Gage R&R Calculation Spreadsheets - Which one here is valid?

GilesBee

Registered
Hi all,

I'm after some help in finding the definitive Gauge R&R calculation! I have seen so many different spreadsheets both on here & other places and they all give me a different answer, varying from 6% to 88%.

I am measuring a station wagon tailgate on an e-cube type fixture that gives me the result as a deviation from nominal (mm) as opposed to an absolute number. 10 parts, 3 operators, 3 trials so fairly standard stuff

Can anyone on here point me to the correct spreadsheet?

Thanks in advance
 

Marc

Fully vaccinated are you?
Leader
A quick "Bump". It is a problem with respect to files being uploaded here. Many are files people have asked to be "assessed" (typically with respect to cell formulas in spreadsheets) by the great people who help out here. That makes searching the Files Attachments listing a "hit or miss" situation. Because of this, there are a lot of files with "Gage R&R" being in the title that aren't relevant per se. There is a link to the discussion thread the file is attached to a post in for context when you're searching the Free Files Attached to Posts in Threads - Post File Attachments listing.

I wish I could easily fix this, but it's the nature of having a forum with file attachment capabilities. This is not to mention the different Gage R&R types.

My Thanks in advance to anyone who can help in this specific case.
 
N

NumberCruncher

Hi GilesBee

There isn't a single "correct" sheet. The AIAG manual has two different methods for calculating the R&R statistics. One is the average and range method which uses the range and some fiddle factors to estimate the standard deviation of the data. The other method is the analysis of variance (ANOVA) method which uses the standard deviation of the data to estimate the standard deviation of the data.

However, there is also this article by Donald Wheeler.

http://www.spcpress.com/pdf/DJW189.pdf

This explains his criticism of the AIAG methods, and some people will have spreadsheets that will follow his method.

I have attached a simple variables ANOVA spreadsheet which has the merit of being pre-loaded with the data from the MSA manual (4th edition) and giving the same results.

This sheet automatically decides if the interaction term is significant and calculates the estimates of variance accordingly. In the case of the data from the manual, the interaction term is not significant, so the results are those given on page 127.

I'm sure that there will be people who will say that Donald Wheeler is correct rather than AIAG. I am pragmatic on this. If the results satisfy your customer, and can't be criticised by the auditor, I don't really care!

NC
 

Attachments

  • Gauge R&R - ANOVA - 4th edition data.xls
    22 KB · Views: 695
D

Dave Mac

I have been in this situation myself, where I have struggled to determine the correct calculation and I've never really found the definitive solution to it.

But I can share one thing that has helped me - which is to report the Process Capability along with the Gage r&r. (I am defining Process Capability as the number of times that the 6 sigma spread of the process fits into the spec limits.)
This is because if I have a good Process Capability, even when measured with a high variation gage, then the true spread (with perfect measurement) of the process can only be better (ie smaller). And if my measured process is centered with a reasonable process capability why would I spend $ on trying to improve the gage? - I have minimal danger of making scrap and not knowing it, or making good product and thinking it is scrap - because my process does not produce product anywhere near the spec limits.
So, what I mean is that high gage r&r for a process with a good Process Capability may not be the highest priority.

On the other hand if process capability is poor or marginal, then I have a very real danger of measuring a good part and calling it scrap, or measuring a bad part and calling it good. So in this case I definitely want to know how much variation is caused by the gage, and how much is driven by the process. If I find a high gage r&r here, then reducing the gage r&r for this process becomes a high priority, because of my high risk of passing scrap as good and good product as scrap. Also, if I am going to do process improvements, having a low variation gage here makes my improvement efforts easier to manage, because I have a clearer view of the real process values.

This is a bit difficult to explain, but basically if you have high/poor gage r&r then I don't think you can get a full picture of the impact unless you take into account the process capability. - My opinion.
Hope this helps
 
Top Bottom