Gage RR acceptance criteria.
Six Sigma – iSixSigma › Forums › Old Forums › General › Gage RR acceptance criteria.
- This topic has 10 replies, 6 voices, and was last updated 13 years, 9 months ago by
Klerx.
-
AuthorPosts
-
August 12, 2008 at 2:19 pm #50732
Hi,
I am about conduct a gage R&R on a piece of equipment. I am using Minitab as the software package to analyze the data.
I have not measures any product with this gage yet so I dont have a historical sigma to include in the variation. Can I continue with study?
What acceptance criteria should I use? What is the industrial standard?
Is there a reason why I would change the study variation number of standard deviations from 5.15 to 1 when I calculate the resulted. I was given a reason before but totally forgot.
Thanks in advance
paul0August 12, 2008 at 2:41 pm #174785Hai Paul,
Historical Sigma is under options so it is Optional. If you use it (Xor the Tolerance) Minitab calculates an extra column for you where the % are calculated with Sigma(process) in the denominator (instead of Sigma(data encountered) ). This column can tell you how good your gage system is in reference to the production Xor Customer specs instead of in reference to the data that you used during the analysis (needed when you can’t use representative data for your analysis). So Answer1 = Yes
Answer2: Standards are:Ford: gage r&R% <10% for perfect; <30% for acceptable; otherwise: do a GB project on the Gage System first.
gage r&R % <15% for OK.
The gage r&R value can be found in the Session Window output of gage r&R in the column (%SV) the first(=upper) value.
I can see no reason to change the 5.15 into a 1. Maybe you remember it wrongly: some teachers say you should change it into 6. Minitab14 uses 5.15 as Default; Minitab15 uses 6. The 5.15 (or 6) has to do with changing Sigma-values into Range-values and is an approximation anyway. The 5.15 corresponds with 99% and the 6 with 99.7% of the data being within the covered Range (100% is unreachable because the Normal Distribution is an infinite distribution).
And remember: the gage r&R is not an investigation of the equipment only!
Hope this helps0August 12, 2008 at 4:10 pm #174792No one has a standard of
0August 12, 2008 at 9:01 pm #174813Hi
Thank you both for the respoce. it has been most helpful. i am assuming the AIAG standard can be carried across to the manufacture of medical devices.
thank you again0August 13, 2008 at 2:42 am #174822
Michael MeadParticipant@Michael-MeadInclude @Michael-Mead in your post and this person will
be notified via email.Maybe I don’t have the latest edition of the AIAG Measurement SystemsAnalysis, but the First edition uses 5.15
0August 13, 2008 at 3:06 am #174823You are not current – it changed about a decade ago.
0August 13, 2008 at 9:56 am #174831Hi
One more question (actually 2). Is it acceptable to make product for the Gage R&R which is outside specification. e.g. specification 21±1mm and our product would be 20±1mm. it is the same toleremce and i would be able to make product which would cover the entire range. 10 samples at different lengths across the 2mm.
Question 2. Is there any danger in taking a sample of product from normal manufacturing process. this would not cover the full range of the tolerence.
Thanks again, very helpful.
Paul0August 14, 2008 at 7:54 am #174863Hai Paul,
both questions relate to the issue of ‘Representativeness’. And the answers are: it depends.
Remember: when you do an analysis the conclusion is only valid ON THE DATA YOU USED. Generally you try to make the data you use a representative sample of your population. And then you decide that the conclusions on your sample are also valid for your population. With Confidence Interval you can also take into account the probability of being (a bit) wrong due to sample variation (i.e. how correct will you be in your predictions).
And remember: data from production is not necessarily representative for ALL production (you could have data in a smaller range or with special causes or without,..).
Only YOU can decide whether the data you use is representative and if the conclusion is valid for your population. In the gage r&R there is an option to correct a wrong Range (data-range population range): The Tolerance option.
Another issue is: does the measurement system work in the same way for all the products (also those out of specs). An example is a machine which not only measures but also tweakes the product into spec if it is very close to the specs. In fact the eqt is then not only a measuring machine but also a repair machine (=production process).
hope this helps0August 14, 2008 at 9:07 am #174866
Michael MeadParticipant@Michael-MeadInclude @Michael-Mead in your post and this person will
be notified via email.Hello Paul.
Q1: Technically no. A gage study matches a gage to s specific product and process. Practically, maybe. The only real issue is whether the gage is lenear. Does it have the same errors at 20mm as it does at 21mm? Probably, In my experience, many companies perform :Gage Studies” on a gage, using one representative product. It ain’t rite, but it is what they do.
Q2: This is also probably acceptable. However, it is best to cover the range of expected product. If everything has the same observed measurement, you can’t do a gage study.0August 14, 2008 at 11:56 pm #174891
Joe PeritoParticipant@Joe-PeritoInclude @Joe-Perito in your post and this person will
be notified via email.Paul, here’s the correct info. I verified the facts having just referenced the 3rd and 4th editions of the AIAG MSA manuals: 5.15 is used because it represents 99% of the area under the normal curve. This is stated in the manual. When you use the calculations sheet in the (current) 4th edition it has a footnote that says see the 3rd edition for an explanation of the 5.15 constant. The 3rd edition has a loose leaf errata attachment for corrections to the book. It says the reference to page iv should have been vi. On page vi you will find the explanation that 5.15 had always historically been the constant used for 99% of the area under the curve. It does not say the constant was changed. But it does go on to say that you “may wish to use 6.0”. It further explains that if you do use 6.0, you should make it clear to people how you are doing the calculations.
0August 21, 2008 at 11:39 am #175066Where did you read that the AIAG default 5.15 changed to 6?
The third edition of the AIAG manual is inconsistent. Pages 188 and 190 use 5.15. Page 116 uses 6. Page vi mentions both but says 5.15 is used “historically, by convention”.
On p. 116 MSA is referring to Process Varition (based on a Capability Study) which means Process Variation = 6 x sigma so it is clear the new (to be used) Total Variation = Process Variation / 6
Minitab is using 6 as default for Study Variation which has NOTHING to do with Process Variation based on Capability Study; you call Process Variation based on Capability Study: the Historical St Dev.
So I don’t see any need for using 6 as default.0 -
AuthorPosts
The forum ‘General’ is closed to new topics and replies.