Viscometer Gage R&R (GR&R) Questions

[jmarcus - 2008]

I'm fairly new to the GR&R process. I've ran two so far with not great results. I've ran a Digital Caliper and a Viscometer. My question is about the viscometer

We don't have many lots of product to choose from, so I figured if I can't have 10 samples and 3 trials I would do 3 samples and 10 trials.

Our product is solvent based, has a tolerance of +/- 30 mPa and it seems to change over the 20 minutes it takes to measure it, not much 2-3 units, but still changes.

I plug it into our GR&R spread sheet and I get numbers in the 40-50%.

I do not understand how the statistical date is working. I would think a 5% change could still be controlled with us knowing that it will change slightly over time.

Plus all the readings were well within Tolerance.

Is there a trick in running GR&R on a product that isn't 100% stable?

Thank you in advance.

[David DeLong]

In some situations, a R & R study may not be appropriate and your may be in one of those situations. Statistical studies on variable gauges are to reflect variations in the gauges and I would perform a stability study on this gauge rather than an R & R.

Get a sample and measure it every day for 25 days. Plot it on an individuals and moving range chart and calculate the control limits. Are your readings within the control limits?? Great! Now that is easy.

[jmarcus - 2008]

How is measureing the sample over 25 days different then measuring it 25 times over a single day and running those charts?

Also would these types of test be exceptable for an external TS16949 audit?

Thanks again.

[David DeLong]

In Reply to Parent Post by jmarcus

How is measureing the sample over 25 days different then measuring it 25 times over a single day and running those charts?

Also would these types of test be exceptable for an external TS16949 audit?

Thanks again.

Running 1 per day for 25 days is different from 25 time during a day since we have a greater interval between using the gauge. We could run 25 consecutive readings but that is not really a good feel of the gauge over time.

Stability studies are included in the MSA 3rd edition and I would always have those handy when 3rd party audits are involved.

Some gauges, such as hardness testers, spectrometers, and viscosity measuring systems are not conducive to R & R studies. One would probably have difficulty getting a good R & R study if your samples are are not much different from each other. One must have samples are both ends of the specification limits to achieve a good result.

I would change it to a stability study.

[Miner]

In Reply to Parent Post by jmarcus

Our product is solvent based, has a tolerance of +/- 30 mPa and it seems to change over the 20 minutes it takes to measure it, not much 2-3 units, but still changes.

I plug it into our GR&R spread sheet and I get numbers in the 40-50%.

I do not understand how the statistical date is working. I would think a 5% change could still be controlled with us knowing that it will change slightly over time.

Plus all the readings were well within Tolerance.

Is there a trick in running GR&R on a product that isn't 100% stable?

Thank you in advance.

When you state that the readings change over the 20 minutes, does this mean that one measurement takes 20 minutes, or that it takes 20 minutes to perform all of the R&R measurements?

Does the product eventually stabilize after 20 minutes, or does it continue to change?

You would definitely need to use the method for a destuctive gage study. If the product eventually stabilizes, perform the measurements after it has stabilized. If does not stabilize, you may need to get inventive and perform a regression study versus time, then apply a correction factor to your measurements based on the time.

Please review earlier posts in this forum for a discussion on collecting samples for an MSA based on whether the gage is used as an inspection gage, or as a process control gage. David made a good point about the effect that the variation of your test samples has on your results. Make sure that you are using the correct metric (P/T Ratio or %GRR) and that your samples represent the process variation if %GRR is used.

Please note that a stability study is intended to assess the stability of the measurement system. The basic assumption is that the standard itself is stable and unchanging. This seems to violate that assumption. A similar study would be worthwhile to assess the stability of the product over time, but this has a different purpose than a gage stability study.

[jmarcus - 2008]

It takes about 20 minutes to take all the readings, and it never truly stabilizes. The solvent changes the viscosity slightly over time. Internally during production check it every two hours and reset the viscosity if it's out of our specification.

Are all of these statistical methods found in the MSA / SPC books?

Thanks again

[Miner]

The MSA book covers all of these methods except the regression analysis.

See NIST Engineering Statistics Handbook for information on regression analysis.

[DsqrdDGD909]

There are some products that are shear sensitive. Is it a standard ASTM method?