Severity Reduction in a Design FMEA (DFMEA)


Trusted Information Resource
I think it's not the question of elimination of cause reducing severity. It is elimination of effect of failure that is in discussion. Doesn't it reduce severity?


No, it does not reduce severity, it only reduces occurrence (elimination of cause of failure). If the cause of failure is totally eliminated, then "the cause of failure is eliminated through preventive control" and occurrence = 1 if you are discussing an assembly containing a valve.

If a part requirement for a valve is to "flow xx cm3/sec", potential failure modes are "flows < xx cm3/sec" or "does not flow any volume". The effect of failure may be "torque gun cannot cannot meet torquing requirements." A potential cause for both failure modes is "the valve is assembled backwards". Designing in a method so that it can only be assembled one way does not change the fact that it will not function (supply torque gun torquing effort) if it is assembled backward, it means that the cause "assembled backwards" has been prevented (Preventive Control). There is no reduction of severity because the effect of failure has not changed. There may be other causes which will make the valve "flow < xx cm3/sec" or "does not flow any volume" and they will have the same effects of failure.

Do not mix DFMEA levels. Stay within the scope. the lower level DFMEA for the value component assembly itself, carries the Potential Failure mode "Assembles backward" which can have the failure mode eliminated by the design "one-way" change. The assembly of the valve in a higher level assembly cannot. The cause from the higher level DFMEA , "Assembles backward", becomes the failure mode at the valve component assembly level.

"Effects of Failure" only ranks the failure mode, it is not an actionable item.


Trusted Information Resource
"Effects of Failure" only ranks the failure mode, it is not an actionable item.

I don't see any reason why it is not. Electrical Fuse is the best example I can think of. In overload condition, it sacrifices not to prevent overload but to minimize the consequences of overload. If presence of such aspects in my design is mitigating the most serious effect scietifically and logically, why shall i measure risk in my system as much as the one without such safety aspects. While whether or not Severity rating can be altered is a debate one can keep talking about, many inventions in automobile such as airbag, seatbelt might not have come up if effect of failure were not actionable. These features do make a difference to safety of a vehicle, don't they.


When dFMEA is not addressing severity reduction, I guess then it is weak because further down, no other FMEA is so strong to mitigate severity. I also guess this is where the notion comes up that severity cannot be reduced.

Zbigniew Huber

"Effects of Failure" only ranks the failure mode, it is not an actionable item.

Effects of failure is related to Severity rank. Severity rank = most serious effect of failure mode. I hope we all agree to that :)

In DFMEA actions related to severity reduction (it means reduction of effects when failure mode occurred) See page 67 of AIAG Manual.
"To Reduce Severity (S) Ranking: Only a design revision can
bring about a reduction in the severity ranking."

Severity is reduced by design change when effects of failure are changed.

For example:

Failure mode = short-circuit
Effect(s) = no output,loss of primary function(8), unit on fire (10)
Severity = 10
Cause = power transistor breakdown...
Add fuse in power supply line, to eliminate fire risk.
Severity (revised) = 8

When fuse is added, in case of short-circuit (failure mode) the fuse will blow causing unit to turn off (loss of primary function severity = 8) but ELIMINATE the next worst effect - fire (severity = 10) .
The action reduces severity (the failure mode and the cause is still present).

There are many designs that reduces severity not the cause nor failure mode:
- air bags
- redundancy (secondary system takeover the control in case of primary system failure)
- fail-safe mechanisms ( i.e. fuses, shields etc.)

Elimination of failure mode by requirements change and redesign to eliminate the cause are correct methods, but keep in mind that original post of this discussion was about design change that reduces effect of failure (failure occurred, causes of failure not changed by design) and how to write it down into FMEA form.


I can clearly see that my example may not have been a good example; I should have continued it with the next levels to make it clearer.
Still, I don?t see how a severity can remain 10 if the consequences (total destruction) of the failure mode are no longer possible. The failure mode can occur all-day long, but would no longer result in a catastrophic loss.
I think my example would be like taking a valve that requires a flow of xx cmm3/sec and redesigning it so flow is no longer required. Sure, the flow may go low, but the redesigned value would no longer be affected.
Of course my trainer has made it clear that the redesign must also be evaluated and I agree with him completely on this point.
There were additional actions in my example that reduced the occurrence; however, they were not relevant to the severity reduction.


In my opinion the severity could only be change with a change in design. For example, falling hard brick on someone's head from a 3rd floor is a severity 10 because the person can die. If we ask all people to wear a hard hat, we are just reducing the occurrence so the severity stays. However, if we make a design change and change the brick material to foam, now the falling brick is no longer a hazard and the severity could be reduced.
Top Bottom