GRR (Gage Repeatability and Reproducibility) for Material Strength Tester

[Stevenli]

Folks,

I knew there are many discussion about non-repeatable GRR or destructive GRR, one of the classic way is use so-called "NESTED" way e.g. in Minitab. Other than that, I try to get the help from you guys whether we can only monitor the stablity of test equipment over time with standard cupons or the actual samples from a short time window. Very typical equipement is material strength tester.

I believe it works in some cases, but want to double confirm from your expertise.

Thanks

Steven

 

[Miner]

Re: GRR (repeatability and reproducibility) for Material strength tester

Your question is unclear. Are you asking whether stability testing is an acceptable substitute for an R&R study?

The two types of studies measure different aspects of a measurement system, and are not interchangeable. A stability study assesses whether a measurement system's bias (i.e., location) will drift over time whereas an R&R study assesses the spread or variation of the measurement system.

If your question is "Is it necessary to perform an R&R study?", it depends on many factors: your industry sector, do you have multiple operators or just one?, do you average multiple test results or throw out the highs and lows?, etc.

My MSA blog may be of assistance.

 

[Stevenli]

Re: GRR (repeatability and reproducibility) for Material strength tester

Your question is unclear. Are you asking whether stability testing is an acceptable substitute for an R&R study?

The two types of studies measure different aspects of a measurement system, and are not interchangeable. A stability study assesses whether a measurement system's bias (i.e., location) will drift over time whereas an R&R study assesses the spread or variation of the measurement system.

If your question is "Is it necessary to perform an R&R study?", it depends on many factors: your industry sector, do you have multiple operators or just one?, do you average multiple test results or throw out the highs and lows?, etc.

My MSA blog may be of assistance.

Miner, thank you. In fact, I want to clarify both items as you stated. Many people will ask it's a must to do GRR for above situation. Or we only verify stability (bias) with other variation source controlled by standard work etc.

Will review your MSA blog for more details.

 

[Bev D]

as Miner stated the two approaches you suggested as alternatives do not provide the same information as a Gage R&R. specimen prep, operator technique, any fixtures all contribute to measurement variation which you wouldn't see with coupons or ' tracking' actual results. using coupons (which are far less variable than real parts) is a calibration (accuracy) test, whereas R&R tests repeatability.

IF you track your actual results for a period of time AND the total variation remains well within tolerance then you know your measurement variation is negligible.

Since well designed R&R studies provide a wealth of information about your measurement systems, the real question here is: Why don't you want to do a Gage R&R?

 

[Stevenli]

Since well designed R&R studies provide a wealth of information about your measurement systems, the real question here is: Why don't you want to do a Gage R&R?

Yes, we'd like to do GRR for those wealth of information, however as the measurement system is non-repeatable/destructive, we don't think there is a well-deployed method cross the industry, either in Atuomative or other commercial industry. Another consideraton is the cost for destructive test. Your thoughts?

 

[Bev D]

Not sure why you think that the methods for destruct R&R are not well established...I can understand that the AIAG study design (10 parts measured 3X by 3 Operators) and analysis method are not very user friendly or informative.

Please see Measurement Systems Verification and Validation for a better study design and analysis methods that allow you to get the most information from your design.

What type of testing are you doing?
Some tests must be nested (the entire sample or part is consumed) but others can be done using the same sample (sub samples from a homogenous batch, multiple test locations for hardness or tensile testing...)

You can save money in the beginning by using only one operator and testing 15 parts twice. You can also perform the study by simply taking an extra reading for each sample submitted for standard testing...the MSA doesn't have to be completed all at once.

 

[Stevenli]

What type of testing are you doing?
Some tests must be nested (the entire sample or part is consumed) but others can be done using the same sample (sub samples from a homogenous batch, multiple test locations for hardness or tensile testing...)

You can save money in the beginning by using only one operator and testing 15 parts twice. You can also perform the study by simply taking an extra reading for each sample submitted for standard testing...the MSA doesn't have to be completed all at once.

Not sure whether you have any experience to share the GRR of tensile testing? This would be much helpful. Thanks.