iSixSigma

what is right way to explain GRR

Six Sigma – iSixSigma Forums Old Forums General what is right way to explain GRR

Viewing 18 posts - 1 through 18 (of 18 total)
  • Author
    Posts
  • #31759

    Leon Zhang
    Participant

    I have a question about interpreting GR&R result.
    My mind:
    when we say GR&R is 10% and  that measurement system is acceptable.  this conclusion is only applicable on the charateristics which are covered by  the GR&R study?
    Am I right? if not . what is it? because I found so many instrument’s descrimination only half of tolerence of some parts. but the guys still say the GR&R is aceptable.
    I am really confusing about that.
     

    0
    #84090

    Jamie
    Participant

    Leon, Gage studies can hardly to boiled down to one number. The true complexities can not be easily covered in one post, but let me give you a few thoughts. First I must ask this what exactly is the 10% you report (always best to be clear). Is it percent contribution, percent study or precision/tolerance. All three are reported as %’s and will typically yield very different value, make sure you are clear on what is being reported.
    Now once you know this, you can interpret it within the context of the gage study. Yes the percent contribution and percent study apply only the the range (really variation) you tested in the gage study itself. If the range of values (really the variation) for the gage study are different then the process then you can not apply this percent directly back to the process. So if you have chosen to measure parts that vary significant higher then the process this number will appear much better then it really is, the converse is true. A common mistake is to measure different products and include them in the study, this causes the part variation to be huge and conversly make these percents look very small.
    Two good number to look at are the precision (5.15 * S(gage)) and precision/tolerance. These values are not influenced by the range of parts measured. These often tell me more then any of the previous percents that were described.
    Hope this helps, Jamie

    0
    #84102

    Gabriel
    Participant

    I hope this mental example helps.
    Let’s imagine that I had a 200mm digital caliper that, if any people measured repeatedly a part of true size X, delivered a normal distribution of sigma = S(r&R) = 0.010mm, and that that was true for any characteristic of any size X of any product type.
    In this case S(r&R) would be a characteristic of the instrument, but not %r&R. You wouldn’t need to make the study for every characteristic of every product, but you wuold need to calculate %(r&R) for each characteristic of each part knowing S(r&R) together with the tolerance (to get the % of the tolerance) or the process variation (to get % of the process). Of course, the tolerance and the process variation will be different for each characteristic of each product, and hence the %r&R will also be different. So the 10% you mentioned would be the same only for characteristics with the same tolerance width and/or the same process variation.
    Now, the bad news. Give me any real life 200mm digital caliper and I bet that S(r&R) will not be constant for any characteristic of any size X of any product type. It will be different for 10mm and 180mm, for spheric features and for flat features, for steel parts and for plastic parts, for outer and inner dimensions, etc. In that case you would need to make the study for each characteristic of each feature. Yet, when you measure similar characteristics of similar size and similar in similar products from similar processes, many times you can use the same S(r&R). However, and here is where priorization appears, you don’t use SPC charts, process capability studies, poka-yoke, etc for every characteristic of every product, do you?
    And as Jamie said, %r&R is not the only number that would make your instrument acceptable or not. You also have bias, linearity, stability and discrimination. An instrument with a discrimination of 1/2 of the tolerance (as you said) is unacceptable for most purposes.

    0
    #84126

    Leon Zhang
    Participant

    Thanks  Jamie and Gabriel
    you make me confident about GR&R. I work in a mechanical dominated copmany,  I found too many unacceptable GR&R here. but no body is clear about what is right . I think only useful numer is R&R(sigma) value to evalute the diviation of measurement system. but no guys believe and say that it is impossible to get an acceptable GRR% .
    In our case .we measure hardness, dimension of products in most of time. but around 50% of SPEC for these characteristics are half of descrimination of instrument. for example, the descrimination is 1 HRC for hardness tester. but the spec is  2 HRC . some time they use different parts to get good GR&R. and also no one know the difference clearly among percent contribution,percnet study. or precison/tolerance.So guys may choose an accepable value. I have a MSA handbook from FORD,GM, and Chrisler. i had read it but still got confused on the issue I mentioned.
     
     

    0
    #88927

    Ashman
    Member

    How long we are required to perform GR&R once? 6 month? 1 Yr?

    0
    #88930

    DaveG
    Participant

    GRR is one tool in a Measurement System Analysis (MSA), which is used to initially and periodically evaluate your measurement system capability.  Think of measurement as a production process like any other for which you would:
     
    ·        Do an FMEA to identify sources of variation and failure
    ·        Implement controls for the above
    ·        Periodically check the output with a Control chart or equivalent
     
    I recommend, as a minimum:
     
    ·        Do the GRR at each calibration cycle
    ·        Do stability checks at least weekly
    ·        Apply all the other MSA tools accordingly
     
    http://www.aiag.org/publications/quality/dcxfordgm.asp

    0
    #89108

    Patrick
    Participant

    Thank you for your information.

    0
    #89632

    Adithia
    Participant

    Could you give me further explanation how ofthen we should perform GR&R study and why?

    0
    #89634

    DaveG
    Participant

    MSA, of which GR&R is a subset, is SPC for a measurement system which determines if a measurement system is capable of accurately and precisely yielding true values of a measurand.  Initially, it is done to decide if the measurement system may be put into service:  the 3 outcomes are 1) accept, 2) modify the system and redo the MSA, or 3) reject the system.  After the system is accepted into service, periodically check the output.
    [email protected]

    0
    #162394

    Ashley
    Participant

    Can GR&R be perform on holding fixture?    Holding fixture here means act as a jig to hold the part for measurement. 

    0
    #162395

    GE
    Participant

    If your measurement variable is a continuous data type then yes a GR&R else if discrete use an AR&R

    0
    #162400

    Ashley
    Participant

    My understanding GR&R is intended to measure the amt of variation in the measurement system arising from measuring device and the people taking the measurement.  And u will repeat if there is a new operator or if we suspect there is inconsistency in the operator.  As holding fixture is use only to facilitate the measurment,  I am not sure if there is a need to perform a GR&R there is no change to the operator and measurement device. 
    Btw, what is AR&R? 

    0
    #162401

    Szentannai
    Member

    Hi Ashley,
    we did a GR&R once where we found that the difference between operators (a quite big one at that) was due to the different styles they had when fixing the part. One of them tightened the screw as far as it went and caused a measurable deformation in the part, the other did not.BTW we were told not to bother with an R&R in the first place because the measurement was automatic… It sure was but we still found this big reproducibility issue due to the holding fixture :).Regards
    Sandor

    0
    #162402

    Ashley
    Participant

    Hi Sandor,
    Thanks for sharing.  What abt 2nd fixture of the same kind?  Do u still conduct GR&R using the same measuring device and operator?  If no, how do u justify there is no need for one.
    Warmest regard..Ashley

    0
    #162406

    Taylor
    Participant

    Ashely
    First establish Bias and Linearity for each fixture. Once you have established that Bias and Linearity are acceptable, a Gage R&R for each fixture may not be needed. Perform a normal Gage R&R and if your results are marginal, you may want to dig further and perform multiple R&R’s to be certain of the issue. This can sometimes become a tedious task, but in the end will assure gage function for the product.
    Chad

    0
    #162436

    Szentannai
    Member

    Hi Ashley,I actually would do the GR&R for the second fixture as well – although logically there seems to be little justification for that.
    On the other hand you might find something interesting and the effort of making the test seems to be small once you do it for the first fixture anyway – so in the worst case you are no worse off then before but in the best case you learn something useful – quite a bargain in my book :).Regards
    Sandor

    0
    #162441

    Mikel
    Member

    The fixture is part of the measurement system. Do the R&R part and see what it tells you. I agree with Sandor, some of the worse measurement capabilities I have ever seen was on conputerized half million dollar equipement – the fix was in the fixturing.
    If you don’t know what to do with the R&R result, post it here. You’ll get help.

    0
    #162495

    Ashley
    Participant

    Hi, thanks for all the valuable inputs.  I really appreciate it. 

    0
Viewing 18 posts - 1 through 18 (of 18 total)

The forum ‘General’ is closed to new topics and replies.