By Jon M Quigley
We have discussed the Failure Mode Effects technique a few times in the past. Though Failure Mode Effects and analysis seems to be a powerful tool, the problem is you do not know if the FMEA is effective and perhaps you will never know. The Failure Mode Effects Analysis tool, theoretically, allows us to critique the product before we test. How do we prove something did not happen because of a well conducted FMEA? I have seen failures arise in testing and post launch. In both instances I found it difficult to believe we missed the issue in the development or testing work as the failure seemed obvious. I would later discover the FMEA was either missing or performed with a check box mentality, both of which are poor execution and not necessarily a testimonial of the apropos of the technique. Of course this is only anecdotal evidence.
For me the real benefit is when we wed the FMEA to our testing. We theorize something in the FMEA, a particular failure will have a specific set of consequence, but we may not really know. We set about testing the more egregious failures. The severity of the failure may not be what we have attributed. The same is true for the occurrence. Perhaps what we thought was going to be an insignificant probability of occurrence, upon testing; we find the failure to be much more common. We are now in a position to take action based upon this evidence.
As Kim Pries say, it takes sound design and harsh testing. However, I would add harsh critiquing goes along with sound design and harsh testing. Unfortunately, that is tough to come by and FMEA’s can descend into what I call a bunch of glad handing due to poor execution.
It is certain if there are benefits from doing an FMEA, if you do not them or do them poorly those probably benefits disappear.