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.

Gage R&R and R chart interpretation

Discussion in 'Gage R&R and MSA - Measurement Systems Analysis' started by lello_1983, Jan 26, 2017.

  1. lello_1983

    lello_1983 New Member

    Joined:
    Jan 26, 2017
    Messages:
    1
    Likes Received:
    0
    Trophy Points:
    1
    Hi,
    First of all I would like to thank you for the great community you have here!!

    I would like to have a clarification about the interpretation of Gage R&R results, used to evaluate the adequacy of a measurement system;
    In particular, I would like to better understand how to interpret the average Rbar value shown in the R chart;

    How can I use the R value of the R chart in order to understand if the measurement system is able to discriminate between different parts?

    Thanks a lot in advance!!
     
  2. Englishman Abroad

    Englishman Abroad Member

    Joined:
    Mar 29, 2016
    Messages:
    25
    Likes Received:
    4
    Trophy Points:
    2
    Lello,

    The R chart itself can not tell you much about the measurement system. It can tell you whether there are any non random (strange) values in your values.

    The R limit is the Control limit of the range of values measured by each operator for each part. If you have a value above the R limit then you should investigate why before proceeding with using the values to calculate GRR etc.

    Hopefully Miner and Bev D will add something to my very simple explanations, but look at Miners Blog on MSA, to have some good background.
     
  3. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    605
    Likes Received:
    663
    Trophy Points:
    92
    Location:
    Maine
    The range chart can really only be used to detect 'outlier' differences between repeated measurements that might indicate an invalid measurement or data entry error. As Englisman Abroad stated...

    The Range chart itself doesn't really enable you to understand the 'discrimination' of your measurement system. For a better graphical display of R&R results try the Youden Chart. I describe this in my paper of MSA in the Resources section of this forum. Miner also covers this in one of his MSA blogs...
     
  4. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    261
    Likes Received:
    168
    Trophy Points:
    42
    Location:
    North Carolina
    Cautionary statement: Bev is correct about the Youden Chart. But don't just gloss over it. There is a LOT of information condensed in the thing. To master it, I coded the darn thing in Excel then ran a bunch of different R&R scenarios in it to see what it did. In other words: what does it look like when the measurements get noisy? What does it look like if Operator B gets bias relative to A and C? What does it look like if ONE measurement has an outlier (like from a typo). It shows all those things. But if all you do is look at good examples, you won't train yourself well to spot the bad ones.
     
  5. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    576
    Likes Received:
    492
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    In addition to the preceding comments, the Rbar value is not typically used in isolation to interpret the measurement system. It is an intermediate calculation that can be used to calculate the various control limits as well as the standard deviation for Repeatability, although I recommend using the ANOVA method instead.

    The R chart is used to visually spot differences in Repeatability between operators. It can due this by flagging outliers, or runs by operator.
     
  6. tony s

    tony s Well-Known Member

    Joined:
    Sep 10, 2015
    Messages:
    1,350
    Likes Received:
    1,054
    Trophy Points:
    112
    Location:
    Laguna Philippines
    The R Chart can provide information whether the measurement system has an acceptable discrimination. If your R-values in the chart can be grouped in 5 or more data categories, then your measurement system has an acceptable discrimination. The example below shows that the 2nd chart has better discrimination than the first one with only 3 data categories (i.e. 0.0, 0.1, 0.2).
    upload_2017-1-31_18-25-37.png
     
    Miner likes this.
  7. ncwalker

    ncwalker Well-Known Member

    Joined:
    Sep 21, 2015
    Messages:
    261
    Likes Received:
    168
    Trophy Points:
    42
    Location:
    North Carolina
    I am honestly confused now. tony is this an "Rbar" chart or a "run" chart. I'm not sure what I'm seeing with the blue dashed line at the bottom. Doesn't Rbar have one control limit, the upper? It could very much be formatting, I don't recognize the package.
     
  8. Bev D

    Bev D Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    605
    Likes Received:
    663
    Trophy Points:
    92
    Location:
    Maine
    It looks like EXCEL - those are the old default pink and blue markers and lines...it is possible for the range chart to have a lower control >0, but the limits on the upper chart certainly don't look right!
     
  9. Miner

    Miner Moderator Staff Member

    Joined:
    Jul 30, 2015
    Messages:
    576
    Likes Received:
    492
    Trophy Points:
    62
    Location:
    Greater Milwaukee USA
    While the charts are not great examples, tony's comments are correct, and I am glad he pointed them out. It's something I automatically check for, but forgot to point out. One additional requirement is that no more than 25% of the values can be zero.

    This is what tony's was trying to portray. I used the AIAG dataset, truncated the Operator A data at 1 decimal place, Operator B at 2 decimal places, and left Operator C at 5 decimal places.
    The resolution of 0.1 only provides 2 possible values under the UCL_R. Unacceptable resolution. The resolution of 0.01 provides 13 possible values under UCL_R, so this resolution is acceptable. Obviously 0.00001 provides many more possible values under UCL_R, so it also is acceptable and probably overkill.
    Elsmar2.jpg
     
    ncwalker, Bev D and tony s like this.
  10. tony s

    tony s Well-Known Member

    Joined:
    Sep 10, 2015
    Messages:
    1,350
    Likes Received:
    1,054
    Trophy Points:
    112
    Location:
    Laguna Philippines
    Honestly, I just used a simple MS Excel line chart just to drive my point that measurement systems with more acceptable discrimination can be viewed in the R-chart.;)
    Miner thanks for providing a better presentation.

    Having less data categories can also result from rounding-off the R-values. The top chart was actually the rounded values of the bottom chart that I presented.
     
    Last edited: Feb 2, 2017