Gage R&R calculator
Enter your operator × part × trial readings and get %GRR, the equipment and appraiser variation (EV / AV), part variation (PV), the number of distinct categories (NDC), and a traffic-light verdict. Uses the AIAG MSA average-and-range method. Everything runs in your browser.
Study setup
Full tolerance width (USL − LSL). Adds %GRR against tolerance.
If known, total variation is built from this sigma instead of the parts in the study.
Measurements
One reading per operator, part, and trial. Tab through the grid; results update live.
Result
Method: AIAG MSA 4th ed. average-and-range (X-bar & R). Variations reported as % study variation, i.e. the 6σ (6-standard-deviation) study-variation convention, which equals the ratio of the sigmas. The average-and-range method does not estimate the operator × part interaction; ANOVA does.
Gage R&R, in plain terms
What gage R&R tells you
A measurement system analysis answers a simple question: when your numbers move, is that the parts changing or the measurement changing? A gage R&R study takes a handful of parts, has a few operators measure each one a few times, and splits the total variation into part-to-part variation (the real signal) and measurement variation (the noise). The noise itself is two things: repeatability, the equipment variation (EV) you get when one person measures the same part twice, and reproducibility, the appraiser variation (AV) you get when different people measure it. Add those in quadrature and you have the gage R&R. The headline, %GRR, is the share of the variation that comes from the gage rather than from the parts.
%GRR thresholds
AIAG reads the result against a familiar traffic light. Below 10% the measurement system is generally acceptable. Between 10% and 30% it is conditionally acceptable: whether you keep it depends on the application, the importance of the characteristic, the cost of the gage, and the cost of repair. Above 30% the system is unacceptable and needs work, whether that is a better gage, better fixturing, or operator training. These are guidelines rather than pass or fail lines drawn by a standard, so a safety-critical measurement may demand tighter than 10% and a rough screening check may live with more. You can read %GRR against the total variation of the parts you measured (does the system resolve these parts?) or against the tolerance (is it good enough to judge against the spec?); they answer different questions, so this tool shows both when you provide a tolerance.
NDC (number of distinct categories)
Number of distinct categories estimates how many non-overlapping groups of parts the measurement system can reliably tell apart across the range you studied, computed as floor(1.41 × σ_part / σ_GRR). AIAG recommends an NDC of 5 or more for a system used to control a process or sort parts. An NDC of 1 means the gage can barely separate parts at all; 2 or 3 means it can only bucket them coarsely (low, medium, high). NDC and %GRR usually agree, but NDC is the more intuitive way to say what resolution you actually have.
The average-and-range method (and its limit)
This calculator uses the AIAG MSA average-and-range (X-bar and R) method: the equipment variation comes from the average measurement range divided by a d2* constant, the appraiser variation from the spread of operator averages corrected for repeatability, and the part variation from the range of part averages. It is transparent and matches the worked examples in the AIAG manual. Its honest limitation is that it cannot estimate the operator-by-part interaction, whether some operators measure some parts differently from others. ANOVA can, and is the more rigorous method when you have software for it; for most studies the two agree on the verdict. References: AIAG Measurement Systems Analysis (MSA), 4th edition.
Common questions
-
What does a gage R&R study tell you?
A gage R&R (Measurement System Analysis) study separates the variation in your measurements into the part-to-part variation you actually want to see and the measurement system variation that is just noise. The noise has two parts: repeatability (equipment variation, EV) is the scatter when the same operator measures the same part again, and reproducibility (appraiser variation, AV) is the scatter between different operators. Together they are the gage R&R. The headline number, %GRR, tells you what share of the total observed variation comes from the measurement system rather than from the parts.
-
What are the %GRR thresholds (acceptable, marginal, unacceptable)?
The AIAG guideline reads %GRR against the total (or the tolerance): under 10% the measurement system is generally acceptable; 10% to 30% is conditionally acceptable depending on the application, the cost of the gage, and the cost of repair; over 30% the system is considered unacceptable and needs to be improved. These are guidelines, not hard limits. A safety-critical measurement may demand better than 10%, and a screening check may tolerate more than 30%.
-
What is NDC (number of distinct categories)?
NDC estimates how many distinct groups of parts the measurement system can reliably tell apart across the part range. It is computed as floor(1.41 x sigma_part / sigma_GRR). AIAG recommends NDC of 5 or more for a system used to control a process or sort parts. An NDC of 1 means the system can barely distinguish parts at all; 2 or 3 means it can only sort into coarse buckets such as low, medium, high.
-
What is the difference between %GRR against total variation and against tolerance?
%GRR against total variation (the % study variation) compares the gage R&R to the spread of the parts you measured, so it tells you whether the system can resolve the parts in front of it. %GRR against tolerance compares the gage R&R to the specification width, so it tells you whether the system is good enough to make pass or fail decisions against the spec. They answer different questions and can disagree. This calculator reports the study-variation figure always, and the tolerance figure when you enter a tolerance.
-
Is the average-and-range method as good as ANOVA?
The average-and-range (X-bar and R) method is the classic, transparent hand-calculation method and is what this tool uses. Its main limitation is that it cannot estimate the operator-by-part interaction (whether some operators measure some parts differently), and it can slightly differ from ANOVA on the same data. ANOVA is the more rigorous method, captures the interaction term, and is preferred when you have software for it. For most studies the two methods agree closely on the verdict.
-
How many operators, parts, and trials should I use?
The common AIAG study uses 3 operators, 10 parts, and 2 or 3 trials, with the parts chosen to span the normal range of production. Smaller studies (2 operators, 5 parts, 2 trials) are valid and faster but give a coarser estimate, so the d2* constants and the NDC are less stable. Pick parts that represent the real spread you measure; a study run on near-identical parts will overstate %GRR because there is little part variation to compare against.
This calculator applies the AIAG MSA average-and-range method to the readings you enter and reports % study variation, %GRR, and NDC. It does not estimate the operator-by-part interaction (use ANOVA for that), does not replace the AIAG Measurement Systems Analysis manual, and is not your quality system. Treat the result as a working estimate and validate it against your documented MSA procedure before acting on it.
Keep the measurement data with the record
Axiospec keeps every calibration, reading, and certificate in one tamper-evident ledger built for ISO/IEC 17025. See it on real calibration data, no signup.