PFMEA vs MFMEA or DFMEA on equipment/machinery

tugboat

Registered
It's a process failure exercise. The AIAG manual is wrong on several fronts, imo, and this is one of them. Ask yourself how the process might fail, and what the consequences (effects) might be. Product defects are not process failure modes.

This is starting to sound to me like these are philosophical differences that people have innate preferences for. Would it not be helpful in a PFMEA to brainstorm product defects and how those defects would affect the process? Or do you accommodate those, just in a different column (failure effects?)?
 

Jim Wynne

Leader
Admin
This is starting to sound to me like these are philosophical differences that people have innate preferences for. Would it not be helpful in a PFMEA to brainstorm product defects and how those defects would affect the process? Or do you accommodate those, just in a different column (failure effects?)?
There are bound to be differences in approach. I'm just telling you what's worked best for me for 30+ years from both the supplier and OEM ends of things. I think that brainstorming potential product defects is, or should be, the second step after you determine ways that the process might fail.
 

tugboat

Registered
There are bound to be differences in approach. I'm just telling you what's worked best for me for 30+ years from both the supplier and OEM ends of things. I think that brainstorming potential product defects is, or should be, the second step after you determine ways that the process might fail.

Let’s approach this a different way. The one of the links between the control plan and the PFMEA is the potential cause column, correct? This would be input in the control plan as a critical process characteristic? If so this would mean that in my case we would want the control plan help us to monitor things like solution temperature, rinse water cleanliness, etc… to get a clean part in the cleaning cycle, rinse the cleaner off in the rinse cycle, etc…

Using this, this would mean that the following would be the case for my application:

Requirement - Failure Mode - Failure Effect - Potential Cause

No soil contaminants on part - Soil contaminants on part - Paint won’t stick; strip & repaint - alkaline cleaner too hot

This would put Alkaline Cleaner temp on the control plan as a process characteristic to be set to 160F, measured real time with hard mounted temp probe connected to the machine monitor, control method with auto-adjusting PLC system.

If the following would be the case:

Requirement - Failure Mode - Failure Effect - Potential Cause

Alkaline cleaner 160F - Alkaline cleaner too hot - Paint won’t stick; strip & repaint - Temp probe misreading temp

This would put Temperature probe calibration on the control plan, process specification being calibrated Per XX spec, frequency Every 6 months, control method being (for example) auto shutoff of machine if calibration is overdue

To me the first example is a PFMEA and the second is a DFMEA on a painting machine.
 

Jim Wynne

Leader
Admin
First, part of the purpose of the PFMEA process is to determine what goes on the control plan. Normally, not everything that's cited in the PFMEA will get to the PCP. It's a matter of risk evaluation.

Next, if I were doing it using your example, it would look like this:

  • Process Requirement: No soil contamination
  • Potential Process Failure Mode: Alkaline cleaner too hot
  • Potential effect: paint adhesion failure
  • Potential Cause: Temp probe malfunctions
  • Current Process Controls (Detection and Prevention)
  • Recommended Actions (if any)
Calibration of the probe should be handled separately; if the probes need to be periodically calibrated but they aren't, that's an item for the Recommended Actions columns.
 

tugboat

Registered
First, part of the purpose of the PFMEA process is to determine what goes on the control plan. Normally, not everything that's cited in the PFMEA will get to the PCP. It's a matter of risk evaluation.

Next, if I were doing it using your example, it would look like this:

  • Process Requirement: No soil contamination
  • Potential Process Failure Mode: Alkaline cleaner too hot
  • Potential effect: paint adhesion failure
  • Potential Cause: Temp probe malfunctions
  • Current Process Controls (Detection and Prevention)
  • Recommended Actions (if any)
Calibration of the probe should be handled separately; if the probes need to be periodically calibrated but they aren't, that's an item for the Recommended Actions columns.

I should've disclaimed the "I know not everything goes on the control plan", I was just throwing out an example.

But then, assuming this is high risk what goes on the control plan? Is your control plan controlling for the potential cause or the process failure mode? Ultimately the control plan should be controlling the inputs to the process, which in this case would include cleaning solution temperature. If you pretend we need to measure temperature manually then this would need to go on the control plan, measuring X frequency with ABC temperature probe.

If I'm following you then this means the control plan would be controlling the failure mode?

And would the Product characteristics come from the DFMEA? In this case, as specified by design, things such as dry film thickness, full coverage, adhesion properties, etc...
 
Last edited:

Jim Wynne

Leader
Admin
But then, assuming this is high risk what goes on the control plan? Is your control plan controlling for the potential cause or the process failure mode?
Your control plan should show the control(s) for detection/prevention of the cause.
Ultimately the control plan should be controlling the inputs to the process...
The control plan should show controls for process steps that need control based on output from the PFMEA.
...If you pretend we need to measure temperature manually then this would need to go on the control plan, measuring X frequency with ABC temperature probe.
Makes sense to me.
If I'm following you then this means the control plan would be controlling the failure mode?
The control plan should show methods for controlling the cause(s). The idea is prevention.

And would the Product characteristics come from the DFMEA? In this case, as specified by design, things such as dry film thickness, full coverage, adhesion properties, etc...
A DFMEA, when available, should be used to inform the PFMEA.
 

tugboat

Registered
Here's a somewhat alternate viewpoint in my brain which may actually end up jiving with what's been said here.

The process outputs (the opposite of which are failure effects along with what impact said failure effect has on production, eg. "Paint doesn't stick, strip and repaint") are things like adhesion properties, dry film thickness, hardness, drips/runs, etc... since that's what the customer cares about.

The process inputs (the opposite or "bad" state of which are failure modes) are soil contamination level, rust contamination level, part surface dryness, paint viscosity, etc...

The process paramaters, which if out of spec would be potential causes (which would then be on the control plan as Process Characteristics) would be things like alkaline cleaner concentration, alkaline cleaner temperature, drying tank temperature, drying time, paint temperature, paint:water ratio, etc...

If you control for things like alkaline cleaner concentration, drying tank temperature, and paint:water ratio then that's what you want as if these are all in spec then you'll achieve your process outputs.

Let me know if I'm on the right track or if this is crazy talk.
 

Jim Wynne

Leader
Admin
Here's a somewhat alternate viewpoint in my brain which may actually end up jiving with what's been said here.

The process outputs (the opposite of which are failure effects along with what impact said failure effect has on production, eg. "Paint doesn't stick, strip and repaint") are things like adhesion properties, dry film thickness, hardness, drips/runs, etc... since that's what the customer cares about.

The process inputs (the opposite or "bad" state of which are failure modes) are soil contamination level, rust contamination level, part surface dryness, paint viscosity, etc...

The process paramaters, which if out of spec would be potential causes (which would then be on the control plan as Process Characteristics) would be things like alkaline cleaner concentration, alkaline cleaner temperature, drying tank temperature, drying time, paint temperature, paint:water ratio, etc...

If you control for things like alkaline cleaner concentration, drying tank temperature, and paint:water ratio then that's what you want as if these are all in spec then you'll achieve your process outputs.

Let me know if I'm on the right track or if this is crazy talk.
I think you're overthinking the whole thing.
  • For each step in a process, determine as best you can how the process might fail.
  • Consider the potential consequences (effects) of the failure(s) and relative severity.
  • Consider the likelihood of failure.
  • Mitigate the risk(s) based on severity and likelihood.
  • For those potential failures where process controls are deemed necessary or prudent, transfer them to the control plan.
 

Bev D

Heretical Statistician
Leader
Super Moderator
In addition to Jim’s wisdom, I would add that teh critical downfall of many ProcessFMEAs that I’ve seen is that the function that is listed is often related to the part and not the process. This creates a lot of confusion.
For example adhesion properties is an output of several process functions or steps. in painting a cleaning step is often used to 1. Remove contaminants and 2. Prepare the surface for better adhesion (e.g. creating a specific surface roughness). These are functions of the cleaning step. A failure mode is that the cleaning process fails to remove all contaminants. The effect is that paint adhesion may fail to meet specifications. A cause of the failure is that the cleaning solution is too dilute (or did not last long enough, the rinse phase was too weak or to short…). A mitigation might be that we test the output (paint adhesion). Another - better - control is that we use error proofing to ensure settings for duration, test for potency, etc. Perhaps we move from a manual cleaning solution mixing process to an automated one.
 

tugboat

Registered
In addition to Jim’s wisdom, I would add that teh critical downfall of many ProcessFMEAs that I’ve seen is that the function that is listed is often related to the part and not the process. This creates a lot of confusion.
For example adhesion properties is an output of several process functions or steps. in painting a cleaning step is often used to 1. Remove contaminants and 2. Prepare the surface for better adhesion (e.g. creating a specific surface roughness). These are functions of the cleaning step. A failure mode is that the cleaning process fails to remove all contaminants. The effect is that paint adhesion may fail to meet specifications. A cause of the failure is that the cleaning solution is too dilute (or did not last long enough, the rinse phase was too weak or to short…). A mitigation might be that we test the output (paint adhesion). Another - better - control is that we use error proofing to ensure settings for duration, test for potency, etc. Perhaps we move from a manual cleaning solution mixing process to an automated one.

Very helpful, thank you Bev! That’s my thing, I want to make sure this is an FMEA on the process and a DFMEA on the equipment or the part but want to be as crystal clear as possible on the distinction between the three to effectively create an FMEA on the process.
 
Top Bottom