Cp, Cpk, Pp, Ppk Process Capability

🏭

Industry Use

How these indices are applied in pharmaceutical manufacturing

Capability indices compare specification width to process spread. Cp and Cpk use the within-subgroup estimate σ̂ (from control-chart data) and describe short-term capability under statistical control; Pp and Ppk use the overall standard deviation s and describe delivered performance across all sources of variation. Cpk and Ppk additionally account for centering; the gap between the pairs (Cp vs Pp, Cpk vs Ppk) measures how much special-cause or between-run variation is costing.

Where this is used:

Process qualification (Stage 2, PPQ) and continued process verification (Stage 3) under the FDA process-validation lifecycle.
Trending capability of CQAs — tablet weight, assay, fill volume, hardness — over time.
Justifying sampling plans and release strategies from demonstrated capability.

∑

Calculation Explanation

Mathematical basis and regulatory context

Short-term (control-chart σ̂)

Cp = (USL − LSL) / (6·σ̂) · Cpk = min[ (USL − X̄)/(3·σ̂), (X̄ − LSL)/(3·σ̂) ]

Overall performance (total s)

Pp = (USL − LSL) / (6·s) · Ppk = min[ (USL − X̄)/(3·s), (X̄ − LSL)/(3·s) ]

Within-subgroup sigma estimate

σ̂ = MR̄ / d₂ (individuals, d₂ = 1.128) or R̄/d₂ , s̄/c₄ (subgrouped data)

Calculator scope: this calculator takes individuals data and always estimates σ̂ as MR̄/d₂ across the entered values. The R̄/d₂ and s̄/c₄ forms are shown for completeness — they apply to rationally subgrouped data, which this calculator does not accept; capability from subgrouped data should use the subgroup statistics from your X̄ chart.

From the normal model the indices convert to a sigma level (≈ 3 × the index at the nearer limit) and a theoretical fraction outside specification (ppm). Both are model statements — see Assumptions.

Thresholds: ≥ 1.33 general and ≥ 1.67 for critical characteristics are widely applied industry conventions, to be set in your quality system. The FDA process-validation guidance names no index and sets no numeric threshold — it asks for statistical evidence of a state of control. Capability is only meaningful for a process in statistical control: establish control (I-MR or X̄-R) first, or σ̂ is not interpretable.

FDA Process Validation (2011) ICH Q10 NIST e-Handbook §6.1.6 USP <1210>

📋

Variable Definitions

All input parameters and their meaning

Symbol	Variable	Units	Description
USL / LSL	Specification limits	product units	The registered acceptance range for the characteristic.
X̄	Process mean	product units	Mean of the observations in the study window.
σ̂	Within-subgroup standard deviation	product units	From control-chart statistics — common-cause variation only. Drives Cp/Cpk. This calculator computes MR̄/d₂ on individuals data (R̄/d₂ and s̄/c₄ apply to subgrouped data, not accepted here).
s	Overall standard deviation	product units	Sample SD of all data — includes between-run and special-cause variation. Drives Pp/Ppk.
d₂	Control-chart constant	dimensionless	1.128 for a moving range of 2 (individuals data); subgroup-size-dependent otherwise.
Cp / Cpk	Capability indices (result)	dimensionless	Potential (ignores centering) and actual (nearer-limit) short-term capability.
Pp / Ppk	Performance indices (result)	dimensionless	The same construction on overall s — delivered performance.
Sigma level / ppm	Normal-model conversions (result)	σ / per million	≈ 3 × nearer-limit index; theoretical fraction outside specification under normality.

📖

Step-by-Step Tutorial

Worked example: tablet weight capability

Scenario: tablet weight, USL = 510 mg, LSL = 490 mg. Fifty individuals from an I-MR chart: X̄ = 499.8 mg, overall s = 1.85 mg, MR̄ = 1.94 mg.

Estimate σ̂ from the control chart.

Calculation
σ̂ = MR̄ / d₂ = 1.94 / 1.128 = 1.72 mg
Compute Cp and Cpk.

Calculation
Cp = 20 / (6 × 1.72) = 1.94
Upper: (510 − 499.8)/(3 × 1.72) = 1.98 · Lower: (499.8 − 490)/(3 × 1.72) = 1.90
Cpk = min(1.98, 1.90) = 1.90
Compute Pp and Ppk.

Calculation
Pp = 20 / (6 × 1.85) = 1.80
Upper: 10.2 / 5.55 = 1.84 · Lower: 9.8 / 5.55 = 1.77
Ppk = min(1.84, 1.77) = 1.77
Convert to sigma level and theoretical ppm.

Calculation
Sigma level ≈ 3 × 1.77 = 5.3σ
Theoretical fraction outside specification ≈ 0.08 per million (normal model, overall s)
Interpret: both pairs comfortably exceed the common 1.33 convention; Cpk ≈ Ppk indicates little between-run drift, and Cpk slightly above Ppk is the expected direction. Report Cpk and Ppk together — a large gap is itself a finding (between-run variation).

✓ Cp = 1.94 · Cpk = 1.90 · Pp = 1.80 · Ppk = 1.77 · ≈ 5.3σ, ≈ 0.08 ppm — capable by the common conventions

⚠️

Assumptions and Limitations

What the indices presume — and what they do not cover

Normality: sigma level and ppm are normal-model statements; heavy tails or skew (impurities, particle counts) invalidate them — transform or use distribution-appropriate methods.
State of control first: without statistical control, σ̂ has no meaning and Cp/Cpk are not interpretable.
Sample size: n ≥ 30 recommended before quoting an index; small n makes indices unstable.
One characteristic per index; a batch is capable when every CQA is.

📚

References

Primary sources for this guide

FDA, Process Validation: General Principles and Practices (January 2011) — lifecycle stages: 1 Process Design, 2 Process Qualification, 3 Continued Process Verification. fda.gov
ICH Q10, Pharmaceutical Quality System. database.ich.org
NIST/SEMATECH e-Handbook of Statistical Methods, §6.1.6 (process capability). itl.nist.gov
USP General Chapter <1210>, Statistical Tools for Procedure Validation — statistical context.

PharmaCalc