Failure modes for Inspection processes

[somashekar]

Risk management is across the medical device life cycle.

 

[Bev D]

@Tidge So I believe in this case what I getting is that there are two groups one promoting to add inspection as a process with their respective failure modes and also as the control for upstream manufacturing process. The other group feel that inspection is only a control and should not be added in PFMEA as a separate process.

The graft we have has 100% visual insepction to detect the defects. Inspection is perfomed with a an appartus that had to be installed during the duration of inspection (handling comes during this point too).

Let me describe a scenario:

Holes or any damage to the graft can occur throughout the manufacturing line at any stage and inspection process at the end of the line is designed to detect those defects. If itself the inspection process is damaging the product than what is the purpose of having an inspection. But still handling is a factor during inspection too, how will one eliminate that factor?

We have the validation report for that inspection process (test method).

In your case inspection is both a control (which must be evaluated for it’s effectiveness at Detection) and a process that can potentially damage the “product” (and/or patient?). The potential for damage must be assessed like any other failure mode or hazard. It doesn’t get a pass just because it’s a control. There are many inspections and tests that have a potential for damaging the product. The purpose of the FMEA is to evaluate and improve the process. So if your inspection can damage the product you need to either improve the process so that damage doesn’t occur OR select a different inspection method OR improve the upstream process such that inspection is not needed as a control.

Science and Data trump opinion. Science is not a democracy…

 

[Quality Engineer]

In your case inspection is both a control (which must be evaluated for it’s effectiveness at Detection) and a process that can potentially damage the “product” (and/or patient?). The potential for damage must be assessed like any other failure mode or hazard. It doesn’t get a pass just because it’s a control. There are many inspections and tests that have a potential for damaging the product. The purpose of the FMEA is to evaluate and improve the process. So if your inspection can damage the product you need to either improve the process so that damage doesn’t occur OR select a different inspection method OR improve the upstream process such that inspection is not needed as a control.

Science and Data trump opinion. Science is not a democracy…

But then what will be the controls of that inspection process?

 

[Jeroki]

But then what will be the controls of that inspection process?

As Bev indicates, due to the impact it has in the product, it should be considered a additional process, and in your FMEA, taking into account repetibility, severity and detection define actions to improve the risk of such process (whether this is implementing a second quality inspection, modifying the design, modifying the tool to be used, etc.)

 

[Miner]

Can I get any example?

In our inspection process we check for holes, detachment of subassembly etc.

If we write "Hole not detected during inspection", is this the correct failure mode?

which eventually means that we accepted a part which had a hole in it.

You do not necessarily need to list every possible detection failure, but list the most serious, most frequently occurring, etc. Your controls could be things like calibration, gage R&R, attribute appraisal, etc. for accept bad/reject good. The other failure modes I listed would have different controls.

 

[Bev D]

1. The function of inspection is to detect defects (nonconforming dimensions, features, properties or functions). These characteristics are identified in the FMEA at the process or product functional step where they are created. The only failure mode of an inspection therefore is to either fail to detect a defect or to falsely identify a defect where none exists. My experience, belief and practice is that since the inspection is placed as a control to an upstream function there is no need to re-state the defect in the FMEA form where the inspection is documented.

2. Detection controls: Write this down. It is a (mostly?) exhaustive list for any detection control.
First perform a MSA. Not just a Gauge R&R.
Begin with an assessment of the detection’s ability to ‘measure’ the characteristic of interest. This can typically be done through logic and engineering knowledge (not engineering judgment which is just fancy talk for an engineer’s opinion). For example if the characteristic is weight, are you measuring weight or volume? Is the viscosity/density of the thing consistent enough that volume is a substitute for weight? If the method is not suitable, find another one.
Next perform the MSA for repeatability, reproducibility between operators, equipment, fixtures, etc. determine the false positive and false negative results. <- this is the occurrence rate of the detection failure modes. If the failure rate is too high improve or replace your method. Remember that the intent of FMEA is to test not guess.
Next assess the sample size if not 100%. Can it reliably detect a defect rate that is too high? And remember this is the RQL NOT the AQL.
Controls for detection methods: Does the method require that appraisers be certified or qualified? Some tests do require this by regulations. Is the defect severe enough and the test effected by appraiser method? Strongly consider adding your own qualification test (calibrating your appraisers like you calibrate equipment). Next establish a calibration protocol and cycle for your equipment. Additionally if the defect is severe enough and the method potentially variable enough consider using a known defective part to inject into the inspection stream to ensure that the method can still detect it. (Usually this is a ‘just defective’ part as it poses the greatest challenge to the system).

3. Any new failure modes that can be introduced by the inspection process itself (like handling, aggressive sampling, etc.) should be handled as any other process induced failure mode…

 

[Quality Engineer]

1. The function of inspection is to detect defects (nonconforming dimensions, features, properties or functions). These characteristics are identified in the FMEA at the process or product functional step where they are created. The only failure mode of an inspection therefore is to either fail to detect a defect or to falsely identify a defect where none exists. My experience, belief and practice is that since the inspection is placed as a control to an upstream function there is no need to re-state the defect in the FMEA form where the inspection is documented.

2. Detection controls: Write this down. It is a (mostly?) exhaustive list for any detection control.
First perform a MSA. Not just a Gauge R&R.
Begin with an assessment of the detection’s ability to ‘measure’ the characteristic of interest. This can typically be done through logic and engineering knowledge (not engineering judgment which is just fancy talk for an engineer’s opinion). For example if the characteristic is weight, are you measuring weight or volume? Is the viscosity/density of the thing consistent enough that volume is a substitute for weight? If the method is not suitable, find another one.
Next perform the MSA for repeatability, reproducibility between operators, equipment, fixtures, etc. determine the false positive and false negative results. <- this is the occurrence rate of the detection failure modes. If the failure rate is too high improve or replace your method. Remember that the intent of FMEA is to test not guess.
Next assess the sample size if not 100%. Can it reliably detect a defect rate that is too high? And remember this is the RQL NOT the AQL.
Controls for detection methods: Does the method require that appraisers be certified or qualified? Some tests do require this by regulations. Is the defect severe enough and the test effected by appraiser method? Strongly consider adding your own qualification test (calibrating your appraisers like you calibrate equipment). Next establish a calibration protocol and cycle for your equipment. Additionally if the defect is severe enough and the method potentially variable enough consider using a known defective part to inject into the inspection stream to ensure that the method can still detect it. (Usually this is a ‘just defective’ part as it poses the greatest challenge to the system).

3. Any new failure modes that can be introduced by the inspection process itself (like handling, aggressive sampling, etc.) should be handled as any other process induced failure mode…

@Bev D
I am confuse in your first point. Do you mean to say that the failure mode of defect should be listed where it is created and not under the inspection process where it is detected because the inspection is listed as the control of that defect?

Let's say you are manufacturing a product that has a critical feature of a hole being drilled in the correct location. The PFMEA for this product identifies the potential defect of the hole not being drilled in the correct location. The upstream function that creates this feature is the drilling process.
The inspection for this defect would be to visually inspect the product and ensure that the hole is in the correct location. The only failure modes of this inspection are to either fail to detect a defect, meaning the hole is not in the correct location but the inspection does not find it, or to falsely identify a defect, meaning the hole is in the correct location but the inspection incorrectly identifies it as not being in the correct location.

Here the failure mode "Hole drilled on incorrect location" which is the failure mode of the upstream process and the failure mode "Incorrect locaton of hole not detected or Part accepted with incorrect hole location" which is the failure mode of inspection process, how the failure modes are same? The aspect of both the failure mode looks different.

I just want to clear my concept of this, this is too confusing.

 

[Bev D]

@Bev D
I am confuse in your first point. Do you mean to say that the failure mode of defect should be listed where it is created and not under the inspection process where it is detected because the inspection is listed as the control of that defect?

Let's say you are manufacturing a product that has a critical feature of a hole being drilled in the correct location. The PFMEA for this product identifies the potential defect of the hole not being drilled in the correct location. The upstream function that creates this feature is the drilling process.
The inspection for this defect would be to visually inspect the product and ensure that the hole is in the correct location. The only failure modes of this inspection are to either fail to detect a defect, meaning the hole is not in the correct location but the inspection does not find it, or to falsely identify a defect, meaning the hole is in the correct location but the inspection incorrectly identifies it as not being in the correct location.

Here the failure mode "Hole drilled on incorrect location" which is the failure mode of the upstream process and the failure mode "Incorrect locaton of hole not detected or Part accepted with incorrect hole location" which is the failure mode of inspection process, how the failure modes are same? The aspect of both the failure mode looks different.

I just want to clear my concept of this, this is too confusing.

Yes that is exactly what I mean, it is exactly what I said. Defects are listed for the functions that CREATED them; not for teh steps where they are detected (or missed). And its not just me - this is the teaching for decades.

I am not clear what you are saying here: you seem to be contradicting yourself. Creating the defect and detecting/missing the defect are completely different failure modes, they aren’t the same at all…I’m really not sure what you are confused about unless someone else is (incorrectly) telling you that the defect is listed at teh inspection step where it is detected?

Here the failure mode "Hole drilled on incorrect location" which is the failure mode of the upstream process and the failure mode "Incorrect locaton of hole not detected or Part accepted with incorrect hole location" which is the failure mode of inspection process, how the failure modes are same? The aspect of both the failure mode looks different.

 

[Tidge]

There should not be confusion about what @Bev D wrote: There can be many outputs of a process (e.g.a drilled hole) that don't need to be inspected because the process was validated and has controls which don't require additional inspections for risk reduction.

 

[Quality Engineer]

Yes that is exactly what I mean, it is exactly what I said. Defects are listed for the functions that CREATED them; not for teh steps where they are detected (or missed). And its not just me - this is the teaching for decades.

I am not clear what you are saying here: you seem to be contradicting yourself. Creating the defect and detecting/missing the defect are completely different failure modes, they aren’t the same at all…I’m really not sure what you are confused about unless someone else is (incorrectly) telling you that the defect is listed at teh inspection step where it is detected?

@Bev D
Let see if I am getting something now or not

1. "Hole drilled at incorrect location" will addressed under the Process where operation is performed.
2. "Failed to detect incorrect location of holde" will be addressed under the Inspection process.

I believe the failure mode is related to deviation in performing process!

Is this correct?

----------------------------------------------------------------------------------------------------------------------------------------------

Also, In our product mostly the failing defects are handling related, listed belolw:

1. Split/tear
2. Holes
3. Excesisve densification
4. Excessive stretching
5. Wrinkles
6. Contamination
7. Delamination

Others are related to deviation in performing process
8. Incomplete weld
9. Incorrect trimming etc.

Failure mode: "Material damaged - split/tear" will be addressed under the Processes where operations are performed on the products.

Handling defects can occur throughout the manufacturing process under different operations. Then an inspection is performed in the end to "Visually, dimensionally and functionally" inspect all above listed defects.

But then as said earlier that Handling failure mode should be addressed in the similar way they are addressed in other processes and I see that Split/tear, holes, densification, stretching and contamination can be introduced during the handling at inspection. So this means:
one failure mode under inspection is "Split/tear not detected" and then the second failure mode is "Material damaged - Split/tear"

Defects are listed for the functions that CREATED them; not for teh steps where they are detected (or missed). And its not just me - this is the teaching for decades.

Then I see your this point too.