PFMEA Severity - What is Process FMEA Severity estimation based on?

[Phil Huber]

I've been away from the Cove for a bit, have read all the comments in this forum and will make sure to visit much more often!! Great work to all!!

I have found in doing FMEA's within a number of industries that the team will get hung up on one column and forget that we are looking for priority numbers to ferret out the most significant problem(s). Keep in mind that severity numbers of a 9 or 10 typically involve safety or damage concerns either with or without notice. A design change may be required to eliminate these failure modes.

Don't be afraid to use high severity numbers. If the occurance is low and detection is high, the RPN will tend to drop this failure mode below radar. If this is not the case, well then the FMEA did as intended.

Consider the use of severity x occurance to calculate the Initial Risk. Typically, IR's of 35 or greater require attention; use the RPN to sort out which ones to work on first. This gets away from the old rules still hanging around out there of RPN's of 100 or greater (or whatever). Some how when those rules are applied it is amazing to find that there are no RPN's that reach that value!

Keep up the great work!
Phil

 

[PaulMac]

Perhaps you are misunderstanding the purpose of the Severity column. The focus is to evaluate Severity impact on the customer and the end user, not on you. (The Detroit 3 don't care about your impact). The focus is to evaluate the Severity impact that a process failure like a broken tap would have. You compare the rankings in the table to the expected impact (ie: like a broken tap) on the customer assembly and the end user. The rankings in the table give you some guidance on what ranking value to assign. It is not a precise science, but is a legitimate analysis and assigning of risks, to help you focus attention on the more significant issues.

Thanks Helmut,
I have had thoughts that we may be overcomplicating the pFMEA process by trying to focus both on external and internal effects. My life would be easier if I resolved to only consider external effects, except in cases of safety in the building of the product. I may end up there yet. But I wanted to explore a bit more the value in using pFMEA for addressing internal process impacts which can help lower costs for us and the customer. However, I'm not convinced that the benefits outweigh the extra complexity and questions raised - especially if it causes the pFMEA to be bogged down, which is one of the complaints of my current employer.

I think one way to make the internal ratings less confusing is to express process FMs as process failures. This essentially brings the nature of the process failure to the forefront of the FMEA analysis, allowing for a better internal impact assessment without going to the next step - causes. That way we can focus on the high impact stuff first. This also does not limit the team from identifying the end user effect, which now becomes the potential end product failure, due to the product symtom "effect", due to the process FM in the pFMEA. This approach seems to focus the pFMEA team more on understanding the relationships between product and process requirements - which is a good thing.

Still on the fence though. But getting closer...

Thanks all.

 

[PaulMac]

Any other thoughts out there on this topic in regard to my comments/questions? This has been a great help so far. Thanks !

 

[karthikkandasamy]

Dear All,

I am Karthik Kandasamy working as Quality Engineer in India's leading Grey and S.G Iron Auto Components Manufacturing Organization. We are supplying Steering Knuckle, Brake Drums and Brake Discs to all Major OEM's.

While Preparing FMEA document we found very difficult to match the Occurance ratings provided in AIAG manual with current occurance level in our manufacturing process. Usually we are a casting manufacturing foundry we are operating at 7 ~ 10% rejection level ( Overall rejection say Blowholes, sand drop, inclusions, shrinkage etc as our foundry defects which occur in all manufacturing processes). When compared to AIAG manual our current occurance rating falls 9 and 10. So it reflects the RPN ratings above 350 ~ 600. Our customers are asking to take action for High RPN numbers above 100. World class casting foundries are operating at 5% rejection level. 5 to 10% rejection in casting foundry is very common and it is a good benchmark level compared to world class foundries.
So any body working in casting foundry may help me
to take actions or any occurance table specifically made for foundry process. pls do needful.

 

[Jim Wynne]

Dear All,

I am Karthik Kandasamy working as Quality Engineer in India's leading Grey and S.G Iron Auto Components Manufacturing Organization. We are supplying Steering Knuckle, Brake Drums and Brake Discs to all Major OEM's.

While Preparing FMEA document we found very difficult to match the Occurance ratings provided in AIAG manual with current occurance level in our manufacturing process. Usually we are a casting manufacturing foundry we are operating at 7 ~ 10% rejection level ( Overall rejection say Blowholes, sand drop, inclusions, shrinkage etc as our foundry defects which occur in all manufacturing processes). When compared to AIAG manual our current occurance rating falls 9 and 10. So it reflects the RPN ratings above 350 ~ 600. Our customers are asking to take action for High RPN numbers above 100. World class casting foundries are operating at 5% rejection level. 5 to 10% rejection in casting foundry is very common and it is a good benchmark level compared to world class foundries.
So any body working in casting foundry may help me
to take actions or any occurance table specifically made for foundry process. pls do needful.

The ratings in the manual are explicitly identified as suggestions, not requirements. The manual says that if you want to use an alternate method, you can, so long as you use it consistently. With regard to occurrence, you can use a simple 1-10 scale where 1 means that there's practically no chance of occurrence and 10 means occurrence is inevitable.

If your customers insist on the use of the PFMEA manual suggestions, first try reasoning with them. If that doesn't work, and you're being asked to change the laws of physics, you either have to change the laws of physics or "adjust" your PFMEA documents such that they reflect what the customer wants to see. Such adjustment involves prevarication, of course, but if that's what your customers are (unwittingly) demanding, give it to them.

 

[bobdoering]

The ratings in the manual are explicitly identified as suggestions, not requirements. The manual says that if you want to use an alternate method, you can, so long as you use it consistently. With regard to occurrence, you can use a simple 1-10 scale where 1 means that there's practically no chance of occurrence and 10 means occurrence is inevitable.

You may also consult your customer specific requirements. Some have their own rating system they expect you to use.

If your customers insist on the use of the PFMEA manual suggestions, first try reasoning with them.

It is good to have some understanding of this prior to the quoting process (as in feasibility phase). The reason is there are 3 participating factors to the RPN. Your ranking for Occurence may be accurate. What is the Severity? Is it accurate? Most importantly, what is the Detection? If it was a 1, you would likely have a decent RPN number. But, a low detection number requires some very sophisticated implementations - especially in casting. That being the case, the cost would be significantly impacted. Here's the rub - your attempt to prepare a process with a strong, low RPN will be readily undercut by people who ignored the point, and then they feud with the customer for the life of the part. Unfortunately, customers encourage that with their cost control measures - never looking comparing FMEAs of cheap product to consistent product. Reality clashes with theory.

 

[Stijloor]

<snip> If your customers insist on the use of the PFMEA manual suggestions, first try reasoning with them. If that doesn't work, and you're being asked to change the laws of physics, you either have to change the laws of physics or "adjust" your PFMEA documents such that they reflect what the customer wants to see. Such adjustment involves prevarication, of course, but if that's what your customers are (unwittingly) demanding, give it to them.

Very sad but true. :mg:
Adjusting and compromising the integrity of the FMEA process to please an ignorant customer (representative).
Defeats the true purpose of FMEA.

And so it goes.....

Stijloor.

 

[Chennaiite]

Dear All,

I am Karthik Kandasamy working as Quality Engineer in India's leading Grey and S.G Iron Auto Components Manufacturing Organization. We are supplying Steering Knuckle, Brake Drums and Brake Discs to all Major OEM's.

While Preparing FMEA document we found very difficult to match the Occurance ratings provided in AIAG manual with current occurance level in our manufacturing process. Usually we are a casting manufacturing foundry we are operating at 7 ~ 10% rejection level ( Overall rejection say Blowholes, sand drop, inclusions, shrinkage etc as our foundry defects which occur in all manufacturing processes). When compared to AIAG manual our current occurance rating falls 9 and 10. So it reflects the RPN ratings above 350 ~ 600. Our customers are asking to take action for High RPN numbers above 100. World class casting foundries are operating at 5% rejection level. 5 to 10% rejection in casting foundry is very common and it is a good benchmark level compared to world class foundries.
So any body working in casting foundry may help me
to take actions or any occurance table specifically made for foundry process. pls do needful.

Before taking a dig at your Customers, I would suggest you to make an introspective study of your PFMEA. Few things I can technically suggest (I have some sort of exposure in Foundry):
As far as Potential Failure Mode(PFM) is concerned, try to consider your process failures rather than part. I take an example, low Mould harness as a Process failure can potentially cause Sand Inclusion (if I am correct), a part failure. Better to consider the former as the PFM while you analyse Moulding process. This is because, Inclusion is a Part failure and it can also happen due to other processes such as Pouring, Core Setting, etc. Directly applying your overall rejection percentage of the defect to decide the Occurrence rating may exaggerate the risk.
If you are really confident about your PFMEA, you always have a point to take up with your Customer. Keep the so called adjustment of PFMEA as your last option, as you will be trapped in a situation to do so with every single Customer representative.

 

[karthikkandasamy]

Thank you very much for your replies.

In all casting process severity rating may be 7 or 8 because if there is any deviation in process parameters, portion or 100% components have to be scrapped or it may gets detected during machining or assembly. Inprocess check is carried out at some stages but product assurance is done only at final inspection. Here our final inspection is visual check for surface defects and gauge check for dimensions. So our detection rating is also falls 6 ~ 8. Error proof check for foundry casting process is too costly and at present there is no any error proof check available in markets. So the RPN values exist between 350 ~ 600.

All our customers are OEM's. They are well aware of machining and assembly process, not completely aware of casting foundry process. In machining and assembly process there is possiblity to provide prevention and detection error proof for all major process and the rejection level is also well below 1%. So RPN value can be maintained below 100. But it is not possible to maintain the RPN value in foundry casting process.

One customer insist us to develop our own strategy for occurance rating and get approval from all customers. But we dont know how to make such occurance rating. Is there anybody have such occurance ratings for foundry casting defects, pls share with us.

 

[speck]

Re: Severity estimation in a PFMEA is a bit confusing

My own opinion is that process FMEAs should be concentrated on processes, not products. There will be disagreement in this regard, but to me, it's fairly simple. If a customer has done a DFMEA, and the output of that process is reflected in the product specifications, and the product is manufactured in accordance with the specifications, everyone should be happy, all else being as it should be. This means that I, as the producer of the product, should be able to concentrate on designing a process that will produce product that meets the specifications, and I shouldn't need to concern myself with what might blow up in end use if the product doesn't conform--that's the designer's responsibility.

So to answer your question, I think that the PFMEA Severity factor should be based on the risks of the producer, not the customer.

I know this is an older post, but I found it directly relevant to the matter at hand and the new AS9100C risk management requirements. After reading the entire post, I think this specifically defines my question, but I don't think I can yet answer that end question that I started to search the cove for.

I believe that as the new requirements of AS9100C are undertaken by me (us), my (our) approach for our heat treating facility is to base my (our) risk management processes off of our PFMEA's (we are design exempt). There will be other factors and aspects added in and taken into account for sure. I am reviewing and revising what we have and realize that I am in total agreement with Jim, but many people would say we are wrong. This troubles me. Currently, our PFMEAS really do not focus on end user risks as some would have it.

As a commercial heat treater, I am certainly not qualified to say that the design responsible party does not, or has not already taken into account what aspects of heat treating could affect their customer or the products end use. I know nothing about their widget. I in turn will, and have, (or attempted to) reduce those risks that I directly contribute to or create, through controlling our processes.

For example we may receive parts with specific hardness, case depth, flatness, core hardness, retained austenite and maximum T.I.R. requirements for key characteristics.

They have already determined that Intergranular oxidation beyond .0005” deep is acceptable (to others) it is not; otherwise they would have required that in the specifications or requirements to us.

In a perfect manufacturing world, that customer would or should have contacted their professional heat treater which is more competent in heat treating than they are, and asked for, or taken a copy of our PFMEAS and incorporated them into their DFMEAS. They have to account for all possible problems in production resulting from all processes, no? If the design company performs this with all its supplier processes they will incorporate those issues into their DFMEAS and all bases will be covered to THEIR end user through a very thorough DFMEA.

While reading this entire post one question kept nagging me that no one brought up or asked; if we should be mitigating end user risks in our PFMEA’s, why are there still PFMEA’s and DFMEA’s. Why doesn’t everyone in the processing chain pretend to know exactly what the widget is used for, the constraints in which it will be used, what design flaws could lead to death or just dismemberment etc. etc. We could all perform the new “process and design FMEA’s” and do away with PFMEA and DFMEA’s individually.

Maybe I am oversimplifying things but in my mind it seams as though the DFMEA is a compilation of ALL the processes FMEA’s with the end user product FMEA’s mixed in, which only the design authority can fully understand.

I have to admit I am not nearly as trained in automotive PPAP and FMEA’s as others here but I believe my question is valid. When I read Jim’s original statement things became crystal clear, but soon realized there are many people who disagree with our logic.

Thoughts?