When to update a PFMEA (Process FMEA)

[Helmut Jilling]

...that the P FMEA should be updated when important changes are needed and not just a clerical thing.

If I've made no process changes...I see no need to update the P FMEA.

I still struggle at times with the PFMEA. I look at the PFMEA as a way of identifying potentials failures with machine process parameters or subsequent operations during the manufacturing process....

.....We now use a form that comes from Engineering that when mold changes are done the Engineering group based on the change will know what features cold be changed and the lab is verify all before PPAP.

How does this go into a P FMEA? I look at this a more of a system failure and not a process failure that goes into a P FMEA.

You are correct on each statement. The PFMEA is a potential process failure analysis exercise, and should be updated when your analysis has determined a potential failure risk has chnaged.

It is not intended as a clerical punishment. There is too much unnecessary documentation being put into our TS programs. Unnecessary documentation is "unnecessary." Keep it simpler.

 

[Jim Wynne]

It is not intended as a clerical punishment. There is too much unnecessary documentation being put into our TS programs. Unnecessary documentation is "unnecessary." Keep it simpler.

I like the phrase "clerical punishment." Unfortunately, there's a lot of that sort of flogging going on, and sometimes automotive suppliers are forced into it by customers, and by reluctance to question auditor findings that don't make sense. If everyone took your pragmatic approach, everyone would be a lot happier, especially the customers.

 

[Stijloor]

I like the phrase "clerical punishment." Unfortunately, there's a lot of that sort of flogging going on, and sometimes automotive suppliers are forced into it by customers, and by reluctance to question auditor findings that don't make sense. If everyone took your pragmatic approach, everyone would be a lot happier, especially the customers.

PPAP documents (that are rarely reviewed) and the ongoing explosion proliferation of customer-specific requirements are examples of "clerical punishment." All suppliers complain and/or suffer in silence, few push back, and on and on it goes. Annual automotive Customer-Supplier conferences do not help. Maybe the economic downturn may create a much needed change to eliminate the non-value added requirements that are imposed on the suppliers.

Just some thoughts.

Stijloor.

 

[Kevin Mader]

Jerry,

It seems to me that the issue is with regards the interpretation of the three subject customer complaints. By your assessment, these items were inclusive in your original assessment. By the way the NC was written the auditor seems to be suggesting that the subject complaints are a) new and not previously identified and b) original assumptions are not challenged (perhaps the S-O-D needs revisiting?).

To combat this, do your procedures establish an algorithm to determine when a complaint identifies a new, previously unidentified failure mode? Also, do your procedures indicate how you challenge established PFMEAs (including RPNs)?

What might not be clear to the auditor is how you have connected the complaint process back to risk management, design controls, process controls and perhaps how you document a ‘no action needed’ decision in your records.

I think your response should have been successful if these things are in place. If that's the case, then you just got unlucky with your auditor.

Regards,

Kevin

 

[Helmut Jilling]

Jerry,

It seems to me that the issue is with regards the interpretation of the three subject customer complaints. By your assessment, these items were inclusive in your original assessment. By the way the NC was written the auditor seems to be suggesting that the subject complaints are a) new and not previously identified and b) original assumptions are not challenged (perhaps the S-O-D needs revisiting?).

To combat this, do your procedures establish an algorithm to determine when a complaint identifies a new, previously unidentified failure mode? Also, do your procedures indicate how you challenge established PFMEAs (including RPNs)?

What might not be clear to the auditor is how you have connected the complaint process back to risk management, design controls, process controls and perhaps how you document a ‘no action needed’ decision in your records.

I think your response should have been successful if these things are in place. If that's the case, then you just got unlucky with your auditor.

Regards,

Kevin

While you may be correct, the auditor may like this, it is an example of the overdocumenting I was complaining about.

I argue our procedures should give the guidelines and best practices that we have learned over the years. I do not recommend they should be so heavily defined that they attempt to answer every question and situation.

If intelligent personnel are not permitted to use good judgement, we have lost the benefit of our systems.

 

[Kevin Mader]

Helmut,

Oh, I don't disagree that we seem to spend inordinate amounts of time over proving the reasonable person theory. Inevitably, however, it is what the Objective Evidence demonstrates what happened versus what we say happened. Some auditors remain reasonable while others demand the OE. In regulated industries, 'suggestions of what occurred' are a last resort. Inspectors for the governments respond to what is written and verifiable. Vagaries are generally frowned on in these industries. But to your point, no algorithm will be complete and arguably, the more complete the more complete the trap for when you have missed something an auditor finds relevant.

The key is to establish a method commensurate to the task at hand and the risk involved. Loose requirements followed up by loose specifications lead to undesirable levels of variation and interpretation. This is why, of course, we painstakingly establish procedures against guidance documents. We are specifying our method.

Regards,

Kevin

 

[Helmut Jilling]

...
Oh, I don't disagree that we seem to spend inordinate amounts of time over proving the reasonable person theory...

...But to your point, no algorithm will be complete and arguably, the more complete the more complete the trap for when you have missed something an auditor finds relevant.

The key is to establish a method commensurate to the task at hand and the risk involved. Loose requirements followed up by loose specifications lead to undesirable levels of variation and interpretation. This is why, of course, we painstakingly establish procedures against guidance documents....

I like to recommend the Goldilocks approach:

Too loose is not enough,
"painstaking" is too much,
and we want to find "just right."

The OE should bear out what "just right" should look like.

 

[Kevin Mader]

Helmut,

That makes sense to me. Any auditor worth their weight should be able to rely on the OE to determine compliance.

Regards,

Kevin