GRR for Attribute Gauge

[Ron Rompen]

Good morning: Looking for some guidance (or suggestions) on the best way to do an evaluation of an attribute gauging system.

First, the parameters of the problem:

The gauging system in question is a fully-automated attribute measurement system.
There is no potential for operator influence on the results of the evaluation. There are several fully automated processes before the measurement cycle.
Parts are fed from a single source, moved/loaded/removed by a single robot.
The accuracy of the gauge has been established and confirmed.
The repetition of motion of the robot has been verified

Now for the hard part:

A normal attribute study would be 30 - 50 pcs, evaluated several times, and the evaluation results compared to the 'referee' value.
I have 3 sample parts (1 just below the lower tolerance, one within the tolerance band, one just above the upper tolerance). It is NOT possible to obtain additional samples (parts are extremely large, extremely expensive, and extremely difficult to manufacture)

Customer preference (as well as my own) is to calculate the Kappa value of the gauging system (target value of ≤ 0.70).

Ideally I would like to use Minitab to perform the analysis, however an Excel spreadsheet which provides the Kappa analysis would be acceptable.

Anyone have any suggestions?

 

[Miner]

Before jumping into the attribute analysis recommendations, I have a big question. You say this is an attribute gage, but also stated that it was accurate, which is a statement about continuous data. I work with a lot of automated testers that report out a Pass/Fail (attribute) result that is actually based on a continuous measurement. That continuous measurement can be accessed and downloaded. Is there any chance that yours works in a similar fashion? If so, you can use a Type 1 Repeatability Study.

 

[Ron Rompen]

Thank

Before jumping into the attribute analysis recommendations, I have a big question. You say this is an attribute gage, but also stated that it was accurate, which is a statement about continuous data. I work with a lot of automated testers that report out a Pass/Fail (attribute) result that is actually based on a continuous measurement. That continuous measurement can be accessed and downloaded. Is there any chance that yours works in a similar fashion? If so, you can use a Type 1 Repeatability Study.

Thanks for the quick response Miner. Your suggestion mirrors the initial thoughts that I had regarding the gauge - I have been informed (by the software creator for the system) that it is not feasible for me to access the continuous data. That would have made it MUCH easier, I agree.

 

[Bev D]

You should tell the software person to make the data available…

 

[optomist1]

hmm, that's challenging...Mr. Obvious here...by the time the attribute gage/measurement device starts rejecting parts, it is too late. Again, continuous data is/are far superior to attribute. and echoing fine Bev's post, if the part is large and expensive, obtaining continuous data should be an easier sell...

 

[Miner]

Over the years we started uploading the continuous data to a quality information system that allows quality and engineers to access the continuous data over time. The floor associates still see Pass/Fail but the data is available on demand.

 

[gakiss2]

Thought I would take a look at the problem from a different perspective. Won't help with this gage but maybe in the future for similar gages. When the gage is first ordered have your purchasing department put into the contract that the gage has to meet XYX R&R requirements in order for the vendor to receive payment (usually half up front then half after acceptance). Of course your purchasing department has to stand firm with you on the conditions.

The really great part is that now you have the vendor do the study and prepare the report and you just put it into your PPAP docs and send it off to the customer. All parties happy

 

[Ron Rompen]

You should tell teh software person to make the data available…

In a perfect world I could/would. But as they don't work for us, I have no control/influence.

 

[Ron Rompen]

Thought I would take a look at the problem from a different perspective. Won't help with this gage but maybe in the future for similar gages. When the gage is first ordered have your purchasing department put into the contract that the gage has to meet XYX R&R requirements in order for the vendor to receive payment (usually half up front then half after acceptance). Of course your purchasing department has to stand firm with you on the conditions.

The really great part is that now you have the vendor do the study and prepare the report and you just put it into your PPAP docs and send it off to the customer. All parties happy

Yes, that would be an ideal situation. But many times we are constrained by the limits of the software design company. There are a LOT of 'extra' features that I would like to see in Excel (as an example). But no matter what kind of requirements I write in my specifications, Microsoft is NOT going to rewrite the code just to make me happy. Either I find another supplier who can give me what I want/need (not feasible in this case) or I figure out how to 'make do' with what I have.

 

[gakiss2]

Indeed, you have to have a vendor willing to go along with you for that method to work. Good luck with your project