GRR tells you!!!!

What GR&R Tells You
A gage Repeatability and Reproducibility Study conveys 6 basic pieces of information. They are:
· Part variation
· Repeatability and reproducibility (R&R)
· Repeatability
· Reproducibility
· Percent of total variation
· Percent of tolerance
Part Variation is a measure of how much the process is actually varying. If an infinite number of parts made by this process were measured using an Ideal Gage were used 99.0 % (5.15 sigma) of the parts would be within an interval Part Variation wide. This is, of course, only an estimate. This is important because unless the part variation is large compared to the gage variation (the R&R) the assumption that the sample (observed) standard deviation, mean, etc. are approximately the same as the population standard deviation, mean, etc. is not valid. Note that the part variation is always less than or equal to the total variation. So if the observed process is well within specification limits (and the Cpk is good) but the Part Variation is not large compared to the R&R, your process is better than it seems. This condition results in SPC not doing as much good as it would with more precise gages.
Repeatability and Reproducibility (R&R) is a measure of the total variance caused by the gaging system as opposed to the manufacturing process. This number, expressed as a percentage of tolerance, is the most important "warning flag". If R&R is a large percentage of total variation but not a large percentage of tolerance, both the gage and the process are good and the process standard deviation is even smaller than it appears (meaning the process spread is smaller and Cpk larger than reported). This also means a more precise gage would allow tighter control limits for earlier detection of loss of control. The following table shows typical interpretation
% of tolerance Interpretation
0 - 10% Gage is suitable
10 - 30% May be acceptable based on importance of the measured part, cost of gage repair, etc. look at R and R separately
over 30% Gage system needs improvement
Reproducibility is a measure of operator difference. The most common cause of a poor (large) Reproducibility is insufficient operator training. A gage that is sensitive to different techniques or is difficult to use would also cause a large Reproducibility.
Repeatability is a measure of how much the gage readings vary when the same operator measures the same part several times. A large value might indicate gage wear, improper measurement policy (but all operators are making the same "mistake", it is not a matter of individual operator skill), or a gage without sufficient resolution or with a defect. This number is best used by comparing it with reproducibility.
Percent of total variation
The above 4 numbers are listed first as raw numbers, then as percentages of total variation and finally as percentages of tolerance. The percent of total variation shows how the variation is distributed. Note that these numbers are not supposed to add to 100% (see equations in appendix). These numbers are most useful in pinpointing a problem or showing where to concentrate improvement efforts.
Percent of tolerance
The first 4 numbers above are also displayed as percentages of total tolerance. It is these numbers that most companies use to determine the usefulness of the gage or if improvement is necessary. As with percentage of total variation, the numbers will not total 100%.