Sorry for this topic again but for our company this is a something that has to be cleared.

This is a quotation from SAE-J-1739 and AIAG:
Page 13 form SAE, DFMEA Occurrence evaluation criteria:
"Likely failure rates over the design life:
... per thousand vehicles/ items..."

"Occurrence is the likelihood that a specific cause (I understand "root cause")/ mechanism will occur during the design life"

Why they use counting of the failed vehicles?!! Isn't it just the effect of the failure?

Can you explain what do they mean by "vehicles" and "design life"?

How to assess the occurrence for a new device for example - based on problems with similar devices or in some other way?

Sorry to bother you again.
Thank you very much for your help!

This is a quotation from SAE-J-1739 and AIAG:
Page 13 form SAE, DFMEA Occurrence evaluation criteria:
"Likely failure rates over the design life:
... per thousand vehicles/ items..."

"Occurrence is the likelihood that a specific cause (I understand "root cause")/ mechanism will occur during the design life"

Why they use counting of the failed vehicles?!! Isn't it just the effect of the failure?

We can't assume that a vehicle will "fail," nor can we assume that the potential effect will happen every time the potential cause does. Calculation of the likelihood of interplay between cause and effect is a different matter. To put it another way, there are two separate (but related) questions:

What is the likelihood that some potential cause x will happen?
If cause x happens, what is the likelihood that it will result in an undesireable effect?Can you explain what do they mean by "vehicles" and "design life"?

I think the explanation above should clear up the "vehicle" question. As to "design life," it's the normal life expectancy of the product, in each of its possible iterations (a part or assembly might be used in more than one application). I have one of those old Ecko manual can openers at home; the same basic design has been in use for longer than I've been alive (~150 years, if you're only as old as you feel :D ). I'm sure there have been electric can opener designs that are long since obsolete, however. The question to be answered is, how long will a basic design be used to manufacture product?

How to assess the occurrence for a new device for example - based on problems with similar devices or in some other way?

You have to use the best data available--experience with similar products, or an estimation based on the expected life spans of components.

We can't assume that a vehicle will "fail," nor can we assume that the potential effect will happen every time the potential cause does. Calculation of the likelihood of interplay between cause and effect is a different matter. To put it another way, there are two separate (but related) questions:

What is the likelihood that some potential cause x will happen?
If cause x happens, what is the likelihood that it will result in an undesireable effect?

I think the explanation above should clear up the "vehicle" question. As to "design life," it's the normal life expectancy of the product, in each of its possible iterations (a part or assembly might be used in more than one application). I have one of those old Ecko manual can openers at home; the same basic design has been in use for longer than I've been alive (~150 years, if you're only as old as you feel :D ). I'm sure there have been electric can opener designs that are long since obsolete, however. The question to be answered is, how long will a basic design be used to manufacture product?



You have to use the best data available--experience with similar products, or an estimation based on the expected life spans of components.

Can you give me understandable example of failure mode and occurrence ranking based on some number of vehicles/items failed in the field?
:rolleyes:

Can you give me understandable example of failure mode and occurrence ranking based on some number of vehicles/items failed in the field?
:rolleyes:

Perhaps you're still struggling with the "vehicles" concept as it relates to potential failures in the field. Just because a component fails in a vehicle, it doesn't mean that the vehicle has failed. For example, a radio might malfunction, which will have no affect on the operability of the vehicle. On the other hand, there are component failures that will result in catastrophic vehicle failure.

"Vehicles/items" is just an overly vague construction, and the AIAG documentation is full of those (http://elsmar.com/Forums/showthread.php?t=18626). Look at it this way: You have a component. You've identified potential failure modes. You've identified potential causes of those failure modes. What is the likelihood that (A) any given potential cause will happen in end use (i.e., in an operating vehicle), and (b) what is the likelihood that the cause, if it does happen, will result in failure of the component?

Our Quality manager ask me a questions. :)
How to interprete this in terms of the customer returns?
We have product (IC) with known wickness during the design.
Due to this wickness we receive customer returns (field returns). We have certain rejection rate (3ppm).
So our QM asks: How these 3ppm should be linked with the occurrence rating in the DFMEA?

Our Quality manager ask me a questions. :)
How to interprete this in terms of the customer returns?
We have product (IC) with known wickness during the design.
Due to this wickness we receive customer returns (field returns). We have certain rejection rate (3ppm).
So our QM asks: How these 3ppm should be linked with the occurrence rating in the DFMEA?

By number of items (as opposed to number of vehicles) perhaps? But whether it's "items" or "vehicles" would depend on how many of the failed items find their way into vehicles. In other words, you might have something that breaks down like this:


5% of the manufactured ICs will be likely to have the defect in question.
Of that number, 95% will be detected in internal (your company) testing.
The remaining 5% will be shipped, but 95% of those will be detected by the customer prior to his final assembly.
The remainder will find their way into vehicles, but might not cause failures of the assembly.It can get quite complex, as you can see. Maybe it's best to look at it this way: if you know that your present design will yield a certain level of part failures, your customer wants to know how that will affect him. Note also that the evaluation criteria (the vehicles/items bit) in the manual are "suggested" and intended as a "guideline." The manual explicitly states that
The team should agree on an evaluation criteria and ranking system, which is consistent, even if modified for individual product analysis
This means that you should use a ranking system that makes sense in your application, and not do this: :frust: trying to figure out how to make a square peg fit in a round hole.

Our Quality manager ask me a questions. :)
How to interprete this in terms of the customer returns?
We have product (IC) with known wickness during the design.
Due to this wickness we receive customer returns (field returns). We have certain rejection rate (3ppm).
So our QM asks: How these 3ppm should be linked with the occurrence rating in the DFMEA?

Didn't the AIAG FMEA manual, once upon a time, display a chart of Occurance ratings based on ppm or other numerical rates of incidence? Brutas, do you have access to this manual? I don't have a copy handy here, but my (bad) memory :notme: would have me guessing that a 3 ppm rate would equal a 1 or 2 for an Occurance rating. This would most likely render a rather low RPN #, which would leave you working on other issues for defect prevention, other than this particular example. Hope that helps clarify some of what Jim has been saying. :2cents:

Didn't the AIAG FMEA manual, once upon a time, display a chart of Occurance ratings based on ppm or other numerical rates of incidence?

It does, but for process FMEAs. For DFMEAs the criteria are based on number of failures per x "vehicles/items," which is why I suggested that some other method might be more appropriate.