Hi everybody!

Just had A PSO from one of my customers to release a new product, during PFMEA he asked for corrective actions for activities with severity 9 or higher, Is this right? I mean do I have to do this?

I looked all over FMEA manual and gives no information if i should or not, hope someone can help me in this matter.

Thanks in advance! :bigwave:

Every good PMEA has a recommended action for all potential failure modes. A nine is a nine, so fix it in the design.

Am I right in thinking that you cannot change the severity rating ?

Only reduce the occurrence, or improve the detection ??

Am I right in thinking that you cannot change the severity rating ?

Only reduce the occurrence, or improve the detection ??

You are correct.

Like Ben implies, severity rankings can only be changed with a design change.

Tattva, If your customer requires it in their CSRs or contract, you have to have to do it. Unfortunately I've seen far too many suppliers "fudging" numbers to avoid customer requirements like that. Realistically, you should consider improvement to failure modes that have a severity that high, but blindly requiring it doesn't seem right. If your severity is 9 with occurrence 1 and detection 1, is it really necessary? Can you give us some idea of the RPN or occurrence/detection rankings for these?

My answer is "It depends". I'm not sure how your customer defines a "9" but the AIAG guidelines say it is a Safety item with warning. If you have a Detection level of 1 (Poke-Yoke), so you are guaranteed to catch the failure and not ship it to your customer, and your management can "live with" whatever the Occurence value reflects, your customer should be fine. There is nothing wrong with a note in the Recommended Actions that simply states "No action to be taken at this time" to show that it was reviewed.

Sam is correct - you cannot reduce Severity on a PFMEA.

Every good PMEA has a recommended action for all potential failure modes.
I don't buy that!!!

Bill

Actually RPN is 18 Occ=1 and Det=2, the problem is that this is for physical test activities and he argued this was a very important issue

There are times when the severity is high but the likely of occurence is sooooo rare, or ease of detection soooo obvious that nothing further is to be done.

There are no rules attached to specific RPN numbers, nor detection or occurence numbers, that require action. This is a thought process, where each issue is considered on its own merit. The numbers just serve as guidelines.

I've worked with companies that "pareto-ize" the RPNs, and simply work on the top 10 or 15. Others say anything above 6 for each category requires action. I've tried to help all of them see that the value of the FMEA is in the exercise, not the precise numbers derived. :ko:

There is a second aspect of the 9 rating and perhaps this is how your customer looks at it


This ranking results when a potential Failure Mode results in a final customer and/or a manufacturing/assembly plant defect. The final customer should always be considered first. If both occur, use the higher of the two severities.

(Manufacturing/ Assembly effect)
Or may endanger operator (machine or assembly) with warning. 9

From the Ford FMEA training manual


Path/Step 1 of the Working Model is completed by considering

appropriate Recommended Actions to:

a. Eliminate the Failure Mode

b. Mitigate the effect

To reduce Severity, consider this action:
Change the design (e.g., geometry, material) if related to a product characteristic or change the process if operator safety is involved or if it relates to a process characteristic.
To accomplish this:
· Eliminate Failure Mode(s) or change its effect on the product performance.

If you believe that it is a design issue then you should return this to your customer for his action.
If it is your process then you should deal with it.

A nine is a nine, so fix it in the design.
Yup!

Actually RPN is 18 Occ=1 and Det=2, the problem is that this is for physical test activities and he argued this was a very important issue

It looks like you've got a pretty decent handle on the "9". With an Occurence of "1", you can't reduce the RPN without some type of 100% testing (as in the electronics field). But not knowing what you're producing, what the "physical testing" involves (cost, time, etc.), or what the actual safety issue is, I can't offer much else.

Rob Nice response. I've been on too many teams where the bickering about numbers is ridiculous. I've also been on teams that "understate" RPNs to reduce workload (or to satisfy a customer "bogey") and have it come back to bite them/us. It's taken a few years but we now have a core group that understands the potential of the FMEA methodology when we "hunker down" and do as thorough a job as possible at the outset.

Bill

Tattva,

It does not jump out at me in the above posts...... Who is design responsible?

IMO, since this is regarding Process FMEA, a severity of 9 may implies the process design is possibly causing a major / serious hazard.

Thus, reducing a severity ranking is possible here. Lets look at this way. The customer is responsible for the product design. But as a manufacturer / supplier, you are responsible for your own process design. If the design of your process may have high potential of hazards (hence the severity =9), you may want to look at ways of redesigning your process / machines..

Just my opinions...

I mostly disagree with that. The only case I would agree with it is if the 9 is due to a safety issue with production personnel, not with the customer/end user.

If a failure mode is possible with the customer's part design, nothing you can do in the production process can reduce the severity ranking. All you can do is reduce the probability of occurrence or increase the probability of detection.

I agree with Howste on this one. I can see where YKT is coming from but regardless what changes are made to your process, the severity of any given failure won't change. If the failure mode (let's say unthreaded bolt) results in a wheel falling off, as long as the failure mode exists there will always be the same end resulting possibility - the wheel might fall off - therefore the severity is always the same. If I redesigned my process so that unthreaded bolts could only happen once in a billion times and when it did we could find it automatically, the severity, if it ever did happen, never changed. The only way to reduce the severity would be to redesign the wheel so it couldn't fall off. Now the severity of an unthreaded bolt might be reduced to a lower number because nobody will die as a result of the failure.
One other point to look at is if I spent a billion dollars and eliminated the possibility of an unthreaded bolt (I mean it would never be possible), I wouldn't list it as a failure mode (because it could never happen) BUT, I still haven't reduced the severity.
JMHO

Dave

Comments above are on track.

One other aspect to consider about the severity, assuming that it's
a product design issue; that's why designers put in redundancy.

With the unthreaded bolt for the wheel example, notice that your
automobile has 5 or 6 bolts connecting the wheel. You can
lose 1 (or 2? or 3? don't know how much safety margin!) without serious risk.

If you look at true "safety critical'" features of automobiles you'll note
that they generally have backups.

So if design is the issue, consider whether there are backup systems
on the product; maybe the initial severity assessment was in error.
(Errors can be corrected, with explanation. As noted above, you can't
just reduce a severity without a change, so you'd have to justify
a reduction.)

Howard's earlier comment regarding hazard to the operator is a totally
different issue; it, too, might require a design change to fix. This time
it could also be a change to the customer's assembly process or the product.


Brad

If you look at true "safety critical'" features of automobiles you'll note that they generally have backups.

So if design is the issue, consider whether there are backup systems
on the product; maybe the initial severity assessment was in error.
(Errors can be corrected, with explanation. As noted above, you can't
just reduce a severity without a change, so you'd have to justify
a reduction.)Brad

I guess I don't follow where the "backyp system" affects the Severity rating. Doesn't it really affect the Occurence?? As Dave said - there is still the Potential for the Failure Mode to happen, even with a backup system, so how has the Severity been impacted?

Just looking for clarification.
Bill

Bill - responding to your question ...

Many of the PFMEA's I've seen (including some that I've done!) reflect
a lack of deep knowledge of the system that the product is installed in.

What I am suggesting is that if you *think* that you have a high severity
for a component that you make, it becomes your duty to study the
full system and learn about the design safety margins. If the severity # is
real then you need to be thinking about 100% automated on line testing
or some other extremely high confidence control system.

I recognize that this sounds good but may be very difficult - I have rarely
seen a design FMEA from *any* of my customers' customers -- North
American auto companies. I won't
name names but one has a 2 letter name and another a 4 letter name,
and neither are owned by Germans. I am often told that these documents
exist but are company confidential. I wonder sometimes....

It's also an interesting point on how this might really be considered as
an occurrence factor. Maybe the horrible consequence only happens if both
the primary and the backup system fail simultaneously. That could be
handled by decreasing the occurrence index.

But to restate what everyone else has said, the only way to change the
severity figure involves either a change to the design, or, as I suggest,
a change to your knowledge of the design.

Regards,
Brad