Gauge R&R and measurement system analysis (MSA)

What MSA looks at

A measurement system is the instrument, the method, the people, the parts and the environment together. MSA studies its bias (average difference from a reference value), linearity (how bias changes over the range), stability (drift over time) and its precision: repeatability and reproducibility. Calibration checks the instrument against a standard; MSA checks the whole process on real parts.

In the automotive industry IATF 16949 requires MSA studies for each type of measurement system named in the control plan (clause 7.1.5.1.1); the AIAG MSA reference manual describes the methods.

Running a caliper GR&R study

1. Pick about 10 parts that cover the real spread of the process, number them where the appraisers cannot see it.
2. Choose 2–3 appraisers who normally do the measurement.
3. Each appraiser measures every part 2–3 times, in random order, without seeing earlier results.
4. Use the same caliper, the same feature and the same position on the part.
5. Analyse with the ANOVA method (preferred) or the average-and-range method, usually in a spreadsheet or statistics software.

With 10 parts, 3 appraisers and 3 trials you collect 90 readings. Customers or internal procedures may set a different plan.

Reading the results

%GRR compares the measurement variation with either the total observed variation or the tolerance. Against the tolerance: %GRR = 6 σ_GRR / tolerance × 100 %. Example: tolerance 0.10 mm, σ_GRR = 0.004 mm → 6 × 0.004 = 0.024 mm → 24 %: marginal, acceptable only if the application allows it.

ndc (number of distinct categories) = 1.41 × PV / GRR (both as standard deviations), rounded down: how many groups of parts the system can tell apart. Below 5, the system cannot follow the process well enough for control charts.

If repeatability dominates, look at the instrument and the technique (force, squareness, zero, resolution). If reproducibility dominates, the appraisers do something differently: align the method and train.

Calipers in GR&R studies

Resolution should be at most a tenth of the smaller of the process spread and the tolerance. A 0.01 mm caliper on a tolerance of a few hundredths usually fails; a micrometer, a fixture gauge or a go / no-go gauge is the better choice (see choosing instruments). Most caliper variation comes from technique: part not deep in the jaws, tilt, uneven force and unchecked zero.

KELIBRON does not run GR&R studies. It trains the technique behind repeatability — checking zero, holding the part square and deep in the jaws, light even force — and shows the size of each mistake in the wrong-use demonstrations.