iSixSigma

Occurrence in DFMEA

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #53827

    Indzhov
    Participant

    Hello All,
    I’m new in FMEA, but I have one question.
    How can we say what is the frequence of occurrence when we make DFMEA and the product is still under disign? (When we have no data about frequence of occurrence!)
    Sorry, if my question is stupid but I wasn’t able to catch it.
    Thank you beforehand for Your all comments!

    0
    #191590

    Strayer
    Participant

    The occurrence rating is not derived from data about actual failures. FMEA is used to identify and prioritize “potential” failure modes. The team is asked to identify everything that “might” go wrong and then, using group wisdom, to agree on the ratings for likelihood of occurrence, potential impact, and detectability so that the RPN reveals the potential failure modes that most demand attention.

    DFMEA and FMEA are the same in that both deal with identifying and prioritizing what might go wrong, not what actually went wrong. The difference is that DFMEA can only consider designs and models since the process does not yet exist. Particpants in an FMEA have knowledge and experience of what actually happens, or might happen, but the exercise is still speculative. If a failure mode is known to have occurred, other tools are much better suited for root cause analysis.

    0
    #191591

    MBBinWI
    Participant

    Straydog wrote:

    DFMEA and FMEA are the same in that both deal with identifying and prioritizing what might go wrong, not what actually went wrong. The difference is that DFMEA can only consider designs and models since the process does not yet exist..

    Regardless of whether the process exists or not (your statement opens up the argument that if it does, then you cannot do the DFMEA), you should always approach the DFMEA with the thinking that “the process is going to produce exactly what the design calls for – and at the extremes of the tolerances specified. Given that the design will be faithfully executed (including tolerance extremes), what can go wrong?

    By assuming that the design will be faithfully executed as specified, it gets the team better focused on what is being specified and off of how it will be produced.

    BTW – while you are technically correc that the occurrence rate is an estimate, usually you can find similar designs or go back to first principles with evaluations at tolerance extremes and get very good estimates.

    0
Viewing 3 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic.