The FMEA Mini-Series: Using an FMEA vs. SxO for Prioritizing

[J.R. Strickland]

Al,

You are correct. I always make sure the team agrees and that the criteria is DOCUMENTED. I also try to make sure they stay within the 10 point scale and cover 1-10.

 

[Help Me]

Thanks,

All along, through my posts I have been saying that a high severity ranking (which, by definition, is the ranking of the severity of the effect of the potential failure-I think I got ripped once for looking at it in this way)is an indication that specific attention should be given to reducing the occurrence and detection rankings, therefore the overall RPN.

I think I referred to this as "Low hanging fruit." And, I would suggest that all the horns and secondary chutes actually, on a system level, do just that. They reduce the occurrence of the mechanism of the potential failure, and provide a means of detection of a failure. So, the redundant systems are a fantastic way to lower an RPN with design changes. But, if I remain cognizant that these items only will come into use if the main system fails, I must also be prepared to address what happens if the back-up systems (ABS, stability control, horns, back-up chutes, and the like) fail. And in the vast majority of cases, the severity of effect of the failure (again, this is the definition of DFMEA severity)of the parent system is inherited by the child, or back-up system.

If I don't, at least, consider what could POTENTIALLY happen if all the wheels fall off,(btw, does anybody remember when I said this was hypothetical? Sometimes exaggeration helps make the point. eg. I'll bet the guy who designed the hydraulics on that plane never dreamed that the main and two back-up systems could be wiped out in one event, either. But, it happened.) I haven't really assessed the seriousness of the effect of all the failure modes. Again, I admit that the wheels falling off was an unlikely HYPOTHETICAL (I am sure I said that before. Didn't I???)

As for System FMEA/ Design FMEA goes, I can't help but look at designs in terms of systems. And the AIAG FMEA reference manual specifically states that, "End items, along with every related system, subassembly and component should be evaluated." So, I don't think I am off base by taking a systems approach.

Perhaps we must agree to disagree on whether or not a severity ranking for the effect of a potential failure (the definition of DFMEA Severity)can be effectively reduced by a design change and get on with our lives.

I will stick to focusing on designs that will favorable impact occerrence and detection, esecially on the high severity items.

I thank all of you for your patience with me on this topic.

Good-bye

 

[Al Dyer]

Help Me,

Don't say good-bye. Join the forum and post your opinions, we all need a wake-up call now and again.

ASD...

 

[Sam]

J.R. Strickland posted:"
In the parachute example, if the chute doesn't open, you probably die and therefore it is a 10 (failure occurs without warning)."

TRUE
"No suppose I design a smartchute that has built in diagnostics that emits a loud audible alarm telling me it is not going to open...It still doesn't open, but warns me that I am about to die, therefore making it a 9 (failure occurs with warning.)"

TRUE but the potential failure effect of the of the alarm becomes 10.

"If I take it a step further, and add a smaller backup chute that deploys, that allows me to land without dying, I can make a case for the severity being a 7 (Item operable, but at a reduced level of performance. Customer dissatisfied)."

TRUE, but the potential failure effect of the backup chute becomes 10; and so on through out the design phase.
The original potential failure effect was "the chute does not open" The ranking will never change but will only be carried over to each design change. IMHO

"The question of action should be based on the RPN, not severity only. I think your ISO organization is wrong if they are trying to force you to lower the severity just because it is a 9 or 10. We often have 9's and 10's on our DFMEA's because many of our products effect safe vehicle operation or a government regulation."

Refer to the AIAG FMEA manual pg. 21step 18 the last underlined sentence in the paragraph.

 

[Help Me]

Sam,

Thank you!

I knew somebody out there had to understand what I was trying to say.

I agree wholeheartedly with the line you refer to from the manual.

And I think that comes through from my posts.


Tom

 

[Al Dyer]

Originally ...

Refer to the AIAG FMEA manual pg. 21step 18 the last underlined sentence in the paragraph..........[/B]

I gracefully ask, what is your point?

The FMEA manual says we can define our own criteria and what do you consider high severity?

I think it is company specific and based upon customer requirements!

Does underlining a sentence that starts with "In General......." make a difference?

I will not go into the other subjects such as "shall", "should", "note", "as applicable" etc..............

ASD...

[This message has been edited by Al Dyer (edited 20 April 2001).]

 

[Marc]

Now you're going beyond common sense. Your original line item was 'the wheel falls off'. It was not 'every wheel falls off at the same time'.

If you will stick to the original line item, the design change will have reduced the severity of that failure mode or, as often happens, eliminated the specific potential failure effect. Your arguement is based upon trying to keep the potential effect of the potential failure.

You keep saying you still have to address the effect of the potential failure.
->But, you still have to look at the POTENTIAL EFFECT of the
->redundant wheels falling off.

Now you're talking a higher level. You're changing the subject.

One effect is for a given wheel to fall off. You want to now say, well, all the wheels might fall off so you have to address that. Well, maybe, maybe not. But let's say you you do want to address all wheels falling off at the same time. That is more of a system FMEA than a design or process FMEA. None the less, we'll say it's the line item you're concerned with.

Potential Failure Mode = All wheels fall off at the same time.
Potential Failure Effects = Car uncontrollable, possible death.

If I redesign the car to ride on an air cushion, the severity of all wheels falling off approaches zero.

->At best, you can find a wall to throw it over so that the
->engineer of another system has to cintend with a high
->severity.

Yes - sometimes there are tradeoffs. Sometimes not.

->Replace the wheels with a cushion of air. Fine, the
->severity fo losing an ornamental wheel is relatively
->minor. But, you have also changed the "Item/Function".

So? Is that somehow illegal? Please explain why that does that not count for reducing the severity of the effect of a wheel falling off. So what if you assign it to another system which can better handle the problem.

-> Whether they fall off
->simultaneously, or one at a time without any detectable
->warning to the operator.
This is not a valid arguement because that's not how FMEAs evolve.

I have never, by the way, seen a top level automotive FMEA (or lower level FMEA, whether design, system, or process) which has as a Potential Failure Mode of "Every wheel falls off at the same time". It doesn't happen in real life. One wheel - yes. Maybe even 2. But 3? Or 4?

Every month you hear of another stability function being added to cars. All wheel drive, ABS, electronic stability control systems - all of these reduce the severity of a wheel falling off. And they are all design changes.

 

[J.R. Strickland]

I hesitated to dive into this one, but what the heck...

The severity applies to the EFFECT only. The effect of a given failure will not change unless you change the design of the system or part. In the parachute example, if the chute doesn't open, you probably die and therefore it is a 10 (failure occurs without warning). No suppose I design a smartchute that has built in diagnostics that emits a loud audible alarm telling me it is not going to open...It still doesn't open, but warns me that I am about to die, therefore making it a 9 (failure occurs with warning.) If I take it a step further, and add a smaller backup chute that deploys, that allows me to land without dying, I can make a case for the severity being a 7 (Item operable, but at a reduced level of performance. Customer dissatisfied).

The question of action should be based on the RPN, not severity only. I think your ISO organization is wrong if they are trying to force you to lower the severity just because it is a 9 or 10. We often have 9's and 10's on our DFMEA's because many of our products effect safe vehicle operation or a government regulation. If the severity is high, we at least think about any changes that might be made. Often times, we have no control on what the vehicle does when our parts fail...this is determined by the car companies and we all know they are infinitely wise in areas of quality and safety. If changes are not feasible, we then focus on occurrence and detection (See my post much, much earlier in this string on what our procedure is) to bring the RPN into an acceptable level.

 

[Al Dyer]

Marc,

You posted:

...Every month you hear of another stability function being added to cars. All wheel drive, ABS, electronic stability control systems - all of these reduce the severity of a wheel falling off. And they are all design changes.

Thank you, severity can be reduced!!!!!!!

ASD...

 

[J.R. Strickland]

Sam posted: "Refer to the AIAG FMEA manual pg. 21step 18 the last underlined sentence in the paragraph."

Yes, the OEM's get a little antsy when 9's and 10's are in FMEA's despite the fact many of them are driven by the system they own. (I wonder what the Firestone FMEA had in it?) If you look closely at my original post, I did say..."If the severity is high, we at least think about any changes that might be made." The actual verbiage in our internal procedure is "...Action plans will be considered and initiated, where possible, for the following FMEA conditions: 1) the top (highest) 3 RPNs, 2) a severity number greater than 8...). Why does it say "...where possible..."? I still stand on my claim that real action is based on RPN. What is a design team to do if they have a S=9,O=1,D=1 making the resultant RPN=9? My answer?...nothing, unless there is some easy and inexpensive change that can be made. My company's engineering resources will be better spent working on the other issues or designing the next product. In general I agree with the underlined statement on page 21, section 18. This doesn't mean I force my design organization into the redesign loop just because I have a 9 or 10.

As for the analysis of my scenarios, I agree that a failure of the main chute AND the alarm constitute a 10. I suppose one could make some argument as to the failure of the alarm itself being a 10, but I would put it further down on the scale. Most FMEA's don't take into account 2nd, 3rd, and 4th level failures in determining severities. While that would be nice, for complex products, it becomes an impractical exercise and none of us would want to pay the cost in sticker price for each part of a vehicle to take the FMEA to that level.

In the end, if I have a 10 and my design is so robust that my occurrence is a 1 and my design controls are so thorough as to be a 1, I move on and really worry about those items that have S=7,O=5,D=6,RPN=210 or 9 and 10 severities that do not have low occurence and detection ratings.

Help Me may be right about agreeing to disagree. However, I also read a lot of aggressive agreement in the posts.