P
Hello all, 
I done gage since years, but I realize that even recently I have done some without really understanding the things, and probably it was not done the best way. I’m still learning things as any situation is different from others. I think the force of a forum is to compile real experience, investigation, interpretation, that’s why I want to discuss with you about this topic as its part of my current activity.
I passed though the 35 pages of threads regarding Gage R&R, and particularly about rule to apply for the choice of nb of operator, nb of pcs, nb of replicates. As readed it seems there’s no strict rule for that…ok let’s say a minimum requirement…
I would share some knowledges and also ask some clarifications that I’m still looking for:
In my previous enterprise, a GRR sampling rule was written in the internal procedure:
- if the nb of pieces X number of operators is >15, then do 2 replicates
- if the nb of pieces X number of operators is within 8 and 15, then do 3 replicates
- if the nb of pieces X number of operators is within 5 and 8, then do 4 replicates
- if the nb of pieces X number of operators is <5, then do 5 replicates
Now in my current company here the rule of one customer:
- nb of operator X nb of replicates X nb of pieces > 60

Question 1:
Did somebody knows or already saw this kind of rule in litterature or norms?
I got only MSA 3rd edition and see nothing like that in it so maybe I missed it somewhere?
Question 2:
Statements in the threads are that more pieces we have for executing the GRR, more we are confident in the results; however did someone already calculate what the confidence level of the basic rule is and how the confidence level is going if we increase the number of pieces?
As additional informations for the sampling:
- Minitab is able to generate Gage R&R study worksheet. However it follows no rule (a GRR combination of 1op X 2reps X 2pcs is possible for example). A guy from Minitab confirmed to me that there’s no rule for that in Minitab.
- Minitab has a Gage R&R sample size guidance in the help section, but in fact there’s no generic rule inside. Interesting advices are:
o “Use more parts with few measurements of each rather than few parts with many measurements of each”
o “Select parts that represent the actual or expected range of process variation”
Yes….the last sentence is very important and was discussed in other threads, and I assume that everybody knows that the “GRR% to process variation” is very sensitive to the selected pieces (as the “%GRR to Tolerance” is not).
As stated in the AIAG “For process control situations (…) the availability of samples over the entire range operating range becomes very important. An independent estimate of process variation (process capability study) is recommended when assessing the adequacy of the measurement system for process control.
Question 3:
Ok for the capability study, but the time range to be assessed for determining the process variation is still little bit confusing for me, I assume to know my process I have to do a long-term analysis (PpK) and not just taking a CpK of the day before…My approach is then to do an assessment of the process since january 2008, then pay attention that the sampling used for the Gage R&R is representative of the long-term process range (more or less of course).
Do you agree with this approach or do you have another proposition?
Question 4:
I believe that the quantity of pieces manufactured has no link with the quantity of pieces to be selected for executing a gage R&R (i.e the
For example the GRR sampling for assessing a measurement process thatmanufacturing of 30 million of component per year or 10000; the sampling for the GRR will be not different.
Did someone can just confirm that?
Thanks to all for any answers.
Fred

I done gage since years, but I realize that even recently I have done some without really understanding the things, and probably it was not done the best way. I’m still learning things as any situation is different from others. I think the force of a forum is to compile real experience, investigation, interpretation, that’s why I want to discuss with you about this topic as its part of my current activity.
I passed though the 35 pages of threads regarding Gage R&R, and particularly about rule to apply for the choice of nb of operator, nb of pcs, nb of replicates. As readed it seems there’s no strict rule for that…ok let’s say a minimum requirement…
I would share some knowledges and also ask some clarifications that I’m still looking for:
In my previous enterprise, a GRR sampling rule was written in the internal procedure:
- if the nb of pieces X number of operators is >15, then do 2 replicates
- if the nb of pieces X number of operators is within 8 and 15, then do 3 replicates
- if the nb of pieces X number of operators is within 5 and 8, then do 4 replicates
- if the nb of pieces X number of operators is <5, then do 5 replicates
Now in my current company here the rule of one customer:
- nb of operator X nb of replicates X nb of pieces > 60
Question 1:
Did somebody knows or already saw this kind of rule in litterature or norms?
I got only MSA 3rd edition and see nothing like that in it so maybe I missed it somewhere?
Question 2:
Statements in the threads are that more pieces we have for executing the GRR, more we are confident in the results; however did someone already calculate what the confidence level of the basic rule is and how the confidence level is going if we increase the number of pieces?
As additional informations for the sampling:
- Minitab is able to generate Gage R&R study worksheet. However it follows no rule (a GRR combination of 1op X 2reps X 2pcs is possible for example). A guy from Minitab confirmed to me that there’s no rule for that in Minitab.
- Minitab has a Gage R&R sample size guidance in the help section, but in fact there’s no generic rule inside. Interesting advices are:
o “Use more parts with few measurements of each rather than few parts with many measurements of each”
o “Select parts that represent the actual or expected range of process variation”
Yes….the last sentence is very important and was discussed in other threads, and I assume that everybody knows that the “GRR% to process variation” is very sensitive to the selected pieces (as the “%GRR to Tolerance” is not).
As stated in the AIAG “For process control situations (…) the availability of samples over the entire range operating range becomes very important. An independent estimate of process variation (process capability study) is recommended when assessing the adequacy of the measurement system for process control.
Question 3:
Ok for the capability study, but the time range to be assessed for determining the process variation is still little bit confusing for me, I assume to know my process I have to do a long-term analysis (PpK) and not just taking a CpK of the day before…My approach is then to do an assessment of the process since january 2008, then pay attention that the sampling used for the Gage R&R is representative of the long-term process range (more or less of course).
Do you agree with this approach or do you have another proposition?
Question 4:
I believe that the quantity of pieces manufactured has no link with the quantity of pieces to be selected for executing a gage R&R (i.e the
For example the GRR sampling for assessing a measurement process thatmanufacturing of 30 million of component per year or 10000; the sampling for the GRR will be not different.
Did someone can just confirm that?
Thanks to all for any answers.
Fred