Severity Reduction in a Design FMEA (DFMEA)

Matt_B said:

I am having a disagreement with the person who is training me to teach FMEA classes. I wrote a blog post in which I used an example FMEA that had a severity of 10. The actions taken involved a design change that eliminated the potential failure effect so I made a new estimate for the severity.

My trainer told me that I was wrong to do this. He said that in the revised metrics section of the FMEA I should have kept the severity at 10 and re-estimated the occurrence and detection based upon the improvements. This particular failure can?t occur due to the listed failure mode so I believe a new severity is appropriate.

We are in agreement that the new design should also be evaluated as a new point in the FMEA; however, I am convinced that the severity can be reduced with a proper design change. The AIAG manual concurs; however, my trainer believes that this new severity is used in a new line and not as part of the revised metrics.

I disagree with my trainer, but have agreed to teach things his way until such time as I can provide concrete evidence that I am correct.
Can anybody please provide that evidence? Or, explain to me why I am wrong.

Thanks!

Click to expand...

Yes, you can change the severity after a design change in the DFMEA.

Here is a description from an internal procedure that I use:

The severity ranking shall reflect the severity of the effect on the customer. Effects assume that the failure mode has occurred and that no controls are in place to detect and contain the effects of the failure mode. Assume that the effect has escaped all design controls and rate the impact on the customer. For in-plant effects assume that no controls exist and rate the impact on production. Severity rankings may only be reduced by making design changes to the product.

Severity rating belongs to the effect, so this typically does not change whatever you do (except when there is new information to impact the rating). For a given design therefore the failure mode you are assessing does also not change severity rating whatever action taken; with a fixed design your actions can only affect detection or occurrence. Changing the design may impact the failure modes and their relationship to the effects, e.g. a certain failure mode cannot occur anymore and others appear that you did not have before (-> delete/add new lines to the FMEA). With a design change also an already existing failure mode may now lead to a different effect (or a similar one with higher/lower severity) therefore getting a new severity rating.

1. The trainer is wrong. You are right.

2. During DFMEA it is preferred to reduce high severity ranking by design change. The design change you described looks like severity reduction (reduction of negative effects when failure mode occurrs).

3. AIAG FMEA Manual rev.4 page 67 says:
"To Reduce Severity (S) Ranking: Only a design revision can
bring about a reduction in the severity ranking."

Hope it helps

I showed him the info from page 67 last week; but it does not specify if the new severity is listed as the new severity on the right side of the FMEA or only in a new entry in the FMEA. I say the area for the new severity is where the reduced severity is listed. I just can't prove it.

Matt_B said:

I showed him the info from page 67 last week; but it does not specify if the new severity is listed as the new severity on the right side of the FMEA or only in a new entry in the FMEA. I say the area for the new severity is where the reduced severity is listed. I just can?t prove it.

Click to expand...

This depends on what the form looks like in the beginning. If there is already an entry on the left side with the SOD numbers entered and the item is revisited, it would be appropriate to use the right side of the form to document the new information. Conversely, if it's a new entry, you would obviously start on the left side.

It pays to remember that the form is not the FMEA. FMEA is a process and the form is a record of it. In that sense it might not make much difference where the information is, so long as the information is there.

Matt_B said:

I showed him the info from page 67 last week; but it does not specify if the new severity is listed as the new severity on the right side of the FMEA or only in a new entry in the FMEA. I say the area for the new severity is where the reduced severity is listed. I just can?t prove it.

Click to expand...

My understanding is following:

AIAG manual explains the FMEA form "column by column".

Page 67 of AIAG manual says about "Recommended Action(s)", next pages also are related to Occ and Det reduction. So it is related to the right side of the form. I think that intention of the manual creators was to put "reduced severity rank" in right side of the form. This approach clearly shows identified high severity rank (left side of the form) and actions taken to reduce it along with decreased severity rank (right side of the form). Same approach is used for Occ and Det ranking changes due to actions taken.

Page 71 says:
"After the preventive/corrective action has been completed,
determine and record the resulting severity, occurrence, and
detection rankings"
It means that severity may be a result of completed actions in same manner as occurrence or detection.


I recommend to put into "actions taken" short info about design change taken and what effect are changed (reasons for severity reduction in next column).

In my opinion: Starting a new entry is reasonable only if there is another failure mode identified (for given function/requirement) or a new cause is identified (for given failure mode). If there is just another effect identified (along with its severity rank) then it should be added into "effects column" for given failure mode and severity ranking updated (if new effect severity is higher than others).

In general I agree with Jim - FMEA is not just a form..

There are no "right" and "wrong".

It all depends on definitions, e.g. to what your severity ranking relates. If it is attributed to the hazardous situation (i.e. how severe can the outcome be) then it can be reduced. If, however, it is attributed to the outcome itself (as it is, for instance, in ISO 14971 - medical devices risk management), it can't be reduced. If, for example, patient death is a conceivable outcome, its severity will always be the same - top. The only possible change in that context is the total elimination of such outcome from the scope (it can be represented by striking through the entry), e.g. when the mitigation is so effective that patient death is no longer conceivable. In all other cases, a patient death is a patient death, and mitigation means can only reduce the probability of occurrence.

Ronen E said:

There are no "right" and "wrong".

It all depends on definitions, e.g. to what your severity ranking relates. If it is attributed to the hazardous situation (i.e. how severe can the outcome be) then it can be reduced. If, however, it is attributed to the outcome itself (as it is, for instance, in ISO 14971 - medical devices risk management), it can't be reduced. If, for example, patient death is a conceivable outcome, its severity will always be the same - top. The only possible change in that context is the total elimination of such outcome from the scope (it can be represented by striking through the entry), e.g. when the mitigation is so effective that patient death is no longer conceivable. In all other cases, a patient death is a patient death, and mitigation means can only reduce the probability of occurrence.

Click to expand...

Totally agree. Thing is that some people really get creative when it comes to defining the worst case scenario. There is always somebody who can think of a scenario that leads to the death of a patient. And then people assess the risk by combining the severity of the (improbable) worst case effect with the probability of occurrence of the failure mode instead of the probability of the worst case scenario...

Matt_B said:

I am having a disagreement with the person who is training me to teach FMEA classes. I wrote a blog post in which I used an example FMEA that had a severity of 10. The actions taken involved a design change that eliminated the potential failure effect so I made a new estimate for the severity.

My trainer told me that I was wrong to do this. He said that in the revised metrics section of the FMEA I should have kept the severity at 10 and re-estimated the occurrence and detection based upon the improvements. This particular failure can?t occur due to the listed failure mode so I believe a new severity is appropriate.

We are in agreement that the new design should also be evaluated as a new point in the FMEA; however, I am convinced that the severity can be reduced with a proper design change. The AIAG manual concurs; however, my trainer believes that this new severity is used in a new line and not as part of the revised metrics.

I disagree with my trainer, but have agreed to teach things his way until such time as I can provide concrete evidence that I am correct.
Can anybody please provide that evidence? Or, explain to me why I am wrong.

Thanks!

Click to expand...

Assuming that the Design change eliminates the most serious effect and not the failure mode or cause, where is the question of re-estimating the Occurrence? In your blog case, irrespective of any number of explosions in exhaust port it is assumed the reactor is unharmed due to isolation. Re-estimating the number of explosions in exhaust port or number of attack doesn't make sense to me. Perhaps your trainer doesn't agree with you on the assumption.

Your trainer is 100% correct and the AIAG manual does not agree with you.
The design change MUST eliminate the failure mode. A design change which eliminates the cause is great, but it does not eliminate the failure mode. Only a design change which changes the requirements and eliminates the failure mode can do that.
When this occurs the DFMEA line with the Recommended Action improvement is archived. It no longer represents the product design as the design requirements have changed. Current, archived, and proposed rows/ data should all be kept in the FMEA for Lessons Learned purposes. The FMEA is then filtered to "current" when publishing the FMEA for common use.