27 Feb 2015 Why we should use Mini-FMEAs.

A wonderful blog at Literally Engineering discusses a course on FMEAs - Failure Mode and Effects Analysis. It points out that even the gurus who make a living training people on FMEA have a hard time pointing out their true value.

My own experience is that generally FMEAs have a negative value. Let me explain. The idea behind them is sound. Get a team together to systematically explore what could go wrong with an engineering structure, system, process then put in place whatever is needed to stop those things from going wrong. It's a great idea. Engineers tend to focus on the hard task of getting a system to do something well, on time, on budget. Focussing on how it might go wrong doesn't come naturally (save for some obvious things like breaking loads in bridges).

So you get a team of insiders and outsiders to go carefully through every aspect of the process, logging how things might go wrong, scoring the severity and, ultimately, taking the highest scoring items and working out how to reduce the risks either by redesign (best) or by control monitoring (less good, because then you have to think what happens if the control monitor goes wrong...).

Gurus abound who can offer you systems, spreadsheets, software to do this. But it's so trivially easy that anyone with a spreadsheet can do it. Indeed, I've done it myself. The problem is not the system or software. The problem is that the whole thing is grindingly tedious. And this is where the negatives immediately set in. Anyone who has flair, imagination, dynamism and go-getting tendencies (i.e. the sort of key people in an organisation) will crawl on broken glass rather than participate in an FMEA. You lose the will to live after the first 15min and there are days and days of FMEA stretching into the future. A certain way to demotivate any bright employee is to tell them that they must participate in an FMEA.

Why has such a basically sound idea become so deadly? Because there is no limit to what an FMEA might naturally do, and because they become part of the official system which needs to be "signed off" as meeting FMEA norms. The people who set FMEA norms are generally the sorts of administrators who set all sorts of other norms that generally clog up most organisations. And "just to be safe" you can always find another layer of FMEA detail that must be done so that boxes can be ticked and lawyers fended off.

The result of this (perhaps) well-intentioned over-doing of FMEA is that less of them get done than would be optimal. The FMEA concept is really good and we should all use it as a routine part of our product development lives. But no one with any sense will allow themselves to get sucked in to the type of FMEA monster that has grown up over the decades. So FMEAs tend either to be bloated, will-to-live-sapping monsters or to not happen at all.

That's why I once decided to invent the concept of the mini-FMEA. The rule was that on all significant projects we'd do a mini-FMEA, that the brightest people in the team were especially required to participate and, therefore, no meeting would ever last more than 1hr. In a normal FMEA, 1hr is just sufficient to decide whether Factor 1 should really be Factor 1A and 1B. In a mini-FMEA you've got an outline FMEA of the whole thing and probably spotted a couple of important issues that would have been missed. Just a couple more 1hr sessions and the job's done.

The point is that a mini-FMEA gets 90% of the benefit for 10% of the work. A real FMEA in principle gets 99% of the benefit for 100% of the work, but in fact, because people do everything to avoid them they get 50% of the benefit for 100% of the work. They are positively counter-productive simply because they are so awful.

Another rule I adopted for mini-FMEAs was that I would chair those that concerned me. Those who know me realise that I am exactly the wrong sort of guy to head up an FMEA. And that is why it was important that I headed my mini-FMEAs. Because I was intellectually convinced of their merits (and events proved me right) I had to show the leadership to make sure they got done. The very fact that a non-natural FMEA type like me was leading the process made the point that these were important.

There was some opposition to these mini-FMEAs. They were undermining the culture of the full FMEAs. But that undermining was nothing to do with me. The poor folk who'd been in the full FMEAs knew that they would never again volunteer or be volunteered to participate in one. And some of them were happy to take part in the mini-FMEAs and their learning from the full FMEA was hugely useful. There was a danger that I was being too mini (arguably micro-FMEAs are too trivial to be useful) and I learned a lot from them.

So I'm not calling for the world's FMEA gurus to be out of a job. All I'm asking is that they start promoting the skills and attitudes needed to foster lots of mini-FMEAs. The world will be a better place if they do.

If you are a lawyer reading this, I have to admit that if I were involved in, say, an oil rig for drilling deep into the Gulf of Mexico then a full FMEA might be more appropriate. This blog is about the vast majority of engineering activities where full FMEAs are positively destructive of engineering excellence.