double-bar) is 1.005 mm, and the average range (R-bar) is 0.012 mm. For a subgroup size of 5, d2 = 2.326.
σ_within = R-bar / d2 = 0.012 / 2.326 = 0.00516 mm
Cp = (1.050 − 0.950) / (6 × 0.00516) = 0.100 / 0.03096 = 3.23
Cpu = (1.050 − 1.005) / (3 × 0.00516) = 0.045 / 0.01548 = 2.91
Cpl = (1.005 − 0.950) / (3 × 0.00516) = 0.055 / 0.01548 = 3.55
Cpk = min(2.91, 3.55) = 2.91
The process is highly capable (Cpk = 2.91) and slightly offset toward the upper spec. The Cp of 3.23 tells you the process is more than capable — the gap between Cp and Cpk suggests the process could be even better if centered. For this non-safety SMT placement, the capability is excellent.
Download Our CPK / Control Chart Template
Download our CPK / Control Chart Template with automatic Cp/Cpk calculations and X-bar R charts. Get the CPK Template