Question on measure used for Gage R&R - Characteristic is a Knurled diameter

[JRKH]

Our customer asked us for a gage R&R and capability study on some parts that we farm out.
The suspect dimension is a knurled diameter .256 +.005/0.

I took 30 of the parts to our supplier and he
Our supplier has provided a gage R&R which uses the knurl as the Characterisitc.

Is this normal and acceptable?

My understanding is that a Gage R&R is designed to determine the accuracy of the gage and operator and that the parts measured for the tests should be of a consistant, known dimension.

I know for a fact that these knurl dims have a .001 variance depending on where you measure on the dia.

Please advise an old inspector as to whether I am correct or going nuts. Overall I have difficulty understanding and interpreting Gage R&R and am likely going to post another question concerning the results of these tests.

James

 

[Jim Wynne]

Our customer asked us for a gage R&R and capability study on some parts that we farm out.
The suspect dimension is a knurled diameter .256 +.005/0.

I took 30 of the parts to our supplier and he
Our supplier has provided a gage R&R which uses the knurl as the Characterisitc.

Is this normal and acceptable?

My understanding is that a Gage R&R is designed to determine the accuracy of the gage and operator and that the parts measured for the tests should be of a consistant, known dimension.

I know for a fact that these knurl dims have a .001 variance depending on where you measure on the dia.

Please advise an old inspector as to whether I am correct or going nuts. Overall I have difficulty understanding and interpreting Gage R&R and am likely going to post another question concerning the results of these tests.

James

GR&R should be done when there's a specific reason for doing it, not just because of a blind requirement. In terms of part qualification, if there are special characteristics, it's usually expected that the supplier will provide evidence of control, and that should include evidence that the measuring system is efficacious. If there's no specific purpose for the analysis, why do it?

 

[JRKH]

Why Do it?
Because the customer asked for it.

I am not going to question my customer's reason for asking for the gage R&R. I just want to determine if what my supplier did is acceptable before forwarding it on to the customer.

JAmes

 

[Jim Wynne]

Why Do it?
Because the customer asked for it.

Well, you didn't say that in the original post. My guess is that the customer isn't likely to scrutinize the report too closely, and since it seems that the analysis is being done just to satisfy a requirement to do an analysis, I would probably accept the supplier's report if it appears to make sense. It seems a little odd that the supplier would choose the knurl as the target characteristic, but if they can measure it consistently...

 

[JRKH]

Actually I did state the customer requirement in the origional post - but let us not quibble - - - :truce:

Consistancy is actually part of my concern - their %R&R = 24.95 which seems very bad to me and I am wondering if I need to do some training with them.

I am trying to gain a better understanding of how gage R&R works so I have asked an additional question in a new post.

James

 

[Jim Wynne]

Actually I did state the customer requirement in the origional post - but let us not quibble - - - :truce:

You're right, mea culpa.

Consistancy is actually part of my concern - their %R&R = 24.95 which seems very bad to me and I am wondering if I need to do some training with them.

I am trying to gain a better understanding of how gage R&R works so I have asked an additional question in a new post.

James

Well, 25% doesn't seem all that bad, given the characteristic. You might want to look at the breakdown--how much operator error (repeatability), part variation and device error (reproducibility) are contributing to the final number. Given a characteristic that you know has significant within-part variation, and without a consistent measurement technique, you're going to have results approaching the limits of acceptability at best, I think.