Gage R&R Calculator

Analyses a crossed gage R&R study, in which several operators measure the same parts several times, to show how much of the variation comes from the measurement system.

How to use

1. Paste the study into "Measurements" (one row per part, columns A1, A2, A3, B1 …; other layouts are recognised). "Load example" shows the format.
2. Check the line that reports parts, operators and trials; if it is wrong, set "Data layout" or "Number of operators".
3. Optionally enter "Tolerance (USL − LSL)" and "Process standard deviation".
4. Switch between "ANOVA" and "Average and range", then use "Copy results", "Save as CSV" or "Print".

How it works

• Average and range: EV = R̄ × K1, AV = √((X̄diff × K2)² − EV² ÷ (n × r)), GRR = √(EV² + AV²), ndc = 1.41 × PV ÷ GRR rounded down, with the AIAG MSA (4th edition) constants.
• ANOVA: the operator × part interaction is pooled into repeatability when its p value is above the chosen level (0.05 by default).
• The 10 % / 30 % limits and ndc ≥ 5 are shown as common guidelines, not a pass or fail decision.

Questions and answers

What is an acceptable %GRR?

The AIAG manual calls under 10 % generally acceptable, 10 % to 30 % possibly acceptable depending on the application, and over 30 % in need of improvement.

Why do ANOVA and average and range differ?

ANOVA uses all the data and separates the interaction; the average and range method uses ranges. For the AIAG example %GRR is 27.86 % and 26.68 %.

What does ndc mean?

The number of distinct categories of parts the system can tell apart: 1.41 × part SD ÷ gage R&R SD, rounded down. At least 5 is commonly recommended.

How many parts, operators and trials do I need?

A common design is 10 parts, 3 operators and 3 trials. Any balanced study with at least 2 parts and 2 trials is accepted.