Dismiss Notice
You must be a registered member in order to post messages and view/download attached files in this forum.
Click here to register.

Gauge R&R for standard measuring instruments- calipers, micrometers...

Discussion in 'APQP and PPAP' started by Sundance, Aug 31, 2016.

  1. Sundance

    Sundance Member

    Joined:
    Aug 31, 2016
    Messages:
    5
    Likes Received:
    1
    Trophy Points:
    1
    I've been a Quality Manager for many years in the precision machining industry, a relatively low volume production environment. Ten months ago I executed a career shift into a high volume production industry where our 70K sq. ft. facility with a total head count of about 70 people ships in excess of 10 million units per month.


    My introduction and acclamation to statistical methods and activities has been fast & furious. I've since been immersed in all manner of APQP/PPAP activities and have a question about Gage R&R requirements.


    Since I've been here I've been told that Gage R&R studies are generally reserved for part or feature specific gauges that have been specially designed and developed for that purpose, but not for standard measuring instruments such as calipers, micrometers, indicators or other equipment used for general measuring purposes that are regularly monitored through a calibration recall system.


    I've been recently instructed by a customer perform Gauge R&R studies for such standard equipment and it makes little sense to me for multiple reasons. Am I wrong in thinking that performing Gauge R&R studies on these types of instruments will have little or no "value added" impact?
     
    S1D3K1CK likes this.
  2. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    397
    Likes Received:
    303
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    Welcome Sundance. Starting off, you have been misinformed. MSAs on standard gauges definitely provide value. However, many do not understand that such a study must be on combinations of the gauge and the feature measured. For example, on your caliper, you will obtain different R&R results on that caliber if your feature is an outer diameter vs. an external plane-to-plane dimension even though the same anvils are used. Your % Tolerance results will also depend on the tolerance of the feature, and there may also be differences in the same feature type (e.g., outer diameter) if the are large size differences, or differences in how the feature was created which affect variation in form.

    My guess, is that what you have been told is a rationalization in order to avoid doing the number of studies required to do it correctly. The best approach is to create gauge families (see #1 below).

    Several good places to start:
    1. My MSA blog. An easy, minimal statistics approach to explain the key elements. I forgot to add part 10, Creating Gauge Families, which may be found here.
    2. AIAG's MSA reference manual, particularly when you are dealing with the automotive industry, or a customer that has been influenced by the automotive industry.
    3. Dr. Donald Wheeler's EMP III manual. A more statistically sound approach to MSA, but not widely recognized. A good approach to use if your customer's do not dictate the AIAG method.
    If you are not convinced due to your "multiple reasons", come back and specify these reasons. We will be happy to address each one.
     
    Bev D, tony s and Atul Khandekar like this.
  3. Sundance

    Sundance Member

    Joined:
    Aug 31, 2016
    Messages:
    5
    Likes Received:
    1
    Trophy Points:
    1
    Thank for the speedy response Miner. Info like you've provided is why I decided to join the forum. I will visit the links you included in an effort to better understand. I've been listening to all the chatter around me and suspected at times my sources here do not really know, but are just telling me how they see it. Beyond that, let me ask another question. Would it be better to use my manufactured parts for the study or various know standards like Gauge blocks, Gauge pins etc? Forgive me if my questions make me sound like a rookie....but I am still a rookie in this arena!
     
  4. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    397
    Likes Received:
    303
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    Definitely use actual product. The product can introduce complexities that affect the R&R results that would not be seen if gauge blocks/pins are used.
     
    Bev D likes this.
  5. Sundance

    Sundance Member

    Joined:
    Aug 31, 2016
    Messages:
    5
    Likes Received:
    1
    Trophy Points:
    1
    Thanks again Miner. I'm thinking I came to the right place.
     
  6. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    260
    Likes Received:
    165
    Trophy Points:
    42
    Location:
    North Carolina
    HA! I am going to argue a bit with Miner because when I do I typically learn something.

    Basically, a Gage R&R is a means to evaluate a measurement system. It can evaluate ANY measurement system, standard micrometers included.

    Why all the hubub? Well, they cost money to conduct and many people (not in quality) don't get the need, so they have this negative stigma. AND people tend to not like the results when they fail. Which is strange. But, it sometimes happens that way. Add to this, we all get busy. And sometimes that means we can't do the full battle rattle on everything like we should. So we set priorities and pick and choose. It is this aspect of it that leads to the statement that they are "only done on specialty gages."

    That is often true, but the reason is cost and rush to production, not invalidity of the study on normal gages. So we pick the "most difficult" gages. The gages that measure key characteristics. Do those only, and call it good. Does this work? Sometimes.

    Let's talk about micrometers. I've used them a lot. If I am controlling a ground pin with a 0.100 mm tolerance with a micrometer, I don't need a Gage R&R to tell me I'm OK. I know I am because I have experience. Does "my experience" meet requirements? It depends on your customer.

    But let's continue with micrometers..... So maybe I am new to them. And I am new to Gage R&Rs. And I want to do a Gage R&R on a micrometer to learn about how this all works. So I pick some, say, 8 mm dowel pin with let's say a 0.010 mm tolerance. A reasonable pin, not terribly tight. I do the Gage R&R and I get a pass. Now, my next product line is a 10 mm pin. I can check THIS with the same micrometer. Do I have to do a Gage R&R? A purist would say yes. I would say no. Because it isn't that different from the 8 mm pin.

    The bottom line is - the reason it is done is because you want to ensure yourself with statistics that your gage is OK. So let's assume you've been using 1" micrometers to measure dowels your whole life and you KNOW how they work and don't feel like you need to do one. Here's some "what ifs" for you:

    1) What if your product you are checking is an oval? And you want to control max and min? That's not the same as a round. That involves an operator correctly FINDING the max and min, not the same as a straight up diameter.
    2) What if you are now measuring a pin that is 120 mm? Now you need a different micrometer. Yes, the mechanism is identical. But the mic to measure this will be large and unwieldy. Not to mention the part itself will be heavy. Maybe you can't do the hold hold the part in one hand and the mic in another.
    3) What if you are measuring a run of the mill 8 mm pin, but NOW it is installed in some other part. Suddenly it's a little awkward to get the micrometer into correct position. Maybe you can't even see the readout and have to carefully drag the micrometer off.
    4) What if you have been using a 0.100 mm tolerance and suddenly you have an application requiring 0.005 mm?

    You can't necessarily say "I know this mircometer is good." Because each of my "what ifs" adds an aspect of the measurement that COULD throw the results off.

    Don't think about it from the sense of "does it apply" like there's some rule book that says use this here, and that there. There really isn't. And the attempts at writing such a rule book fail because of one size fits most. :)

    Consider your gage and what you are trying to measure. If there is any aspect that is NEW.... different tolerance, nonstandard implementation, oddly located or constructed feature, awkward handling that might not make it perform as expected - you need to do a Gage R&R.

    Consider your project - what is the risk of shipping a bad part? What is the cost, both financially and credibility wise? If your certain there won't be an issue, by all means, blow it off. It's your business. I've passed many a gage because of similarity to a feature I have already checked. Key word: similarity.

    And, last but not least, if it's a customer requirement, there you go.

    Sorry that was long. The statement that it does not apply to some type of gaging is blatantly false. I've been in arguments that they "don't apply" to hardness testers. Hogwash. It's a statistical picture of reliability. It applies to ANY measurement. All my droning on above is more to the do you need to do it or not? The answer there depends on the specifics of the situation. And again, purists will say "Yes, always." They aren't wrong. But it's not always necessary. You don't jump in your car before sunrise and haul ass east just to make sure the sun REALLY IS going to rise for your family in an hour. You just assume that it is because you are familiar with it. But if I change your latitude enough, it may NOT rise exactly as expected. So you test until you are familiar with the new situation. Think about it like that.
     
    dibdab, tony s, Miner and 1 other person like this.
  7. Sundance

    Sundance Member

    Joined:
    Aug 31, 2016
    Messages:
    5
    Likes Received:
    1
    Trophy Points:
    1
    Appreciate your insight as well ncwalker. All said and done though, it didn't sound like an argument with Miner, but more like an agreement with additional input. I am in the spring business now which is way different than the precision (aerospace) machining world that I spent years in. Generally our tolerances range from +/- .005" on the tight end to +/- 1/32" and more, with the most critical feature being the load/force factors in most instances. Considering that, it's odd that I've yet to be asked specifically to do an R&R on a force gage. Hmmm.....
     
  8. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    397
    Likes Received:
    303
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    Yes. Where was your disagreement?
     
  9. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    260
    Likes Received:
    165
    Trophy Points:
    42
    Location:
    North Carolina
    You know, I'm not sure. Usually Miner answers with absolute textbook perfection. I tend to be a bit loosey-goosey. So usually after I put something on here, Miner will hit me with a "except for this" and be right. I guess it's finally sinking in. My use of the word "argue" is in the good natured sense. Some of my favorite exchanges have been with him. We don't really argue.
     
  10. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    260
    Likes Received:
    165
    Trophy Points:
    42
    Location:
    North Carolina
    Also - do the R&R on a force gage, if only to increase your familiarity with the device. The gage will have an uncertainty value somewhere in the instructions. This gage reads to +/- 0.5 lbs, or something like that. That was arrived at under controlled situations. If you're checking coil springs and clamping one side of it, that's going to add noise. Also, the springs themselves may not behave exactly the same. Again with the micrometers - typically good to 0.0005 mm on a direct application. But how good would they be measuring the thickness of a rubber band? We'd all squash them differently.....

    The Gage R&R is NOT about how effective the gage is. It is about how effective the gage is at measuring a specific part/measurement. The parts are just as important. You can't measure a rubber band to 0.0005 with a mic. Doesn't mean the mic can't do it. Means the mic can't do it on a rubber band.
     
    Miner likes this.
  11. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    473
    Likes Received:
    514
    Trophy Points:
    92
    Location:
    Maine
    Miner likes this.
  12. Golfman25

    Golfman25 Well-Known Member

    Joined:
    Nov 6, 2015
    Messages:
    715
    Likes Received:
    352
    Trophy Points:
    62
    While it may be "wrong" I generally agree with them. Here's the problem as I see it. You check all of these off the shelf Gage's and they don't meet the "numbers." Now what. Do you spend money upgrading the Gage's to something that will pass? For what?
     
  13. tony s

    tony s Well-Known Member

    Joined:
    Sep 10, 2015
    Messages:
    1,172
    Likes Received:
    867
    Trophy Points:
    112
    Location:
    Laguna Philippines
    The results of the GR&R should also be analyzed against the capability of the manufacturing process for a specific tolerance. Textbooks will tell us that under 10% is acceptable, 10-30% is maybe acceptable and over 30% is unacceptable but if the Cp is to high (i.e. over 2), your measured values are considered reliable and will not cause wrong decisions in product acceptability/rejectability.
     
  14. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    473
    Likes Received:
    514
    Trophy Points:
    92
    Location:
    Maine
    Because if your measurement system is not adequate to discriminate good parts from bad you will reject good parts and ship bad parts. It's not just about the gauge, it's about the system including the appraiser, any fixture get and the part properties...I DO understand that the AIAG popular method typically overstates the R&Ra s the math is incorrect. (Ironically It is an unreliable assessment system). The popular method also doesn't easily enable understanding the weaknesses of your system - which is why the method advocated by Wheeler et al is so much better. (It actually predates the AIAG method as it was developed in the 1950s by Youden and his contemporaries). That weakness acknowledged the 'requirement' for performing a Gauge R&R is a solid principle of quality engineering.

    Almost worst case scenario: I worked on a Problem involving a component that when it was too large led to the equipment failing. This put many lives in danger (100+ for each event). Of course there were also substantial costs to the company operating the equipment even if no one was hurt. It came down to a hand held 'off-the-shelf' micrometer measuring a small feature on line and an expensive CMM that measured each part to 'accept' or 'reject' the part. Before I got involved everyone thought the same as the OP's advisors: both should be good enough and if not, there wa nothing that could be done about it...once the R&R's were completed both systems were shown to be incapable of measuring the parts in any reliable manner, allowing the parts to be shipped and assembled into the equipment. This inability also allowed the problem to continue unknown for several years until the first failures began. (You can imagine that the remediation of the field was time consuming, costly and posed a prolonged consumer risk). The good news is that once we investigated the R&R failures we were able to correct the systems cheaply and quickly and get reliable information. We then rapidly determined the cause and corrected the Problem (for an astonishingly low cost given the effect)

    Of course if you are making parts for products that are benign if they fail, then it doesn't matter. If the person in charge of performing R&Rs doesn't care about the quality of their parts there is a lot bigger Problem to be addressed...sorry if this comes off as harsh, but this 'quality stuff' is important. I abhor poor methods as much a s I abhor poor attitudes...
     
    Andy Nichols likes this.
  15. Golfman25

    Golfman25 Well-Known Member

    Joined:
    Nov 6, 2015
    Messages:
    715
    Likes Received:
    352
    Trophy Points:
    62
    That's all great. But if your measuring parts better than you can make them it's a waste of resources. To get the variability you need for analysis you need to go out several decimal points. So when I tell my people we are .0004 off, there's no way to adjust the process to that level of precision. If we try to make the adjustment we can loose the whole thing. Been there, done that.
     
  16. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    473
    Likes Received:
    514
    Trophy Points:
    92
    Location:
    Maine
    Golfman - not really sure what you are trying to say here - can you give us a specific example?
     
  17. Andy Nichols

    Andy Nichols Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    3,719
    Likes Received:
    1,887
    Trophy Points:
    112
    Location:
    In the "Rust Belt"
    It's not "better". If you use the 4:1 or 10:1 discrimination rule, that's NOT "better" and to use such terminology is to confuse the correct reason for being able to discriminate accurately and, therefore, not go over adjusting!
     
  18. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    260
    Likes Received:
    165
    Trophy Points:
    42
    Location:
    North Carolina
    Invert your logic. Does the statement "If you're measuring your parts AS GOOD as you can make them you are fine" make sense? And remember, if your gage is as good as your manufacturing tolerance, you have a plug gage. That's fine if you are willing to inspect 100%. But you can't do SPC then. You have no idea if you are centered or not.

    If what you are saying is "Don't waste money on overly accurate gages" then I agree. Yes, that is wasteful. I don't need micrometers in a lumber yard. But if what you are saying is your gage accuracy doesn't need to be better than your tolerance, I disagree.
     
  19. Golfman25

    Golfman25 Well-Known Member

    Joined:
    Nov 6, 2015
    Messages:
    715
    Likes Received:
    352
    Trophy Points:
    62
    What I am saying is that, yes to do all of these studies you have to find variation. So you use gages that are 4:1, or 10:1 or whatever. And I run my gage R&R and it tells me I'm "no good" because I'm not 10% (or some other arbitrary number). Or my CPK is not go because I am .0002 "shifted." Now everything stops because the numbers are "no good." Now instead of the $150 off the shelf mic which as been good for years, I need a $5,000 super duper measuring thinginy. All the while, if I ask my production people to reduce the variation by a couple of tenths so the numbers "work," they laugh.

    Now granted, some industries need a certain level of precision. And variations, measurements, and such are really important. But others don't. Our parts have some flex in them and are press fits. So a few tenths really doesn't make a difference. So measuring to all these "numbers" has no value. Don't apply the same criteria across the board. Use a little thinking and common sense. But stuff rolls down hill and it is easier to take the "rule" from the AIAG book and pass it on.
     
  20. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    260
    Likes Received:
    165
    Trophy Points:
    42
    Location:
    North Carolina
    That is much more clear. And I agree with that. Let's all not forget that the other "half" of the equation in both Cpk and Gage R&R is the tolerance. If we increase the tolerance, both will get "better."

    More than once, I have approached a production process, done the hard work, and found that the gage failed Gage R&R AND the Cp was too high. And yet, the product line had been running for years with no issues. You look on the print and here's this tolerance of +/-0.005 mm. Call up the engineer and say "Why so tight?" and the answer is invariable "Dunno, was on the print I copy/pasted the design from." With no statistical experimentation.

    In those cases I:
    1) Get a WORKING gage on the floor.
    2) Monitor the process for a long time. (Long time until I am satisfied I have captured the variation).
    3) Calculate Cp/Cpk (Yes, still low)
    4) Calculate the tolerance needed that WOULD allow them to be 1.33
    5) Go back to engineering and say "Open this up. You're getting this anyway and have been for years."

    Sometimes it works. :)
     
    Candi1024 likes this.

Share This Page