iSixSigma

Gage RR

Viewing 12 posts - 1 through 12 (of 12 total)
  • Author
    Posts
  • #48279

    Sinnicks
    Participant

     
    The back ground to the problem is that I have been challenged with conducting a Gage R&R, the issue is that it does not follow the format that I recall as a good study i.e. 3 operators with 10 parts if my memory serves me well?
     
    What we have is 4 operators with a maximum of 2 parts per week, this would mean the study would run into many weeks. I have asked the Process Engineer to create a template so that the operators measure in the same place every time and Work Instructions on how the Process is to be executed. The peice of equipment being validated is a CMM arm measuring coils.
     
    The main question here is that is it ok for the measurements to be collected over many weeks?

    0
    #162150

    Chad Taylor
    Participant

    Mark
    However long it takes to get a representive sample of the process condusive to run the R&R is OK. If it takes Weeks or Months. So the answer is yes.
     

    0
    #162160

    Sinnicks
    Participant

    Chad,
    Many thanks, it will make the process alot easier.

    0
    #162171

    annon
    Participant

    M,
    In the interest of time and potential problems with a long-term approach, is it not possible to select parts already inventoried for your study? 

    0
    #162174

    Dennis Craggs
    Participant

    You don’t need the constraint of 3 operators – 10 parts. Any GR&R can be considered an ANOVA. In other words, you are trying to assess the contriubtion of operators, parts, and measurements to total variation. ANOVA GR&R is an option discussed in the AIAG Measurement Systems Analysis Manual. Also, Minitab has an ANOVA GR&R.

    0
    #162284

    Sinnicks
    Participant

    Annon,
    Many thanks for your reply, unfortunately some might say that due to the business implementing a “pull” system and running Kanban our inventory levels are always kept to a minimum.
    Mark

    0
    #162286

    fake accrington alert
    Participant

    You  may  use  Excell as  well

    0
    #162313

    annon
    Participant

    You can run a test-retest using a single part as a reference value to determine the accuracy and precision of the gauge.  I would feel comfortable using 5 parts that represent the full range of output for your process (including values slightly outside of your tolerance) to conduct your GR&R.  Select them from your limited inventory for a brief period of time for the GR&R or have them manufactured outside the production schedule (if possible). 
    Good luck.

    0
    #162341

    Erik L
    Participant

    Mark,
    Ideally a Gage R&R should be conducted over a fairly narrow window of time.  The results of the variance components are typically interpreted as an ‘ideal’ of the best your measurement process ‘could’ deliver.  Sort of like capability potential.  However, if the study needs to conducted over a lengthy window, I would definitely take the advice of obtaining a ‘standard’ and assessing any accuracy/bias issues as your study evolves.  With a baseline established, you could then continue until you feel that you’ve captured a representative sampling of your product.  As the new units become available for the study, re-run the unit that is the standard to ensure that there has not been any drift/shift in the measurement process.  This would be similar to running center points in a DOE within a blocking term to assess the presence of a background factor biasing a key response that you are experimenting on.  As long as there is not a significant shift in the standard the results you get from the analysis should be valid to assess the measurement process.
     
    Regards,
    Erik

    0
    #162975

    Dreemr
    Participant

    I have asked the Process Engineer to create a template so that the operators measure in the same place every time and Work Instructions on how the Process is to be executed
     
    I have more concern with this statement that the time period for your gage study.  Are you suggesting a process change form the normal run of things for your gage?  If so, the setup and process used for the gage should be adopted for “normal” production measuring as well.  
     
    The piece of equipment being validated is a CMM arm measuring coils.
     
    I have had the opportunity to test a couple of different CMM machines in my time as a black belt.  I have had only one that passed on the first gage attempt.  Let the CMM operator do what they normally do the first time you run the gage.  Make sure that the parts are actually removed from the CMM table between measurements.  This will force a setup for each measurement.  Most operators will resist this, as they see the repetitive setups to be a waste of time.  I my past studies this has been the largest contributor to variation and has caused a failure in every first gage attempt except for one.  The learning derived from failing that first gage and passing subsequent gages by implementing a SOP is invaluable.

    0
    #162981

    gt
    Participant

    You have to select a representative sample of the process output. As far as number of operators aqnd parts, sometimes you have to deal with what you can get. there certainly a way to get parts from previous runs. i’ve conducted R&R in all possible combination (low volumes, high volumes, non destructive, destructive, 1 to 3 ops, 1 to 3 repetition, 5 to 10 parts ). You have to think of the impact on the business too. (cost to conduct the study, time to conduct the study,…) would you wait 6 months to get a proper sample when you can assume you ship bad parts to customer?
     
     key metric: number of categories. and standard deviation of the study-measurement system
     
      

    0
    #163019

    Sinnicks
    Participant

    Many Thanks to all that have replied

    0
Viewing 12 posts - 1 through 12 (of 12 total)

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