Requiring action on factors alone of RPN in FMEA



I understand requiring action on a high Severity requirement, has anyone run into resistance based on a high Occurrence or Detection? As in a blanket "If Severity, Occurrence, or Detection is greater than _ , corrective action is required"?

I Keep thinking of the hit by meteor failure mode. S 10, O 1, D 1. We're not going to make this part meteor proof.

Bev D

Heretical Statistician
Super Moderator
in my organization we take action based on severity only.
However, we don't allow meteors, snow-apolcalypses, tornados, revolutions or zombies as 'failure' modes. I tell the engineers that if we - or our Custoemrs or suppliers get hit by a meteor we all have other things to worry about besides how our product wil function.

We do use these as part of our business continuity process. When I worked at Honda we had robust busines continuity processes regarding supplier events - and I saw it put successfully into play when one of our single source suppliers had a catastrophic fire. My current business offers generous help to Vet practices in teh event of hurricanes, floods, etc. and we have processes for weather events that disrupt deliveries...and this is the approapriate place for such events.

The issue with blind application of required action for high occurrence or detection is that it ignores severity. a 1 or 2 severity which may not be noticed and typically have no tangible effect on the Customer hardly 'deserves' a fool proof detection method. This misguided approach is even more wasteful and diversionary if the occurence rate of a failure mode which has a trivial effect will rarely occur. Logic must be applied at some point.

the real issue with blind application of RPN calculations is that it is trying to replace human logic with a simple (and flawed) mathematical calculation. When did it become popular to promote lack of critical thinking?

Englishman Abroad

Involved In Discussions
I agree with Bev,

FMEA is a tool used by people, to think about the product / process risks and to analyse them.

If you look in the FMEA manual p103, it is explained that a RPN threshold rule "
if RPN > X then action needed" is not advised.

Because, FMEA is not a scientific process, different teams will rate something higher or lower than others.

Also if you have a threshold imposed then most probably everything will be rated just under the threshold, and the team will not improve the product design or process.

The FMEA should be used to rank/prioritise. Different FMEA teams will have similar failure mode priority lists, even if the absolute values of RPN, or SxO values are different.

In my company we rank actions firstly based on severity, then SxO, then RPN. Action stops when the residual risk is at an acceptable level (it is not worth the resource needed to resolve it).


Involved In Discussions
If the Severity is high ( 10) , but Occurrence is 1 and Detection is 1, action is not needed any more, because The production capability is so high that no problem will hapen (O=1 or 2 or even 3), the measuring reliability is so high (D=1, 2....that all the problem can be detected and no problem will escape.
On the other hand, if the Serverity is very low (S=1 for example), that means the problem is so small that it is not a problem at all, why we need action =money, time !

Bev D

Heretical Statistician
Super Moderator
If the Severity is high ( 10) , but Occurrence is 1 and Detection is 1, action is not needed any more, because The production capability is so high that no problem will hapen (O=1 or 2 or even 3),
One caution: if your occurrence is based on capability then you must recognize that assignable causes can and will happen that you may not be able to detect or control in which case you have created a high severity defect. Also how is the capability assessed? With a weak Cpk study over a short period of time or is only opinion based? Or was it from a long term stable performance data?

IF you have the highest level detection that is actually mistake proofed you 'may' have a feeling of comfort. but I always question these as too often the rating is opinion and not fact based. was the detection system actually tested to ensure that it was in fact impossible to miss even a low defect rate?


Thanks all, just have a customer asking for a corrective action on any of the factors of the RPN being 7/10 or greater. Seems like a waste of resources on strictly high Detection and Occurrence scores. Which is the whole reason we examine RPN in the first place. I understand the compounding effect of S X O and would understand this policy, but I would imagine a lot of "6" scores when I tell the cross-functional team about our imposed requirements. Which skews the FMEA. Argh.


That type of customer youre talking about has not understand the tool FMEA but is destroying the method.

We once had a similar customer with a nice looking RPN rule.

Everything it did was getting all suppliers to rate exactly so they just produced the paper without any value adding content in it...


I look at FMEAs more generally. They are an improvement tool. Of course high severity items need to be looked at, but if you have a severity of 1 and an occurrence of 10, you might still want to perform action as a continual improvement. Don't limit yourself to items you are "required" to take action on. Take action on items that might not be required, but make sense.
Top Bottom