Gage Rr Percent⚠ unverified

Understanding the Parameters

• Gage R&R variation $GRR$ — the measurement system's own contribution, itself the combination of two parts: repeatability (the same operator measuring the same part with the same gauge gets slightly different readings — equipment variation) and reproducibility (different operators get different readings — appraiser variation). They combine in quadrature: $GRR = \sqrt{EV^2 + AV^2}$.

• Total variation $TV$ — all the observed variation, combining the true part-to-part variation ($PV$) with the gage variation: $TV = \sqrt{PV^2 + GRR^2}$. It is what you actually see in the data before separating signal (parts) from noise (gauge).

• The ratio $\times 100$ — the gauge's share of the total, as a percentage. Because both $GRR$ and $TV$ are standard-deviation-like spreads, the ratio is dimensionless. The AIAG guidelines interpret it: $< 10\%$ the measurement system is acceptable; $10$–$30\%$ may be acceptable depending on the application and cost; $> 30\%$ is unacceptable — the gauge cannot reliably distinguish parts.

• Signal vs noise. A high $\%\,GRR$ means measurement noise swamps real part differences, so process capability studies (Process Capability Cpk), tolerance decisions, and control charts built on that data are untrustworthy. Gage R&R is the gatekeeper that validates the measurement before the quality analysis begins.

Derivation (Approaching a Proof)

The method rests on the additivity of variances of independent sources. The variation observed when measuring parts has two independent origins — the parts genuinely differ, and the gauge adds scatter — so their variances add:

$$\sigma_{TV}^2 = \sigma_{PV}^2 + \sigma_{GRR}^2,$$

where $\sigma_{GRR}^2 = \sigma_{EV}^2 + \sigma_{AV}^2$ itself splits into equipment (repeatability) and appraiser (reproducibility) variances. Working in standard-deviation ("variation") units $GRR = \sigma_{GRR}$ and $TV = \sigma_{TV}$, the gauge's fractional contribution to the total spread is the ratio, reported as a percentage:

$$\%\,GRR = \frac{GRR}{TV}\times 100. \qquad\blacksquare$$

(Because variances add but standard deviations do not, note that a $\%\,GRR$ of $30\%$ in standard-deviation terms corresponds to only $\sim9\%$ of the variance — the metric is conventionally reported on the standard-deviation scale, which is stricter.) A related index is the number of distinct categories (ndc) $= 1.41\,(PV/GRR)$, the count of part groups the gauge can resolve.

Dimensional check. $GRR$ and $TV$ share the measurand's units, so the ratio is dimensionless and $\times 100$ makes it a percentage. $\checkmark$

History and Development

• Measurement systems analysis (MSA). Gage R&R is the core of MSA, formalised by the Automotive Industry Action Group (AIAG) as a required step in the automotive quality process (PPAP). Before trusting any process data, the measurement system that produced it must be shown to be capable.

• Repeatability and reproducibility. The distinction — equipment scatter vs operator-to-operator scatter — is diagnosed by structured studies (typically multiple operators measuring multiple parts multiple times), analysed by the average-and-range method or ANOVA to separate the variance components.

• The 10/30 rule. The $<10\%$ (good) / $>30\%$ (unacceptable) thresholds are industry conventions that make Gage R&R a pass/fail gate on measurement capability, complementing the process-capability metric Process Capability Cpk that assumes the measurements are trustworthy.

Related Concepts: Process Capability Cpk, Relative Uncertainty, Expanded Uncertainty, Coverage Factor, Normal Distribution PDF

Notes: Registry calculator gage-rr-percent (unverified). $\%\,GRR = (GRR/TV)\times100$ — correct as shipped. $GRR = \sqrt{EV^2+AV^2}$ (repeatability + reproducibility); $TV = \sqrt{PV^2+GRR^2}$. AIAG rule: $<10\%$ acceptable, $>30\%$ unacceptable. Reported on the standard-deviation scale. Returns $0$ if $TV\le0$.