View Full Version : Automotive Severity 10 - Supplier Response - FMEA for an electronic product
Beldenray 26th July 2006, 11:08 AM All,
Interesting question.
Automotive customer has reviewed our FMEA for an electronic product, and determined that a severity of 10 exists for a specific failure mode. The customer's systems cannot detect or dianose this failure mode, leading to potential of a severity 10 for this failure mode.
In the past, I have been told that the Vehcile OEM's will not allow 10's to be passed off to their customers. As a supplier, this is causing us grave concern.
Should we force our customer's hand to resolve this? No one wants to expose consumers to risk.
Belden
howste 26th July 2006, 11:24 AM So this failure mode may affect safe vehicle operation without warning? If so, it is what it is. If it's a 10, there may be no way to reduce the severity except through design change. What you need to do is make sure that you have a low occurrence ranking (i.e. mistake proofing) and a low detection ranking to counter the high severity.
Helmut Jilling 26th July 2006, 06:33 PM All,
Interesting question.
Automotive customer has reviewed our FMEA for an electronic product, and determined that a severity of 10 exists for a specific failure mode. The customer's systems cannot detect or dianose this failure mode, leading to potential of a severity 10 for this failure mode.
In the past, I have been told that the Vehcile OEM's will not allow 10's to be passed off to their customers. As a supplier, this is causing us grave concern.
Should we force our customer's hand to resolve this? No one wants to expose consumers to risk.
Belden
A lot of misinformation in your statements. Please refer to the 3 tables in the PFMEA section of the AIAG FMEA book, 3rd ed.
Whether the customer's system can "detect" the failure or not does not effect the Severity rating. The definitions for Severity are pretty clear in the Severity table (pg 43).
The detectability is evaluated by the Detection table on pg 53. Ideally, that failure is soemthing YOUR system should detect. The final variable to lower the overall RPN score is how often does it occur (Occurance Table on pg. 49). The three scores are rated independently, then multiplied.
There is no rule whether customers will accept a part with a 10 severity. However, it is generally regarded that only a design change will lower a severity rating, and I am guessing you don't have design responsibility for it?
Beldenray 26th July 2006, 09:38 PM Actually, this post was written NOT to address the specifics of the RPN (which is pretty low), but rather to address the specific liablity that could be associated with a product that could cause human injury without warning. This seemed like the best place to address this issue, since it is tied in with risk analysis.
In my experience with OEMs, I have been told "Your products do not have severity 10. The system diagnositics cover that issue." However, in this instance, there are absolutely NO system diagnostics with cover this failure mode. Also, the product was designed similar to products that work in systems with dianostics that could catch this failure mode. (Our low occurance comes from this data of previous designs.)
How common is see this sort of risk for automotive products, and how have you responded to them?
Belden
Helmut Jilling 27th July 2006, 12:39 AM Actually, this post was written NOT to address the specifics of the RPN (which is pretty low), but rather to address the specific liablity that could be associated with a product that could cause human injury without warning. This seemed like the best place to address this issue, since it is tied in with risk analysis.
In my experience with OEMs, I have been told "Your products do not have severity 10. The system diagnositics cover that issue." However, in this instance, there are absolutely NO system diagnostics with cover this failure mode. Also, the product was designed similar to products that work in systems with dianostics that could catch this failure mode. (Our low occurance comes from this data of previous designs.)
How common is see this sort of risk for automotive products, and how have you responded to them?
Belden
I have seen Severity = 10 in some cases, but not very often. Seven, eights and nines, also. But, Severity can only be reduced by redesigning the product is an established axiom in automotive, and you don't have that authority. So, the severity is what it is. You can only address detection and occurance with mistake proofing and robust inspection.
By the way, what is this particular failure mode we are discussing, that is so severe and undetectable?
Michael Walmsley 27th July 2006, 09:48 AM A 10 is a 10 is a 10!!!!:caution:
I've worked on most of the automotive electronic systems (Tier 1 and customer level) to know that a 10 is a product recall waiting to happen.
It will. Eventually. Regardless of the Occurence rating in an FMEA.
The only way out is to redesign.
There is something you are not telling us.
How close are you to product release????
You are probably under tremendous pressure by your management and customer management to downplay the issue.
Your management does not want to lose the business by confronting the customer and requesting a delay to redesign. Spineless idiots!
In the long run your reputation will suffer more by not doing the right thing.
Your customer is probably trying to get you to reduce the severity / deleting the mode from the FMEA.
In this way they can always come back later and blame everything on you if and when it does happen.
You , as a release / design engineer are caught in the middle. When a liability lawsuit does happen you will be in the hot seat in a court of law.
Your only option as a supplier is to contact your legal department with the potential issue. You will most likely be requested to write an electronic memo to your customer,explaining the issue,potential outcomes of the issue and to seek their guidance. It is now in their court.
If they seek to release the product as is,there will be "Joint" liability shared.
If not, then pursuing a redesign is the only option.:agree:
howste 27th July 2006, 11:34 AM Actually, this post was written NOT to address the specifics of the RPN (which is pretty low), but rather to address the specific liablity that could be associated with a product that could cause human injury without warning....
How common is see this sort of risk for automotive products, and how have you responded to them?
Belden
Maybe I'm crazy, but I believe that cars are shipped every day with hundreds (if not thousands) of severity 10 potential failure modes. If my steering wheel falls off, that's severity 10. If my gas tank explodes, that's severity 10. The key is that it's not very likely for my steering wheel to fall off or my gas tank to explode.
As far as liability goes, a 10 is a 10 and liability is liability. If a steering wheel falls off, you can bet there will be an investigation and one or more organizations will be held liable. You mentioned that your customer's diagnostics reduce the severity of this failure mode on your other products. If your customer's diagnostics failed to detect a problem with your product, does that mean you have no liability? :nope:
To directly answer your question: A very low percentage of failure modes are severity 10, but there are still thousands of them. In the cases where I've had to deal with them, we have responded by having very good controls to prevent the failure mode, and very good controls to detect them if they occur.
Michael Walmsley 27th July 2006, 11:59 AM I disagree.
NA automotive Steering wheels not only have a lock nut to attach them to the column, but also a lock key as a secondary backup. You will notice freeplay vertically weeks prior to impending release. This acts as a "warning" to the end user. It then places it into a sev level =9.
Fuel systems have shields , bypass vents , labled warnings which also place them into a sev=9 category.
I disagree as to the hundreds and thousands of sev = 10 being on a particular group of vehicles in my experience. Nothing would be sold if this were the case.
The 10 is a red flag. I agree with the need for stringent controls for them,if there are controls that can be had.
I think the original concern was for the lack of any controls to be had.
I also think that they were not digging deep enough to look.
Warning mechanisms and fault/control diagnostics are the chaepest and fastest means for redesign,followed by redundancy.
Helmut Jilling 27th July 2006, 11:49 PM A 10 is a 10 is a 10!!!!:caution:
I've worked on most of the automotive electronic systems (Tier 1 and customer level) to know that a 10 is a product recall waiting to happen.
It will. Eventually. Regardless of the Occurence rating in an FMEA.
The only way out is to redesign.
There is something you are not telling us.
How close are you to product release????
You are probably under tremendous pressure by your management and customer management to downplay the issue.
Your management does not want to lose the business by confronting the customer and requesting a delay to redesign. Spineless idiots!
In the long run your reputation will suffer more by not doing the right thing.
Your customer is probably trying to get you to reduce the severity / deleting the mode from the FMEA.
In this way they can always come back later and blame everything on you if and when it does happen.
You , as a release / design engineer are caught in the middle. When a liability lawsuit does happen you will be in the hot seat in a court of law.
Your only option as a supplier is to contact your legal department with the potential issue. You will most likely be requested to write an electronic memo to your customer,explaining the issue,potential outcomes of the issue and to seek their guidance. It is now in their court.
If they seek to release the product as is,there will be "Joint" liability shared.
If not, then pursuing a redesign is the only option.:agree:
Sounds like you give some very smart advice...
|
|