I have carried out a gauge R/R study see below. The total gauge R/R is 7.82 but my customer has informed me that the study has failed because the %tolerance is above 20%?. The product has a 0.06mm tolerance. What is the %toerance a meassure off?
Thanks in advance for any help.
Source Variance (of Variance)
Total Gage R&R 6.74E-06 [b]7.82
[ Repeatability 6.74E-06 7.82
Reproducibility 0.00E+00 0.00
Part-to-Part 7.95E-05 92.18
Total Variation 8.62E-05 100.00
StdDev Study Var %Study Var %Tolerance
Source (SD) (5.15*SD) (%SV) (SV/Toler)
Total Gage R&R 2.60E-03 1.34E-02 27.96 22.29
Repeatability 2.60E-03 1.34E-02 27.96 22.29
Reproducibility 0.00E+00 0.00E+00 0.00 0.00
Part-to-Part 8.92E-03 4.59E-02 96.01 76.53
Total Variation 9.29E-03 4.78E-02 100.00 79.71
%Tolerance in Minitab is what people often call the precision-to-tolerance ratio. That is, the percentage of the tolerance taken up by the measurement error. In your case, the results are telling you that the measurement error is taking up 22.29% of your total tolerance of 0.06mm (assuming you put everything into Minitab correctly).
I’m not sure what you mean by “the study has failed”. Is the customer saying that your gauge is inadequate because the P/T is > 20% or is there some suggestion that the GR&R study as a whole has “failed” because P/T is > 20%? If it is the latter, I don’t understand that at all. A 20% P/T is very typical. It is no indication of a flawed study.
In your case the 22.29% P/T ratio is on the high side regarding gauge adequacy, especially if you are expecting to get to a high sigma level in your process, but guidelines for P/T are usually set at < 30% with a k of 5.15 like you have. I would not think that this P/T value would indicate an inadequate gauge based on that measure alone.
Overall, with your P/T of 22.29% and your GR&R of 27.96% your process is running at a capability (Cp) of about 1. This corresponds to a sigma level of about 3. Decent, but not great.
You may want to look into a paper by a colleague of mine that does a great job of explaining P/T and GR&R together in a two-dimensional analysis. It may help you get a better understanding.
White, T. Kevin, and Connie M. Borror. “Two-dimensional Guidelines for Measurement System Indices.” Qual. Reliab. Engng. Int. 27.4 (2011): 479-87. Print.
If you are familiar with the concept of NDC using study parameters, this is the same requirement using specifications. The calculation is: ndc = 1.41* PV/GRR The desired minimum of ndc > 5 if inverted and changed to a percent is 20%. Dividing 5.15 * GRR by the Specification Spread results in the percent the measurement error uses of the the total specification. The common practice today is to multiply GRR by 6 not 5.15 which would make the percentage even larger.