I have a question about uncertainty measurement and Gauge R&R. Once they both have been performed and recorded, how often must they be repeated in accordance with ISO 17025?:D

As far as R&R, I would perform it often enough to determine if the process is stable. Do you hire new people? Do variables change in the process? Do you make improvements and such to possibly lower the variance? Or, did it go up?:D

As far as calculating the uncertainty, I would recalculate whenever any of the variables involved with the estimation change. If you receive a new reported uncertainty for the standard, R&R varies, different environmental conditions, etc.

Obviously, I would perform analysis much more frequently at the beginning of the process. The frequency (and possibly number of replications) can reduce once you can demonstrate consistency/ stability.

So, start off often and frequent, and review the data. Once you have established consistency, then consider extending the intervals. For uncertainty, recalculate when there is good reason to conclude a variable has changed.

:2cents:

I would suggest that whoever is planned to use the gage successfully go through the gage R&R study. I would also plan to do a new Gage R&R annually. Just my opinion.:cool:
I have a question about uncertainty measurement and Gauge R&R. Once they both have been performed and recorded, how often must they be repeated in accordance with ISO 17025?:D

Thank you for the input. I started with this company 3 months ago and the uncertainties have not been looked at in over 2 years.

I have an ISO 17025 audit at the end of this month and I wanted to see if I have to explain the gap.:)

Thank you for the input. I started with this company 3 months ago and the uncertainties have not been looked at in over 2 years.

I have an ISO 17025 audit at the end of this month and I wanted to see if I have to explain the gap.:)

Well, I would certainly not stress out too much over it. After all, you are inheriting the system. I would let your management know the status of the program, and any possible concerns that you may have.

If it was me, I would do as you suggest and perform a gap analysis. Also, I would focus on your processes from a risk-based perspective. Try to take care of the important things first; then go from there.

Brad is right about redoing MU when things change.....for example, if your standards go out for calibration, when they get back recalculate your uncertainty.....

Remember uncertainty changes constantly, but the difference is the resolution needed to see the constant change.....

I am in the process of updating all of the uncertainties. Thank you for all of your input, it is greatly appreciated.:thanx:

All, We need both?
Our lab is ISO 17025 and we have just finished a very long GRR study of all of our scope of accreditation tests. We have to perform the MU on these various tests. Can't I just use the GRR to evaluate the MU? It would probably be a larger number than a real MU but it would be safe.
For example: (our worst test)

Gage R&R Study - ANOVA Method

Two-Way ANOVA Table With Interaction
Source DF SS MS F P
Lot 1 4980.63 4980.63 109.283 0.009
Techniciens 2 48.70 24.35 0.534 0.652
Lot * Techniciens 2 91.15 45.58 21.862 0.000
Repeatability 84 175.12 2.08
Total 89 5295.60

Alpha to remove interaction term = 0.25

Gage R&R
%Contribution
Source VarComp (of VarComp)
Total Gage R&R 4.984 4.35
Repeatability 2.085 1.82
Reproducibility 2.899 2.53
Techniciens 0.000 0.00
Techniciens*Lot 2.899 2.53
Part-To-Part 109.668 95.65
Total Variation 114.652 100.00

Study Var %Study Var
Source StdDev (SD) (6 * SD) (%SV)
Total Gage R&R 2.2325 13.3951 20.85
Repeatability 1.4439 8.6631 13.48
Reproducibility 1.7028 10.2166 15.90
Techniciens 0.0000 0.0000 0.00
Techniciens*Lot 1.7028 10.2166 15.90
Part-To-Part 10.4722 62.8335 97.80
Total Variation 10.7076 64.2454 100.00

Number of Distinct Categories = 6

Gage R&R for Melt flow ASTM D1238

So this test can distinguish 6 distinctive products from a total process variation of 64.2454 (6 sigma). So any difference under 10.71 (64.2454 / 6 .. # of categories) is not significant! So my tool's measurement Uncertainty should be (in the range measured) ± 5.35 (10.71 / 2 .. since 10.71 is my full range) .

So am I Ok or I should go back to school?

Thanks

As far as R&R, I would perform it often enough to determine if the process is stable. Do you hire new people? Do variables change in the process? Do you make improvements and such to possibly lower the variance? Or, did it go up?:D


Gage R& R really has no time function - if it is performed correctly at the beginning - which means the parts and the operators cover the variety of possible levels the gage will see over time. If later you get operators less capable than your original operators, then that is significant enough to redo the gage R&R again. Some folks like to R&R every new operator. Again, that is academic (unless their gaging system has been shown to be so operator dependent that it should not likely be used anyway.)

If the gage is changed, or the measurement process changed, then yes, it needs verified. Improvements can fix one thing, and break another. But, lacking any significant changes to the measurement system, redoing them at any frequency is, again, simply an academic exercise.

As far as stability, my preference is to use calibration data - if the same standards and enough measurement points to establish linearity are used -to establish stability over time. That way I can see measurement change not tainted by any measurement error (such as influenced by roundness, etc.)

Gage R& R really has no time function - if it is performed correctly at the beginning - which means the parts and the operators cover the variety of possible levels the gage will see over time.

This is true unless the measurement device becomes damaged or enters the wear out phase of it's life cycle.

This is true unless the measurement device becomes damaged or enters the wear out phase of it's life cycle.

That is the function of calibration to catch those issues. It should catch drift (if stability is captured) from wear and the inability to read correctly. Calibration does have a time function!