FMEA Starting point

[Candi1024]

I am working on a process FMEA. Typically I would use the process specs/outputs as the starting point, then list all possible causes of failure.

However many of the FMEA forms start with process inputs and then work into what the effects would be if those inputs failed.

Is my method not robust? I am working on a retrospective PQ and personally I am only worried about the process outputs and risk mitigation for those outputs.

BTW, I'm really not asking for my own purposes, more just for conversation. I'm happy with my method for what I am doing. However since we won't be able to discuss this in the future, I thought I would ask now.

 

[Bev D]

*I* always think of it as a flow: inputs -> function -> outputs
the function is what fails: in 4 possible ways, complete, partial, intermittent and unexpected.
the inputs to the function are the possible causes
the outputs are the start of the effect chain...

 

[Krocpok]

I think you do it well.
General assumption of FMEA is that inputs of individual process steps (apart from your first process step, for ex. material delivery) are always correct.

 

[Bev D]

I think you do it well.
General assumption of FMEA is that inputs of individual process steps (apart from your first process step, for ex. material delivery) are always correct.

but if the inputs are always correct what can go wrong?

 

[Jim Wynne]

but if the inputs are always correct what can go wrong?

Krocpok is correct. The idea is for the FMEA team to concentrate on what might go wrong in the current step or process. Upstream processes have presumably been addressed, and it's redundant to go over them again.

 

[John Broomfield]

I am working on a process FMEA. Typically I would use the process specs/outputs as the starting point, then list all possible causes of failure.

However many of the FMEA forms start with process inputs and then work into what the effects would be if those inputs failed.

Is my method not robust? I am working on a retrospective PQ and personally I am only worried about the process outputs and risk mitigation for those outputs.

BTW, I'm really not asking for my own purposes, more just for conversation. I'm happy with my method for what I am doing. However since we won't be able to discuss this in the future, I thought I would ask now.

Candi,

My starting point: make sure the tables for prioritizing risk are approved by an authorized person.

John

 

[Bev D]

Krocpok is correct. The idea is for the FMEA team to concentrate on what might go wrong in the current step or process. Upstream processes have presumably been addressed, and it's redundant to go over them again.

ah - so it's semantics about what an input is...?
let's say we are dealing with a sealing process. the material thickness, the adhesive layer etc. are all material inputs to be sealed; they are of the correct thickness, chemical composition, etc. Now the sealing process requires parameter settings of time, temperature, pressure, feed rate etc. are these not input parameters of the sealing process? or do some people regard these as the process?

 

[Jim Wynne]

ah - so it's semantics about what an input is...?
let's say we are dealing with a sealing process. the material thickness, the adhesive layer etc. are all material inputs to be sealed; they are of the correct thickness, chemical composition, etc. Now the sealing process requires parameter settings of time, temperature, pressure, feed rate etc. are these not input parameters of the sealing process? or do some people regard these as the process?

Let me rephrase: The idea is to determine defects that are caused in the current operation or process step. The idea is to isolate causes and not go over things that have been addressed upstream. The automotive PFMEA is based on the discrete steps of the process flow diagram. In your example, the process parameters are not considered inputs as such--they are components of the process and should be addressed as possible sources of failure.

All is general and shouldn't necessarily be considered hard-and-fast rules.