I have two doubts about PFMEA failure modes - How should I describe a "failure mode"

A

astudent

A couple of things are not clear to me about "failure modes" in a PFMEA.
1) Do they rapresent what could not work in a process or what could not be good with the product?
In other words: how should I describe a "failure mode" in the FMEA form: referring to the process or to the product?
A N.C. of the process could be "too hight temperature", a N.C. of the product "burned item" and the temperature could be te cause.
I need a little help as I am confused.
2) Maybe this question is linked with the first one.
Talking about Incoming inspection, let'us imagine we have sampling plans, because the incoming materials can be defective.
What can be a failure mode, cause and effect in this case?

thank you.
 
B

Bill Ryan - 2007

There are different ways to tackle a PFMEA. Our "preferred" definitions are as follows:
Failure Mode = Measureable on the product
Failure Cause = Measureable in the process
Sooo....
"A N.C. of the process could be "too hight temperature", a N.C. of the product "burned item" and the temperature could be te cause." I would treat as "Burnt" is the Failure Mode and "Temperature setting - set too high" as the Failure Cause.


Regarding Incoming Inspection - You could have Failure Modes such as "Accept nonconforming product" or "Reject conforming product" with Failure Causes such as "Sampling Plan ineffective"; "Gage out of calibration"; "Visual Standards unclear"; etc. We also happen to list the product features we wish to inspect as Failure Modes (the biggest reason being that our software forces us to have product characteristics listed as Modes to get them showing up in the correct place on the Control Plan).

As I mentioned, this is how we tackle PFMEAs (that are submitted with PPAPs) at our company - there are others that list failures of the process as Failure Modes. My "issue" with that methodology is that you can get into a "circle" where a Failure Cause turns into a Failure Mode or vice versa.

Does that help any?
 

Jim Wynne

Leader
Admin
Bill Ryan said:
As I mentioned, this is how we tackle PFMEAs (that are submitted with PPAPs) at our company - there are others that list failures of the process as Failure Modes. My "issue" with that methodology is that you can get into a "circle" where a Failure Cause turns into a Failure Mode or vice versa.
While I think the output is the most important thing, I favor using process failures as the modes. The reason for this is simple: if it's prevention your'e after, finding the cause of a process failure is usually more direct than looking for defects, then looking for process failures.
 
A

astudent

Thank you, you have been very clear.
1) These two are the only to ways to tackle a PFMEA or there are others ones?
(just a doubt because you wrote "there are different ways")
I want to understand better the cause-effect chain.

MODE on the PRODUCT:
"Temperature setting - set too high"(Failure Cause) --> "Burnt item"(Failure Mode) --> "mechanical characteristics not appropriate"(Effect)

MODE on the PROCESS:

"bad trained personnel"(Failure Cause) --> "Temperature setting - set too high"(Failure Mode) --> "Burnt item"(Effect)

Where I'm mistaken (if I am)?

2) So, it seems that incoming inspection is treated differently from inspection on the following steps of the process, which are part of the step itself.
It is right?
 

Jim Wynne

Leader
Admin
astudent said:
Thank you, you have been very clear.
1) These two are the only to ways to tackle a PFMEA or there are others ones?
(just a doubt because you wrote "there are different ways")
I want to understand better the cause-effect chain.

MODE on the PRODUCT:
"Temperature setting - set too high"(Failure Cause) --> "Burnt item"(Failure Mode) --> "mechanical characteristics not appropriate"(Effect)

MODE on the PROCESS:

"bad trained personnel"(Failure Cause) --> "Temperature setting - set too high"(Failure Mode) --> "Burnt item"(Effect)

Where I'm mistaken (if I am)?
Not speaking for Bill, but I think when he said "different ways" he was referring to the two ways you illustrate above. It's a matter of preference.

astudent said:
2) So, it seems that incoming inspection is treated differently from inspection on the following steps of the process, which are part of the step itself.
It is right?
If inspection is a discrete step in the flow, it should be treated as a separate operation. In this case, it's not different from incoming inspection, because the assumption must still be made that the condition of the material is unknown (otherwise there'd be no need for inspection smile-a1.gif ).
 
A

astudent

I'm sorry but I can't understand.
It could seem like each inspection of the column for inspections should become a new step.
In every case the condition of the material is unknown.
What are the differences between the two kind of controls?

as to point 1) do you think that my examples are correct?

Thank you because your help in this forum is very important.
 

Jim Wynne

Leader
Admin
astudent said:
I'm sorry but I can't understand.
It could seem like each inspection of the column for inspections should become a new step.
In every case the condition of the material is unknown.
What are the differences between the two kind of controls?

as to point 1) do you think that my examples are correct?

I think your examples are OK; Bill Ryan and I disagree as to how modes should be identified, but the disagreement is minor and either way will be OK so long as the PFMEA process is a conscientious effort at minmizing risk. You should choose one method or the other and stick with it.

There is no "column for inspections" on the standard PFMEA form. There are two columns for "Current Process Controls," one for prevention-type controls and one for detection (inspection) controls. 'Prevention" controls are those that will help to prevent defects from occurring, such as mistake-proofed machine controls, operator training, work instructions and visual aids. "Detection" controls are those that are intended to keep defective product from being shipped, or having value added to it. Inspection should be considered a separate operation on the flow diagram when it is a separate step; that is, when a manufacturing operation takes place and the next manufacturing step doesn't take place until the product has been inspected. If parts are inspected in-process as production is running, there's generally no need to consider it as a separate operation. In that case, material moving from operation "A" is assumed to be "good" upon arrival in operation "B" for purposes of the PFMEA.
 
Last edited:
M

mtncrawler

Jim Wynne said:
While I think the output is the most important thing, I favor using process failures as the modes. The reason for this is simple: if it's prevention your'e after, finding the cause of a process failure is usually more direct than looking for defects, then looking for process failures.

Interesting perspective. I am with an organization now that teaches and practices this method. My past experience (right, wrong or indifferent) has taught me the opposite. Coming from a mainly maching background, I would typically list the process steps (directly from the process flow) and then start looking at critical features being produced at that step (usually from KCC, CTQ's, VOC or the like) and brainstorm the PFM, PEF, for each.

For example;
Process Function - Machine Part
Process Step - Turn .250 +/-.001 O.D.
PFM - Diameter O/S (Diameter U/S)
PE of F - Will not fit at assy, customer dissatisfaction..
PC of F - Improper tool setting, tool breakage,operator training...

The reason I've always favored this approach is that it speaks directly to the key characteristic involved. If I focused on the key input, i.e. tool setting, I know you would be addressing prevention in a more generic methodology for all "machined features", but to me, you loose the focus on specific analysis regarding the critical features.

In my experience in the APQP, this made it real easy to transfer the flow of information into the Control Plan (and vice-versa) as well.

In the end you're still looking at prevention. If you have a high OCC, you will address it.

Again, I don't think either method is wrong. I'm still having to overcome (to engrained!), or at least understand, both ways of thinking.
 

Jim Wynne

Leader
Admin
mtncrawler said:
Process Function - Machine Part
Process Step - Turn .250 +/-.001 O.D.
PFM - Diameter O/S (Diameter U/S)
PE of F - Will not fit at assy, customer dissatisfaction..
PC of F - Improper tool setting, tool breakage,operator training...

I would do it this way:

Potential Failure Modes: Improper tool setting
Tool breakage
(I wouldn't characterize operator training--or lack thereof--as a failure of the machining process)

Effect: (Might differ with the different failure modes)
Diameter O/S
Diameter U/S

Cause(s): Here you might cite training, as well as whatever else might
cause tools to be set improperly, or to break or get dull, or...

The idea being, as I suggested earlier, that if I'm really going to do process control, I want to know how the process might fail. I already know that if it does fail, something bad will happen, or it wouldn't be called "failure."

Your point regarding special characteristics, which customers normally expect to see individually addressed, is well taken, but it doesn't mean that they can't be addressed using the process failure approach.
Also, some customers will expect to see FMEA documents that are done according to The Book, and customer requirements should be honored.
 
Top Bottom