FMEA - DFMEA and PFMEA - nobody knows how it works 1

[umatrix]

I’m curious to see others experience with the FMEA process per AIAG, under the broad scope of the APQP process.

I am under the impression that companies say they align with this to be compliant, however it’s has been my experience that the subject is often very misunderstood. What typically ends up happening is engineers will argue and debate and then finally settle on something no one agrees on. I am amazed how few have studied official process per AIAG, or even know about it. Most try to tackle with their own internal logic, which can be a good trait, but not necessarily aligning with the process.

Is this another topic like GD&t, where it is so confusing that it ends up convoluted and being boiled down to an incorrect simplicity, that often just gets pushed under the rug, due to the investment cost (time, money, interest) of learning it properly?

 

[EdStainless]

One of the problems with FMEA et.al. is that often on one side you have people that have studied the AIAG to death but have never had to apply it so something that they were making, and on the other side you have people that have been making a product for decades and are trying to figure out which data needs to go where.
They are simply speaking different languages.
IF the two sides work together from the start (not just throwing things over the wall at each other) you can actually get it done.
I have worked with aerospace companies that have their own inhouse systems for things like this. Those can be a nightmare.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, consulting work welcomed

 

[MintJulep]

In my experience, historically most of these analysis are done by a small group of number crunchers, locked in a closet in the basement and isolated from the project team at large.

The analysis are done because they are a contractual deliverable, not because management has bought in to the potential value they could add.

They are started too late, long after the analysis results could be used as design inputs.

 

[TheTick]

I've actually been part of a few good, effective DFMEAs and PFMEAs. Like most processes, it's about as good as the people executing it.

At a minimum, it's important that the effort be led by someone knowledgeable of the FMEA process, and that the team include designers and/or process people directly involved.

Part of FMEA competence is recognizing the difference between "D" and "P". If you're throwing in "what if they make it wrong" into the DFMEA ("D" = "design"), then you're doing it all wrong. Likewise, design should not infiltrate the process side.

The places I did FMEA had clearly documented criteria for scoring. The importance of this can not be overstated.

The fact that doing FMEAs is an epic drag can also not be overstated.

 

[GregLocock]

We used to dread doing FMEAs until one day they sent over a senior manager who happened to be rather good at FMEAs. He showed us how to use them in a sensible fashion, rather than as a box ticking exercise.

At one company I had a coworker who had worked in aviation, and he practically had a stroke when it was announced that we needed to do an FMEA. He then described how they did it in aerospace. It was almost exactly the opposite of what I'd been shown, basically bottom up instead of top down.


Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[drawoh]

umatrix,

I have never done FMEA.

I have worked with people who were rigidly process driven. A process like FMEA is presented as a set of procedures. You start from step one. You do steps two and three and so on until you reach the fourteenth and final step. You are now done, and you can stop working, and file the report somewhere.

It would be a shame if thinking about the data, studying the results, and making decisions were not written up as part of the procedure!

--
JHG

 

[CWB1]

IME DFMEAs are standard engineering process. Identify and rank failure modes then analyze or test accordingly to mitigate those failures. It takes the ambiguity out of all downstream analysis and testing, and clearly documents WHY or why you are NOT making critical design decisions or expensive bits of analysis or testing. Given our legal climate stateside, I would NOT want to release any design not covered 100% by a DFMEA. I wouldn't say completing them is thrilling but team reviews are usually educational when you have good dialogue amongst reviewers and worth mentioning as a training opportunity for junior engineers. The biggest issue I see with DFMEAs is that some companies don't store and iterate them, so engineers are left to waste hours creating from scratch rather than simply tweaking the 30 year old DFMEA that's 95% appropriate.

As to PFMEAs, I usually ask suppliers for them but leave the necessity up to others. At previous mega-corps every supplier completed PFMEAs regardless of part volume. In the small consultancy world now the customer often doesn't care and suppliers try to minimize costs for small customers by skipping the documentation. Ultimately that's their decision for me to document, not mine to make. The DFMEA covers me, the PFMEA covers the customer and supplier.