PFMEA for an Assembly Process - Stuck in first steps

[AlbertV]

Good morning Gents and ladies,

I'm leading a team (for the 1st time) that is doing PFMEA for an assembly process, and we are stuck in the first steps of it as we fail to find a rational way of doing it. We have tried different ways, but all of them have led to a doubtful path.

I'll describe the process: we have a steel part with a machined hole to a certain diameter and a rubber seal that must be assembled to prevent air leakage through it. Assembly process goes like this: operator dips seal in soap solution, loads it to assembly machine, loads steel part and the machine nails the seal in the hole. In a further operation, the part is run through a leakage test to check if it is sealed.

There are a couple of logics we have followed that didn't lead us to a satisfactory outcome:

1) Break assembly operation in small bits: first bit is dipping operation, with a Requirement like 'dip complete part'. Failure mode could be 'seal is not completely dipped', effect could be 'seal cannot be completely assembled, therefore it will leak when used by the customer', and cause 'standard work not followed'. We know that our leakage test device effectively protects our customer from receiving it, but it detects the EFFECT of the failure mode, so it should not be considered... Following this logic, we get a high RPN when we know that this failure mode will never reach our customer.

Moreover, one of the following bits of the process is the assembly movement of the machine, with a Requirement 'nail seal into its final position'. One of the potential failure modes for it is 'seal not completely nailed', which is the effect of the dipping operation failure mode. The root cause for this is the same as it was in the dipping operation... Are we working double?

2) Consider assembly operation as a whole: assembly operation requirement is 'nail seal into its final position'. One of the potential failure modes for it is 'seal not completely nailed', the effect is once again 'leakage at customer'. One of the root causes is once again 'part not completely dipped in soap solution'. Again high RPN because leakage test is not considered...

Sorry for the long write. Any help will be very appreciated, as we are running in circles and the team morale is quickly lowering.

 

[AlbertV]

I keep reading about PFMEA's in the Cove. This thread has been very clarifying: How Should I Describe a Failure Mode?. Now I understand that I must decide which way to approach failure modes in PFMEA (as 'product modes' or as 'process modes').

Also, now I see the need to focus the scope of the analysis instead of bouncing around from operation-level (high-level) and step-level (detail-level). Last post on previous thread from adamsjm explains it clearly: https://elsmar.com/Forums/showthread.php?p=462688&postcount=24

Anyway, experienced inputs on this issue are more than welcomed.


Moderator Note: Post edited to add live hyperlinks.

 

[Jim Wynne]

I keep reading about PFMEA's in the Cove. This thread has been very clarifying: How Should I Describe a Failure Mode?. Now I understand that I must decide which way to approach failure modes in PFMEA (as 'product modes' or as 'process modes').

Also, now I see the need to focus the scope of the analysis instead of bouncing around from operation-level (high-level) and step-level (detail-level). Last post on previous thread from adamsjm explains it clearly: https://elsmar.com/Forums/showthread.php?p=462688&postcount=24

Anyway, experienced inputs on this issue are more than welcomed.


Moderator Note: Post edited to add live hyperlinks.

I edited your post to include the links. You're on the right track by searching. I suggest that you also look to past failures in the process to try and understand what needs to be controlled.

As far as format is concerned, you should probably describe the overall process in the leftmost column of the PFMEA form, and then each step or operation underneath that. When trying to determine what constitutes and individual step, think in terms of the possible process failure modes. If a potential failure is unique to a given process step, that process step should get its own line item on the PFMEA form.

 

[Chennaiite]

1) Break assembly operation in small bits: first bit is dipping operation, with a Requirement like 'dip complete part'. Failure mode could be 'seal is not completely dipped', effect could be 'seal cannot be completely assembled, therefore it will leak when used by the customer', and cause 'standard work not followed'. We know that our leakage test device effectively protects our customer from receiving it, but it detects the EFFECT of the failure mode, so it should not be considered... Following this logic, we get a high RPN when we know that this failure mode will never reach our customer.

I wouldn't mind considering detection control of the effect for rpn calculation provided it reflects the risk in appropriate sense. As much as I understand your process, leakage testing is the mechanism to detect failure mode/cause, albeit not directly.

 

[AlbertV]

Thanks both for your help. We have followed your advice and today meeting has been much more productive

Lesson learnt for me is: don't get stuck with how PFMEA should be done, just do it the way it is useful for you. Decide the way that it suits you better and stick to it.

 

[Jim Wynne]

Thanks both for your help. We have followed your advice and today meeting has been much more productive

Lesson learnt for me is: don't get stuck with how PFMEA should be done, just do it the way it is useful for you. Decide the way that it suits you better and stick to it.

So long as your customers don't squawk about your chosen method, you should be just fine.

 

[megahertz]

PFMEA should have more concern on process. As such it is better to break into detail steps rather than putting everything as a whole and it will be more easy if you put in process number accordingly. For an example, assembly process may start from 3000 to 3999 and each sub process in assembly should be numbered with 3001, 3002, 3003 and so on.