To run a gauge R&R study, pick 10 parts that span the real process range, 3 operators who normally use the gauge, and have each operator measure every part 3 times in random order without seeing earlier results. From the ranges you get equipment variation (repeatability). From the differences between operator averages you get appraiser variation (reproducibility). Combine them into GRR and compare it with total variation and with the tolerance. Under 10% is generally acceptable, 10–30% may be acceptable depending on the use, and over 30% means the gauge needs work.
Repeatability, or equipment variation (EV), is the spread you get when one person measures the same part with the same gauge several times. It comes from the gauge, the fixture, and how consistently the part sits in it. Reproducibility, or appraiser variation (AV), is the difference between people measuring the same parts. It comes from technique, reading habits, and unclear instructions.
GRR combines the two. The study then compares GRR with the spread of the parts (PV) and with the tolerance. Because variances add and standard deviations do not, the percentages never add up to 100. Their squares do.
The AIAG Measurement Systems Analysis manual describes a measurement system with GRR under 10% as generally acceptable. Between 10% and 30% it may be acceptable depending on how important the measurement is, what a better gauge would cost, and the customer’s view. Above 30% it is considered unacceptable and needs improvement. The same manual suggests the number of distinct categories (ndc) should be 5 or more, so the gauge can tell meaningfully different parts apart.
Use %tolerance when the gauge is used to accept or reject parts, and %total variation when it is used to control the process, for example on an SPC chart. Customers may set their own limits, so check their requirements. The calibration management guide covers where MSA fits in your gauge control system.
Every capability index includes measurement variation in its sigma. A noisy gauge makes a good process look marginal and can hide real shifts on a control chart. Run the gauge study first, then the capability study with Cpk in Excel or the process capability calculator. Cpk vs Ppk explains which index to report.
The average and range method in this guide is easy to do by hand or in a spreadsheet. The ANOVA method uses the same data and also separates out operator-by-part interaction, so it is preferred when software is available. The two methods usually give similar GRR when there is no strong interaction.
Example: A height gauge measuring a step of 6.00 ± 0.20 mm, so the tolerance is 0.40 mm (fictional data). 10 parts, 3 operators, 3 trials. Average ranges: operator A 0.024 mm, B 0.029 mm, C 0.022 mm. Operator averages: A 6.012, B 6.004, C 6.021 mm. Part averages range from 5.842 to 6.152 mm.
Result: GRR is 17.2% of total variation and 25.5% of tolerance, both in the 10–30% band, with ndc 8. The gauge may be acceptable for this characteristic depending on its importance and the customer. EV is larger than AV, so improvement should start with the gauge and fixture, for example a dedicated locating fixture, rather than operator training.
Both. “Gage” is the spelling used in the AIAG manuals and US automotive practice, and “gauge” is the general English spelling. They mean the same study.
It is the common layout and the one the K constants above assume. Smaller studies such as 5 parts × 2 operators × 2 trials are possible but give a much less reliable estimate. Use the constants for your layout.
The number of distinct categories: how many groups of parts the gauge can reliably tell apart within the process spread. It is 1.41 × PV ÷ GRR, rounded down. A value of 5 or more is the usual guideline.
Set AV to zero. It means the difference between operators is smaller than what repeatability alone would produce.
Calibration checks a gauge against a reference standard for accuracy. A gauge R&R checks the variation of the whole measurement process, including people, method and fixture, on real parts.
COPLAIN AI PLATFORM
When a gauge study shows operators measuring differently, Coplain turns your inspection method into a clear, illustrated job aid so every shift measures the same way.
Generate yours automatically with Coplain →Related Tools & Templates