iSixSigma

What happens if MSA fails?

Six Sigma – iSixSigma Forums Old Forums General What happens if MSA fails?

Viewing 13 posts - 1 through 13 (of 13 total)
  • Author
    Posts
  • #47053

    Ruben
    Member

    Hi, I’m wondering if anyone out there has performed an MSA and discovered that the variation in the measurement system was so wide, it brought into question previous measurements that resulted in “passing” a test when perhaps it should have failed. From a business perspective, what have people done at this point? Thanks, -Ruben

    0
    #156378

    Chris Seider
    Participant

    You are not the only one to run into a gauge that fails.  The short answer on what to do next is “fix it”.
    However, that might be too callous.  Look at if the source of variance is the repeatability or reproducibility and let that lead you on how to proceed to fix the measurement system.
    Don’t forget the idea of signal averaging and taking 4 measurements and taking the average to reduce the variance but a theoretical two. 

    0
    #156383

    Gaurav
    Participant

    Hi Ruben
    Firstly what you mean by MSA FAils, Is it P/T% >30%, or %GRR>30%?
    Now it you are saying that P/T%>30Percent we need to really look on our measurement system. so do PF/CE/CNX/SOP if nothing could be done then go for new Measurement system as P/T is improtant from Customer point of view i.e. how good we are is fulfilling customer requirement.
    In case P/T%30 you can be bit relax. as the system is generally said to be acceptable, as you are able to satisfy Customer requirement but system is not gud in measuring process variation.
    In this case again you sud go for PF/CE?CNX/SOP then its the managment decision what sigme level they want to achive and wether they are interested in investing in new measurement system.
    So Customer requirement is the first then its Process requirement
    Hope it helps
    for detail u can revert
    Cheers

    0
    #156387

    Vidyadhar
    Member

    IF your MSA has failed…. try and find out what that caused variation in the measurement system… Validate your data on MSA failure due to these causes..
    Before taking a call on revamping the measurement system, try if there are any obvious reasons of failure which could be fixed without incurring additional costs…
     

    0
    #156402

    Ruben
    Member

    Thank you all for your advice…I find it very useful. While it is useful to follow up on what is causing the measurement variation to be wide, I’m finding a larger issue is not a technical one but one of perception–the belief that once we find out our measurement variation is wide and given that operate close to our spec limits, we may have said certain items “passed” when in fact they should not have, or certain items had “failed” when in fact they should have passed (I suppose you can call these Type I, II errors in business decisions). Has anyone successfully addressed management on how to address the validity of past items where the status is now questionable?
    Thanks,
    -Ruben

    0
    #156761

    Jonathon Andell
    Participant

    First of all, quantify the degree of variation due to the causes: process, measurement, etc. Make sure you nail those down as best you can.
    Then estimate the quantity of errors or each type. You may want to employ a Monte Carlo simulator to generate an error band here.
    Next, consider the ramifications of each error type. Is the worst outcome limited to mere dollars, or is it as severe as loss of life?
    Finally, evaluate what options you have: everything from “stay the course” through recalls, through active interventions on behalf of customers at risk.
    This can be pretty advanced stuff. It’s better to bring in an expert if the stakes are high.

    0
    #156770

    Carmine Ciriello
    Participant

    If I understand your question correctly, you are concerned with the Change Management aspect of your failed MSA rather than the technical side. I had a case of an analytical procedure that was eventually found to be very poor (through an MSA) for a specific metal assay. The ‘selling’ of this problem was done in several steps, but started with a very simple and clear table / graph of the weakness of that particular system. Do not talk MSA with upper management, as you can quickly lose them – talk $ or % rework. Also, make sure that your MSA was done properly, especially if the implications of your findings can be very damaging or costly (ex: have it reviewed by a peer or expert). In my case, the ‘past’ was not an issue, as we focused our energy on how to fix the problem for the future.
    I do not know the specifics of your problem, but generally, the past is not as critical as the future: (a) Rejects that you passed: should have been associated with high # of customer complaints – is that the case? (b) Good items that were rejected: this means your company absorbed larger rework costs but you can fix this for the future by addessing your measurement system; (c) Finally, you have to accept a certain number of Type I and II errors – try to get a feel for what is acceptable in your industry before going too far.  Hope this helps.

    0
    #156772

    mrkidbb
    Participant

    I think that everyone but Jonathan has missed the point of the question.  I believe that the point is, “What do I do if I know that I have probably sent bad product to my customers in the past (and may be sending it in the future, until I get my measurement system fixed)?”  Ruben, it sounds like you may work in an industry where this could be more than just a financial / customer return issue.  I work in the pharmaceutical industry, and let me tell you, it is a lot bigger deal than, “Go fix it!”  If we are sending product out to consumers that is questionable from a safety or efficacy perspective, we are potentially putting lives at risk.  The FDA doesn’t look too kindly upon that to say the least, and they also don’t look too kindly on companies who know that this has happened / is happening and ignore it.  I don’t know where you sit within your organization, but this is definitely something in our organization that would be elevated to the highest levels of company management to determine whether it necessitated a product recall.

    0
    #156774

    e
    Participant

    A little extremist posting isn’t it?….
    Maybe you had a bad day.  I didn’t see anything about the gage being an output of the entire process NOR did I see it being a life threatening situation.
    just my take..

    0
    #156779

    CT
    Participant

    The question is a little broad, but think we can get there. First, what is the severity of the “potential” defect? Have you received customer complaint? I would probably issue a statement to the customer stating that in a recent effort to improve our quality system we have found that our current measurment system has the potential to pass product that may not meet the customer requirements. State that you are working to remedy the issue and give specifics. Follow up with Data that proves the situation is indeed “fixed” or at least risk reduced. Many companies deal with this situation by asking for a Deviation during improvment process. Deviations can get wicked out of control, so be careful. I have seen deviations run for the life of a product. Never good, but can happen.
     
    CT

    0
    #156780

    Ruben
    Member

    Carmine, that’s exactly what my question is about, and your comments help me tremendously! Thanks for sharing this.

    0
    #156781

    Ruben
    Member

    Actually, I work in the Pharmaceutical industry as well, so your thou hts are both valuable and relevant. Thanks!

    0
    #156788

    BTDT
    Participant

    Ruben:Put together your business case in terms of:1) Bad product released (potential liquidated damages) – type II error (beta)2) Good product reworked (rework and scrap costs) – type I error (alpha)3) in general a SOX violation of insufficient controlsUse the results of your Gauge R&R to calculate the proportion of type I and type II for historical shipments and calculate the dollar impact of the past exposures and rework.Carefully present the process improvement in terms of the financial impact of a SOX violation. I say carefully because it would be poor change management to have the project make it look like you are a whistle blower.Cheers, BTDT

    0
Viewing 13 posts - 1 through 13 (of 13 total)

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