%AV=0 in the MSA GRR

[abdelouahed]

Hi all ,

Please i need your help regarding an issue with %AV in GRR studies, I'm using a Software to calculate the MSA result but in almost cases the result of the %AV is equal 0.
so in the last IATF audit the auditor ask me to explain why the %AV=0 because statistically is illogical.

Is there someone who can give me the reason why ? also if you have a excel sheet that can i use to compare the results between the software and the sheet.

I have try with the GRR example in the MSA manual and the result was the same ( manual vs software ).

Thank you.

 

[Miner]

I would not say that %AV=0 is illogical. It is unlikely, but not impossible. Most likely, it is not zero, but rounds off to zero in your spreadsheet.

What does your graphical output show?

 

[abdelouahed]

Thank you Mr. Miner for your replay.
Illogical it was a comment of the IATF auditor.
The graphic show a little variance between the operators .

 

[John Predmore]

AV% is the portion of the total variation represented by the appraiser (operator). One reason AV% might be zero is if the measurement process does not have enough resolution (# decimal places) to see tiny difference between parts. For example, if you measure the length of 10 parts with a centimeter ruler and the operator records length to the nearest centimeter, your measurements will not see any difference between parts at 3.1 cm and 2.9 cm. Or maybe your sample 10 parts do not show the full range of variation typically seen in production, sometimes it helps to pull a few parts from different batches to see full range. Another possibility is the operator has near-zero influence on the measurement, for example when the part auto-loads into an automatic inspection station and the operator merely pushes the start button.

One of us giving you a spreadsheet will not help if one of these problems is the difficulty. The spreadsheet can only work with the numbers you give it. You may need someone knowledgeable to look at the physical arrangement to comprehend sources of variation in the data which are generated.

 

[Miner]

Thank you Mr. Miner for your replay.
Illogical it was a comment of the IATF auditor.
The graphic show a little variance between the operators .

If you upload your data, we can provide more specific feedback.

 

[abdelouahed]

Dear Mr. John ,

Thanks a lot for your response.
unfortunately, I don't have anyone of the both possibility, because I'm using a digital micrometer with a resolution 0.001 mm to measure a height with a tolerance of +/- 0.05 mm, so the resolution of the equipment is bigger than the tolerance more than 10 time.

bellow the measuring results :

App Trials Measuring results

A 1 2,425 2,458 2,491 2,456 2,555 2,458 2,448 2,588 2,429 2,482
2 2,428 2,459 2,499 2,456 2,569 2,449 2,458 2,586 2,425 2,486
3 2,425 2,456 2,492 2,469 2,564 2,459 2,449 2,578 2,428 2,481

B 1 2,439 2,458 2,482 2,459 2,556 2,459 2,448 2,589 2,421 2,488
2 2,426 2,456 2,486 2,468 2,568 2,456 2,448 2,579 2,428 2,488
3 2,426 2,458 2,496 2,456 2,566 2,454 2,449 2,585 2,429 2,482

C 1 2,425 2,458 2,493 2,466 2,568 2,458 2,458 2,589 2,422 2,488
2 2,428 2,458 2,491 2,455 2,569 2,456 2,446 2,584 2,429 2,486
3 2,423 2,456 2,499 2,461 2,569 2,453 2,446 2,586 2,424 2,482

So the obtained results is :

%GRR ndc %AV %PV %EV
8.836% 15.895 0.000% 99.609% 8.836%

Thanks in advance for your support.

 

[Miner]

I duplicated your results in Minitab using the Xbar/R method and got similar, but slightly different results using ANOVA.

The operator means are slightly different, but the difference is so small that it falls below the level of precision (number of decimal places displayed) and is getting rounded off to zero.

It is not an issue with the precision of the measurement device. That is adequate.