PFMEA for test procedures (ISO 14971)

[Lauren1234]

I'm performing a pFMEA of my manufacturing processes (ISO 14971:2019 compliant) and I'm wondering if I should include an analysis of test methods within my pFMEA. We have a number of analytical and functional tests, none of which are considered a "change to the product". These test methods are referenced as risk control measures for several failure modes identified. I understand that doing a failure modes analysis on these test procedures can potentially put us in a loop, but if the tests are complex isn't it worth performing the analysis to identify whether further risk controls should be implemented within the test procedure, or would a test method validation be sufficient and not warrant a need to perform this sort of risk analysis? Unfortunately, some of our tests do not have a TMV [Test Method Validation] at this time - is it recommended to handle these differently, or can I get away with listing them high-level in my pFMEA as the risk being accepting bad part or rejecting good part with essentially little to no risk control in the absence of a TMV?

 

[Miner]

Inspections and tests have two basic failure modes:

• Accepting non-conforming product
• Rejecting conforming product

However, the effects of these failures will differ by the particular product characteristic or process parameter. Do not attempt to perform the FMEA at the detailed, step-by-step test method level, but at the level of these two failure modes.

 

[Brakeman]

Inspections and tests have two basic failure modes:

• Accepting non-conforming product
• Rejecting conforming product

However, the effects of these failures will differ by the particular product characteristic or process parameter. Do not attempt to perform the FMEA at the detailed, step-by-step test method level, but at the level of these two failure modes.

While that is the common method, if this is a more exacting or multi risk test, I would not recommend your approach. Just because the failure effects are binary doesn't lead that you shouldn't individualize the risks of getting to that point. In automotive, I've seen several supplier PFMEA's just list the risk effect as "customer dissatisfaction" which is so lazy and unhelpful. If you've not going to try and get valued information from the PFMEA, why bother doing one in the first place?

If the test provides type 1 failure because of mis-steps or lack of training, that is a different risk than if the test fails because of poorly maintained equipment or of contaminated chemicals.

I think it boils down to whether there is value in breaking down and analyzing the test.

 

[Jim Wynne]

While that is the common method, if this is a more exacting or multi risk test, I would not recommend your approach. Just because the failure effects are binary doesn't lead that you shouldn't individualize the risks of getting to that point. In automotive, I've seen several supplier PFMEA's just list the risk effect as "customer dissatisfaction" which is so lazy and unhelpful. If you've not going to try and get valued information from the PFMEA, why bother doing one in the first place?

If the test provides type 1 failure because of mis-steps or lack of training, that is a different risk than if the test fails because of poorly maintained equipment or of contaminated chemicals.

I think it boils down to whether there is value in breaking down and analyzing the test.

I agree with Miner's post, although I would add an additional failure mode, which would be failure to properly identify test result status after completion. The more granular aspects can be captured in the causes of failure.

 

[Miner]

Correct Jim. Brakeman, I never said that you couldn't have many causes of the failure modes. That should be done thoroughly and will be unique to the measurement device.

 

[Tidge]

Historically, I have included test methods in PFMEA for a couple of reasons:

• Often, a validated test method will be a risk control for one (or more) potential failure modes
• Some test methods (which are not intended to be destructive) have the potential to introduce defects (e.g. because of 'rough handling')

My main piece of advice is to not confuse a PFMEA and a TMV (test method validation). The PFMEA can motivate a TMV, and can list TMVs as VE of risk controls. Keep clear in your mind that these are different beasts and you will avoid a potential point of being tripped up.