Gage Studies

I’m assuming your choices are to run one anlaysis on just the little rocks vs running two analyses: one on little rocks and one on big rocks.
A lot depends on the nature of the variation in your gage. I would venture to guess that the repeatability would be pretty much the same regardless of the size of the rocks. But, the reproducability could be different if the operators react to the different rock sizes differently.
Whether or not the GR&R is affected by rock size, certainly you’ll need to be able to compare the GR&R to the tolerances of both rock sizes, and more preferably, you’ll be able to compare the GR&R to the total variation, including the part-to-part variation for both large and small rocks. The problem with getting part-to-part variation outside of a Gage R&R study, such as historical data, is that it will include the measurement system variation.
My advice – it is usually easier to run both studies (one for large rocks and another for small rocks) than to mess with all the assumptions you’ll have to make otherwise. If running the Gage R&R is very expensive or consumes resources, then if may be worth the “mess”.
I recommend you have a statistician help you sorting out that mess.
Ken K.

I would vote for two separate studies.
The part-to-part variation between little rocks and big rocks is by design, where p-p variation between either all little rocks or all big rocks is a result of process variation.
One of the major purposes of a Gage study is to compare the R&R (error) with the part to part variation occuring due to process variation. The purpose of a measurement system is to discriminate between products that are different on account of process variation.
If little rocks & big rocks are mixed, this would lead to a huge part to part variation; and wrongly give you an excellent R&R/ P-P variation ratio, putting you in a lull of false security.
Ravi Khare
[email protected]
 
 

Hi Cindy,
If you are just doing gage R & R’s there is only need for one study.  QS states that the “family” of gages can be studied.
 
jp

Hi,
  be very careful with this one. The final R&R value that most stats packages spit out is the ratio of variation in the measurement system v’s the variation in product measurements. i.e. R&R = 15% means that 15% of the variation in product measurements is due to the measurement system.The key phrase here is variation in ‘product measurements’, this figure will be larger if you measure big rocks & little rocks then if you only measure one type, & hence the R&R figure will be smaller. Whatever R&R figure ( or no. of distinct categories figure) you end up with only applies to that range of product measurements that you used in the study. In this case you will need to perform seperate studies for big rocks & little rocks (probably not what you wanted to hear).
If you have time,what I suggest you do is perform 3 studies using the same operators and gage for each. One study on little rocks, one on big rocks, and the other using big & little rocks. You should end up with a smaller R&R figure for the 3rd study (due to the larger figure for overall variation). By doing this you will end up with a better understanding of what the actual results mean, and how part selection can effect the final data.
 
Good luck
Patrick

If the size distribution of the rocks are normal, then you have to do only one anaysis. If it is bimodel, then group them into two and measure independently. You should also define the objectives of the measurements required for analysis.

Show a little initiative! No one has the time to send you everything you want in an easy to digest form — you’ll have to take a class to get that.
Here are some resources from this site (try the search, it’s v. good):https://www.isixsigma.com/dictionary/Gage_R&R-147.htm
https://www.isixsigma.com/st/msa/
Raj

The QS9000 Standard states in 4.11.4.S;
The supplier should repeat gage R&R studies when warranted by measurement system change (including operator) and have a systematic method to improve gaging.
We now do yearly gage R&R studies on each gage that measures our product. The statement above says we SHOULD do a study when we change operators. Does this mean MUST. We change operators often and this would interupt production too much. Can we gather data from our actual measurement of the product of each operator instead of all these studies.
                          Dan

gage study is basically a study of gage by which we are making checks this process will tell us about the perfection of our gage.gage r&r tells us about the repeatibility and reproducibility of our gage and our operator

You can use only one rock (big or little) to measure your measurement fixture (system).  The result will apply for other.
 
Max

We have to measure 2 separate rocks with 2 separate gages.

How does one account for ‘gauge block’ changes while performing a gage study? For example, if one has a test device (the gage) that measures valve crack pressure, but the valve changes each time it is tested due to it’s design, wear, etc. (assuming that the most repeatable gage imaginable were used for the study), the gage will likely indicate a poor R&R result, even though it is actually very good. I assume that one has to have non-changing gauge blocks before commencing the study, but should’t one be using the actual product(s) that will be tested on the device? Is it realistic to create custom, stable gage blocks before beginning such a study to get a true understanding of the gage’s precision?

Gage evolution with repeated measurement is a quite common situation in electronic industry.
One practice seen in publications is to modelize measurement evolution due to repeatability (linear or polynomial regression), and to correct measured values accordingly. Computation of R&R indicators is tehn done considering “corrected values”.
Do you think this paractice is acceptable ?

Jonathan,And you have stumbled upon the inherent weakness of the MSA manual. You get your result and … what next? The results are in terms like NDC and %GRR, but your gage is measuring things like mm or ft-lbs.If you do not have long term process variation to look at, you are correct in using the study variation. Because it is all you have. But your question is “now what..”Answer: Because your Gage R&R is “close” to being unacceptable, it behooves you to do it again in a few months when you have some good, long-term process data.A measurement system is part of a process – no more, no less. We do FMEAs, process studies, etc. to determine which parts of a process we have to watch closely (inspection every hour) from those that we can watch occasionally (inspection every tooling change). I am sure you understand that concept – driven by tolerance you are trying to hit and repeatability of the process. The measurement system is no different.A borderline Gage R&R means you have to keep an eye on it with more calibrations, more remastering, etc. It is simply risk management.Your NDC of 4 is troublesome. I have seen in older literature where NDC of 4 is acceptable. But as time has passed, I have seen that number climb from 4 to 5 to 6 and without a lot of reasons as to why.The NDC is basically the ratio of the noise of your parts to the noise of your gage. (Scaled a bit because it is a vector). You are saying roughly that your gage noise is one-fourth your part noise. Is that good? Sadly, it is really a financial decision.Let’s say you have a spec (for easy math) of 1 mm +/- 0.2 millimeters. Your categories you can discriminate are (roughly)
0.98 to 0.99
0.99 to 1.00
1.00 to 1.01
1.01 to 1.02This is better than a go/no-go gage and certainly accurate enough to show drift using pre-control or another acceptable method. BUT – the outer bands have enough inaccuracy that you could pass a “slightly” bad part. So if the characteristic was important, you are probably ok. But if the characteristic was CRITICAL, you may want to ONLY accept the .99 to 1.00 and the 1.00 to 1.01 bands which would guarantee that you never passed a bad part. Assuming your process was centered, you might be throwing away 1/3rd of your good parts “to be safe”.Your production department would hate you :-).(And I wish I had a better answer. I am trying to work out what to do with the Gage R&R numbers myself. Life would be easier if the process spit out the +/- of the *GAGE*. Then you could just set your inspection limits directly and the “good enough” question would go away.)

Hi Cindy.
You should go ahead and perform two different gage studies. The reason behind is, that there are two different products having different characteristics and the Quality of each product is judged diffrently.
Whenever you have such a doubt just keep this in your mind as a golden rule : Who are the customers of that service or that product . If the customers have different expectations from the different products that you have then those products have to be dealt separately and their characteristics have to be measured separately.The same or different measurement system does not make any difference in your approach. If you have two measurement systems for each product use them respectively for each.
This has helped me a lot and I am sure that will help you as well.
Cheers
Manu