PFMEA for Heat Treat Service Provider - Trying To Develop RPNs

[Tom W]

PFMEA for a Service Provider

Does anyone have any suggestions as to a service provider (heat treat) trying to develop RPNs with the criteria in the AIAG FMEA third edition manual. I find it difficult at times to select the appropriate rating based on the manual being developed for manufacturers, assembly plants and actual automakers. Our customers rarely, if ever, work with us in developing the quality plans.

For example - we have incorrect material being one of our potential problems; which is 100% out of our control. If the wrong material is used or the wrong chemistry was used for the material, and we heat treat to the customer specification and the results are noncompliant, we inform the customer that the material might be in question and then wait to hear from them. There are times that we can rework the material to get the desired results, but we charge them for this rework. Most of the time it could lead to scrap.

Would this calsify us as an 8 on severty or even a 5. If we are dealing with wrong material - it is a 100% problem, there is no sorting. Would I be wrong in trying to interperate the criteria and developing our own terminology to match the AIAG descriptions, or do I have to just suck it up and make it fit?

Square peg in a round hole!!!!!!!!!

 

[Bill Ryan - 2007]

Tom,

It sounds like you might be getting Failure Mode mixed with Failure Cause . As I see it, a couple of Fialure Modes in your PFMEA would be something to the effect of "Part fails to reach hardness specification' and/or "Part exceeds hardness specification". The severity of the Failure Mode would most likely depend on its useage - Safety would get you in the 9-10 range - a structural failure not related to safety (if that's possible) would put you in the 7-8 range.

One of the Potential Causes might certainly be "Customer sent out of specification material" and would probably have a pretty high Occurence rating, if I understand your post. If you choose to have this as a potential cause, you would have a pretty significant RPN value with no "Recommended Actions" as you have no control over what your customer sends you. Another basic "premise" of PFMEAs is to assume the product entering your process flow is "correct". In that case you would have Potential Causes such as "Time setting - Set incorrectly"; Temperature setting - Set incorrectly"; "Quench time - Set incorrectly" (T6??) and so on.

We have the capability of performing "T5" heat treat in-house and I do not address the "out of spec. material" as a Cause at the Heat Treat operation. I have that addressed at the Foundry and Holding Furnace operations.

I don't know if this helps you at all, but my feeling is that if you try to address your customers' issues in your PFMEA, you'll have a big-time nightmare on your hands.

Good Luck,
Bill

 

[Tom W]

Thanks for the reply Bill - we have a unique approach to the point that we have developed what we call generic quality plans for our "standard service" processes. The generics are the ones I am speaking about. I review them once a year and this year I am trying to get better understanding of how and why we came up with the numbers. I will attach one to this post and if you could please take a look at it and give me your thoughts after looking at it I would appreciate it. Thanks.

 

[Bill Ryan - 2007]

Tom,

I'm only going to address the "Receiving" operation for now. Please understand I'm only giving you my thoughts on how I might address things.

As I look at the Failure Mode - "Wrong material or Chemistry / Microstructure" - it seems that you are trying to address your customers' issue. I don't see where you are inspecting the chemistry or microstructure in the "Current Process Controls" column. I also don't see the "Current Process Controls" addressing the Potential Causes. Once again, these causes seem to be addressing your customers' process.

What is the expected outcome of your receiving operation? If it is to ensure lot counts, label legibility, correct packaging, etc., then the Failure Modes for Receiving should address those types of things. If your customer sends you a metal certification, then it would make sense that you might review each certification but I would still handle the columns differently.

I'm going to proceed on the assumption that you don't have a metal certification to review. If you'd like to keep the Failure Mode as you have it, I would address the Potential Cause column as "Customer sent out of spec. material" (or something to that effect). You don't have a "Current Process Control" to detect the out of spec. material unless you perform a hardness or impact test downstream in your process flow. Whatever your confidence level is in detecting a "nonconforming" lot of product (with respect to Metal Chemistry/microstructure, your DET value must reflect that confidence level (I would probably have no less than a 7).

If you'd like to address Failure Modes that you have "control over" (truly, the only ones I'm interested in dealing with), you might have "Shipper quantity differs from actual". Potential Effect" could be "Inaccurate inventory levels" (probably less than an 8 for SEV.). "Potential Cause" - Customer sent incorrect quantity". and the Current Process Control" could be that you "Weigh count each skid".

I know this is a pretty simplistic example but without having a fuller understanding of your Process, I can't go too deep.

A couple of questions I have for you are - Why does the Effect "Discolored, stained, or spotted surface" have a SEV of 8? - Do you truly believe a "Visual Inspection" can have a DET value of "2"?

I hope this makes some sense and is some help to you. I'll be around if you'd like to go into more and, hopefully, a few other "Covers" will jump in with their ideas 'cause I'm far from being a PFMEA "guru" (but I've had good success arguing "content" with each of my customers).

If you'd like, I'll see if I can post an example of how I address Receiving (my software is not the friendliest around but I should be able to get it into a .pdf file).

Until later,
Bill

 

[Tom W]

Thanks Bill - you comments are appreciated. Thanks for taking your time to respond. You bring up some excellent points (some that I have been asking here for a while and just get blank stares back). We have some customers that demand bright clean product out of the furnaces - thus any discoloration or stains would cause them to reject the parts - has no negative effect on fit, form or function, they just don't want it because it doesn't look right.

I am going to shock the world around here and try to put some real numbers on these (we have 33 standard service quality plans for 33 different standard processes that we offer.) The standard servcie quality plans can be used as is or a customer can request specific ones for their part numbers, in which case we use the standard service ones a foundations and build on to them. But for the customers that are not automotive and could care less about a FMEA or Control Plan (i.e. tool steel customers) we can always fall back on the standard service (generic) ones for our own files.

Thanks again for you comments and any more that you could give would be appreciated.

 

[Paul Simpson]

Assume Good!

Hi, guys. I've just dicovered this thread so apologies if I look as if I'm crashing the party.

I do some FMEA training and the first thing I emphasize is you have to know your boundaries. With a process FMEA you draw the boundary around everything you have control of. IMO that means you assume that everything you receive is good. If your customer (who also supplies the parts I presume)wants to deal with the risks of the wrong material getting treated they should deal with it in their PFMEA fior labelling and ID.

Being the customer they may insist you get involved (i.e. write) the FMEA for wrong material getting treated. In which case they have to give you the info on severity of effect (could be 8 for loss of function, could be 10 if safety critical). Likewise they have to give you information on likelihood of occurence - how many times have they mis identified materials in the past!

I'm sure this could ruffle some feathers but that is the way FMEA is intended to work - what can we get wrong in the areas we have control of?

Good lucl!

 

[M Greenaway]

Paul

I believe it says as much in the QS9000 manuals.

 

[D.Scott]

Paul, I suppose I could be reading your post wrong but I have to disagree that the responsibility lies with the customer.

IMO that means you assume that everything you receive is good. If your customer (who also supplies the parts I presume)wants to deal with the risks of the wrong material getting treated they should deal with it in their PFMEA fior labelling and ID.

You need a system to verify that whatever comes in is in fact good. If a customer sends the wrong part, your system should be able to address it. If parts are mixed coming in, something in the process should should address it.
I agree a customer needs to be involved with development of RPN but only in regard to severity. The occurrance and detection should relate to your process. The customer probably knows his process and realizes he sends mixed parts. He needs to know how often does it occur in your process and how likely are you to catch it when it happens.
Control includes everything in my process which includes identification of materials, validation, testing, protection and handling of the customers product, etc. If his process failures can create a failure mode in my process, it is up to me to deal with it, not him. That certainly doesn't mean the customer shouldn't be involved prevention upstream if possible
Now, if you meant addressing the customers process on the FMEA, I agree - N/A.
Just my opinion.

Dave

 

[Paul Simpson]

Who can do it?

Thanks for the replies.

Martin. I believe you are right that it says it in the manual (I don't spend a great deal of time with them). The post was just trying to clear up some general misunderstandings.

Dave. There may be something in your system (e.g. goods inwards inspection) that prevents the failure of your customer's system from having a more severe effect (i.e. by catching it at your site before it gets back to them), the potential severity score is likely to be the same as you cannot guarantee it won't get through your systems and through the customers receiving and manufacturing systems. You cannot do anything about the likelihood of occurence, either as that is all at your customer. All you can do to help out a bit is to do more inspection on their parts at your site, hardly the stuff of quality assurance and efficient operating!

With potential mixed parts you would have to be doing a high level of inspection to find "rogue" mixed parts and I don't believe the customer would pay for that. Better that they get their process right through Process FMEA.

 

[Bill Ryan - 2007]

If his process failures can create a failure mode in my process, it is up to me to deal with it, not him. That certainly doesn't mean the customer shouldn't be involved prevention upstream if possible
Just my opinion.
Dave

Sorry to dredge this back up (not even sure how I ran across it ),but I just have to say that if my customer expects me to sort out the failures of his process, he's going to be paying a "pretty penny" for that "service".

Bill