1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
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.

GR&R Requirements

Discussion in 'IATF 16949:2016 - Automotive Quality Systems' started by gkoenig, Mar 17, 2016.

  1. gkoenig

    gkoenig New Member

    Joined:
    Mar 17, 2016
    Messages:
    1
    Likes Received:
    0
    Trophy Points:
    1
    Hello,

    I completed a PPAP for a customer part that has several critical characteristics. A GR&R was performed for a 1" micrometer that measures one of these critical dimensions. All other dimensions, including the remaining critical, are measured by an optical system, to which a GR&R was also performed.

    The customer is requiring GR&Rs to be performed for all critical characteristics. It seems to me that this is redundant since the same equipment is being used and a GR&R was already performed. The equipment has recently been calibrated and has been in spec. The equipment does not know the difference between one measurement or another.

    Are separate GR&Rs for each critical dimension necessary or is one GR&R study adequate? I would appreciate any input as I have had a hard time researching an answer.

    gjkoenig
     
  2. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    576
    Likes Received:
    492
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    It depends on the measurements themselves. For example, the ability of the 1' micrometer to measure distances between flat surfaces within a limited range should be virtually identical. However, the ability to measure a diameter may be strikingly different from the flat surfaces. Same with the optical system. Discuss with your customer the possibility of grouping flat surface distances together, diameters together, angles together, etc. You may want to perform a few studies within the same group to provide data to help convince your customer.
     
  3. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    261
    Likes Received:
    168
    Trophy Points:
    42
    Location:
    North Carolina
    Hi. I am a typical customer. I will echo Miner's sentiment. If you are giving me, say, some sort of shaft with several bearing journals all different (slightly) diameters. And you are going to check them with the same micrometer, because it can, I would be OK with you doing an R&R on the one with the tightest tolerance. Because that makes sense.

    However, if you are then going to use the same micrometer to check a part that is awkward and very unwieldy, I will want another Gage R&R. Why? Because everyone can hold a shaft easily. But an awkward, out of balance part requires a little more care just because it is unwieldy.

    What I am trying to say is it is reasonable to say a gage is OK a priori IF the circumstances of measurement are reasonably similar. The key is understanding if different measurement circumstances will or will not affect the reliability of the gage.

    Here's another example - Go back to the shaft ... if ONE of my journals requires a 2" micrometer, I may be OK with NOT doing a Gage R&R on that. Because a 2" micrometer is similar to a 1" micrometer. But if there is a flywheel on the end of the shaft that requires a 6" micrometer we have a different story. Comparatively, a 6" mike is much larger than a 1" or 2" mike. You can't manipulate a 6" one handed. It's a different situation.

    The perfect world situation is you do everything. In the real world, cost and timing gets in the way. Ergo you need to have reasonable discussion with your customer.
     
    Steve Davidson, shndcb and Roberticus like this.
  4. Georgi Tsonev

    Georgi Tsonev New Member

    Joined:
    Mar 4, 2021
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    1
    Hello,

    I would like to share with you a research which I have done these days.

    Based on the materials in AIAG and Minitab I have decided to investigate the minimum number of parts that can give us reliable results regarding GRR Study variation.

    Number of parts: As a base, I have followed the article in Minitab: https://blog.minitab.com/en/fun-with-statistics/gauging-gage-part-3-how-to-sample-parts

    which is advising to have a random selection of the parts. So I have started with 30 parts randomly selected from the normal distribution generated in Minitab.

    Number of appraisals: I have started with 5 appraisals and I didn’t add any error coming from the appraisals in order to keep one influencing factor of the results constant. For the selected parts, for each appraisal measure, I have generated a random value using a normal distribution with the same standard deviation for all appraisals.

    Number of trials: I have used maxim 5 repeated measures per part

    Then I have done a DOE using the following control factors: number of appraisals, number of trials, and number of parts(3 levels per factor to capture curvature). As an output, I have done GRRs and collected % study variation and its confidence intervals for it. This gave me the possibility to calculate the difference between lower and higher value and use it in the analysis(I have called it Variation)

    After the DOE analysis, I have found that the factors which have an influence on the variations are the number of appraisals and the number of parts. It is not adding a big benefit if you do more than 2 trials. Based on the results I have created the following graphs(attached)

    What was surprised me that we have bad results in the area which was proposed by most of the theoretical studies so far as “Concepts for R&R Studies” book of Larry B. Barrentine, where it is advised that 2 trials and 8 parts with 2 operators are enough.


    My question is did you manage to find the same behavior in your practice or maybe I have done a wrong conclusion?
     

    Attached File(s): 1. Scan for viruses before using. 2. Report any 'bad' files by reporting this post. 3. Use at your own Risk.:

  5. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    605
    Likes Received:
    663
    Trophy Points:
    92
    Location:
    Maine
    I have empirical experience with this for the last 25+ years. You are correct that we do not improve our estimate of the measurement error by increasing from 2 to 3-5 repeated measurements. This is also substantiated by the formulas for the error associated with the standard deviation. Using 10 or fewer parts can substantially mis-estimate your part to part variation and more importantly provides a very small sample size for the number of repeated measurements that can skew your measurement error estimate OR you may simply miss the important differences across the range of part variation. (While this is rare with measurements fo dimensions adn features it si not when the measurement is a property or function). You can also read the works of Donald Wheeler (go to qualitydigest.com or simply “google” him) who also provides the theoretical and empirical background on performing and understanding measurement system studies.
     
    Georgi Tsonev likes this.
  6. Georgi Tsonev

    Georgi Tsonev New Member

    Joined:
    Mar 4, 2021
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    1
    Thanks for the comments and the advice. I am planning to do dig more on the topic and if you will share the results here.
     
  7. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    605
    Likes Received:
    663
    Trophy Points:
    92
    Location:
    Maine
    I’m not sure what you mean by ‘share the results’. You can find some of my papers on the topic in the resources section here under “practical quality engineering resources”. In there I do have several references to Donald Wheeler’s works and other authors.
     
  8. Georgi Tsonev

    Georgi Tsonev New Member

    Joined:
    Mar 4, 2021
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    1
    Share results mean that I will write them here for discussion. :)