For example, I have a microscope inspection process at the IQC stage, to find defect color change,

The color change is caused by over-time baking at the supplier side. The supplier happens to be our customer. If the inspector happens to be colorblind, he could also fail to find the color change. The color change is an indicator of weak shear, and can cause the part fail to work.

Now, how should I fill in the columns of a PFMEA for this microscope inspection process?

Method 1:
Process Function/Requirements: microscope inspection
Potential Failure Model: Color change
Potential effect(s) of Failure: parts fail to work
Potential Cause (s)/mechanism(s) of Failure: over-time baking
Current Process Control: Microscope inspection

Method 2:
Process Function/Requirements: microscope inspection
Potential Failure Model: Color changed but the inspector failed to find it.
Potential effect(s) of Failure: parts fail to work
Potential Cause (s)/mechanism(s) of Failure: colorblindness
Current Process Control: No

Method 3:
This opinion suggests we should not do a PFMEA for pure measuring process at all, so all the above should not exist.

Which do you think is better, or do you have a better idea?

For example, I have a microscope inspection process at the IQC stage, to find defect color change,

The color change is caused by over-time baking at the supplier side. The supplier happens to be our customer. If the inspector happens to be colorblind, he could also fail to find the color change. The color change is an indicator of weak shear, and can cause the part fail to work.

Now, how should I fill in the columns of a PFMEA for this microscope inspection process?

Method 1:
Process Function/Requirements: microscope inspection
Potential Failure Model: Color change
Potential effect(s) of Failure: parts fail to work
Potential Cause (s)/mechanism(s) of Failure: over-time baking
Current Process Control: Microscope inspection

Method 2:
Process Function/Requirements: microscope inspection
Potential Failure Model: Color changed but the inspector failed to find it.
Potential effect(s) of Failure: parts fail to work
Potential Cause (s)/mechanism(s) of Failure: colorblindness
Current Process Control: No

Method 3:
This opinion suggests we should not do a PFMEA for pure measuring process at all, so all the above should not exist.

Which do you think is better, or do you have a better idea?

#2 is best; you shouldn't do PFMEA for processes you have no control over, as in method #1.

I agree with Jim. Along the same lines - could you also have the Failure Mode of "Color didn't change but operator rejected"? Is color blindness the only Cause that could allow passing or rejection of product in your inspection process?

I'm not 100% sure as to what your process is but if the process is simply pure inspection of the part or item. Then I'd suggest you only have two failure modes along the lines of 1. reject good part or 2. pass bad part. Some companies don't do FMEA on inspection process.

I agree with Jim. Along the same lines - could you also have the Failure Mode of "Color didn't change but operator rejected"? Is color blindness the only Cause that could allow passing or rejection of product in your inspection process?

Yes. We are just starting using PFMEA. Many people in the company are unfamiliar with this tool. Besides, we have many pure measuring processes spotted among the manufacturing processes, so we can’t just follow the method the FMEA manual introduces and ignore them. Currently, blindness is the only cause we have identified. Thank you for your suggestion. I forgot to consider the other side "Color didn't change but operator rejected".

Hello! :)
I wanted to join this discussion because I posted a similar question (http://elsmar.com/Forums/showthread.php?t=20166) recently. I have little training on FMEA so far, but this is my opinion.

My disagreement with Method #1 would be that the "process control" is the same as the "process function". Instead it is my understanding that the failure of color change should be under the baking process, not the inspection process. I have been told before to only include failure modes that you can control at the process you are currently evaluating.

My disagreement with Method #2 is that I've been told that when you are thinking about failure modes that can occur at a certain process, you should assume that all process before operated correctly. So if the prior baking processes operated correctly, there would be no defects to miss. Have I been completely mislead in this? :frust:


I agree most with Method #3 (that inspections should not be a process on the FMEA, but shoud be a control on the FMEA. But so far, everyone that responded to my thread has told me that inspections should be included.

Welcome to this thread. I myself am not very good at PFMEA, so I welcome others join in this discussion. My opinions are as follows:

Hello! :)
I wanted to join this discussion because I posted a similar question (http://elsmar.com/Forums/showthread.php?t=20166) recently. I have little training on FMEA so far, but this is my opinion.

My disagreement with Method #1 would be that the "process control" is the same as the "process function". Instead it is my understanding that the failure of color change should be under the baking process, not the inspection process. I have been told before to only include failure modes that you can control at the process you are currently evaluating.
.
Yes. You are right. But the baking process is at our customer’s side, and we can’t control our customer’s process. So the process of concern is the inspection process, whether or not it can identify the defects caused by our customer.


My disagreement with Method #2 is that I've been told that when you are thinking about failure modes that can occur at a certain process, you should assume that all process before operated correctly. So if the prior baking processes operated correctly, there would be no defects to miss. Have I been completely mislead in this? :frust:
.
I think the assumption that all process before operated correctly applies to manufacturing process, but not inspection process.

For the pure manufacturing processes, whether or not the processes before operated well can’t change you site control a bit, so you’d reduce the complexity of process analysis if you assume they are all OK. But for the inspection processes, it’s better to accept the previous process performance as it is, because it decides the severity of inspection failure.

For example, a manufacturing process, when all elements are under control, produces product with a PPM equaling 100. However, with a minor negligence, the PPM could jump to 1000, and this could happen. If the severity of PPM=100 is 1, the severity of PPM=1000 could be 3. So if you assume all the processes before operated correctly, you might conclude that there is no need for inspection. The design of inspection is precisely for the PPM=1000 case, when minor negligence exists. So if the inspection process exists, and we want to do a PFMEA for it, we had better to assume the manufacturing process before probably did not operate well.

The minor negligence factor is based on a valid knowledge on the process. We know it could happen. We would not unreasonably enlarge the scope, to include other factors.

I agree most with Method #3 (that inspections should not be a process on the FMEA, but shoud be a control on the FMEA. But so far, everyone that responded to my thread has told me that inspections should be included.
The full name of PFMEA is Potential Failure Mode and Effects Analysis in Manufacturing and Assembly Processes. So it is right that inspection should not be included into the PFMEA in this sense. However we need to go beyond this definition, like we do when we evaluate the environment aspects.

I think whether or not an inspection process should be included in the PFMEA document depends on the importance of the inspection itself. That is, we do a PFMEA analysis for every inspection process in our heart.

If we conclude that the process would not possibly go wrong, or even if it goes wrong a little, there would be no impact on our customer, then we are content, and not go further, and most importantly, not create a separate step on the process flow chart.

However, if the above two assumptions are invalid, then we have to get serious, create a separate step in the flow chart, and monitor it.

So it all depends on our judgment Just my 2 cents. :)

Let me ask this question: Can an ("off-line") inspection process have Potential Failure Modes? If so, why wouldn't you evaluate it as a separate processing step and, therefore, have it listed out in the Control Plan?

Another question (perhaps this should be the first question): Do you consider inspection a process?

I would define a process as any step that adds "value" to a product. Therefore I would perform PFMEA on operations that add value.

Inspection does not add value to a product because if you do everything else correctly in the process you shouldn't need inspection/testing. I would also include most handling and movement operations as a non value operation and not perform an FMEA.

Once you start performing FMEA on inspection, testing and handling you will find that you grow the scope of the FMEA to levels that become difficult to manage.

Let me ask this question: Can an ("off-line") inspection process have Potential Failure Modes? If so, why wouldn't you evaluate it as a separate processing step and, therefore, have it listed out in the Control Plan?
Another question (perhaps this should be the first question): Do you consider inspection a process?
Hi Bill, I’m not sure if you are asking me, but I will try to be helpful.
Yes, I consider inspection a process.
Yes, off-line inspection process has potential failure modes.
Yes, I evaluate the off-line inspection process as a separate processing step; however, this evaluation is done in my mind, not in a file. And I may or may not list it out in the control plan or the PFMEA file. It depends on the SOD I evaluated to this inspection process.

If I understand it correctly, the PFMEA is evaluating the following factors of the failure modes:
The severity of the failure
The occurrence of the failure cause
The detection of the occurrence of the failure cause

For inspection process, the failure model, in my opinion, is I fail to identify the defect when it occurs or I identify the product as defective when it is actually OK. If we consider it is defective, we have to take measures, so I fail to identify the defect when it occurs seems a more serious one, because there is no measures following. And the severity of the failure is dependent on the defective product I let pass, so even if the inspection processes are completely the same, the severity rates of failure are not the same.

For example, I have two production lines, each consisting of a cutting process and a dimensional inspection process. The cutting processes make the product’s length dimension and the inspection process check if the dimension conforms to spec.

Line A have a high capability, with a PPM=100, and line B have a relative low capability, with a PPM=1000. Both of them use the same machine, at the same factory. So if you assume the processes before the inspections, that is the manufacturing process, operate correctly, you have to assume both the manufacturing processes have the same capability, but actually they are not.

For the inspection process in the above example, if it fails to function, the product of the low capability manufacturing process pose a greater risk than the product of the high capability manufacturing process, so we may assign a severity rate 1 to the inspection failure corresponding to high capability manufacturing process and 3 to inspection failure corresponding to the low capability manufacturing process.

Then we go to calculate the occurrence of the failure cause. Why I let pass the defective product? Probably I am in a bad mood. I decide to assign the occurrence of bad mood a rate 5.

Then we go to calculate the detection of occurrence of bad mood. How often I know it when I am in a bad mood? I decide to assign a rate 5 to it.

Now we can calculate the SOD of the inspection process, assuming the inspection processes following the two manufacturing process are the same. For the inspection process following the low capability manufacturing process, the SOD is 3*5*5=75. For the inspection process following the high capability manufacturing process, the SOD is 1*5*5=25.

For the inspection process following the high capability manufacturing process, because its SOD (25) is so low, it can’t possible cause me trouble, so I don’t care if it is listed in the control plan. I probably don’t even care it exists. So if the inspection is off line, I don’t write its failure mode analysis down on PFMEA file if it is off line. If it is on line, I may or may not write a separate PFMEA step for it, it doesn’t matter.

For the inspection process following the low capability manufacturing process, because its SOD (75) is high, it might cause me trouble, so it should be listed in the control plan. I probably would suggest it be listed in the control plan if it is not in. I can’t reduce the severity rate, which is decided by the manufacturing process, but I can reduce the occurrence rate of my inspection failure or detection rate of the failure cause. I would write the failure mode analysis for this inspection process down on the PFMEA file, no matter it is on line or off line.

Hi Bill, I’m not sure if you are asking me,
It wasn't directed to anyone, in particular. I was trying to find out how different people "decide" what should/shouldn't be included in Flow Diagram, PFMEA, and CP.

For the inspection process following the high capability manufacturing process, because its SOD (25) is so low, it can’t possible cause me trouble, so I don’t care if it is listed in the control plan. I probably don’t even care it exists. So if the inspection is off line, I don’t write its failure mode analysis down on PFMEA file if it is off line. If it is on line, I may or may not write a separate PFMEA step for it, it doesn’t matter.
I guess my response is that it doesn't matter how capable the manufacturing process is - the inspection process is what it is. Either you address it or you don't.

Here is my take on the documentation aspect - In its "purest" form, the Control Plan would account for most to all controls you have in place for managing your process from receipt of materials through shipment of product. If an item is "worthy" of being addressed in the Control Plan, it should also be addressed in the PFMEA. (By that I mean, if something is so insignificant that it is not addressed in the PFMEA, why would you have that something documented in the Control Plan?)

I would define a process as any step that adds "value" to a product. Therefore I would perform PFMEA on operations that add value.

Inspection does not add value to a product because if you do everything else correctly in the process you shouldn't need inspection/testing. I would also include most handling and movement operations as a non value operation and not perform an FMEA.
The purpose of the PFMEA methodology is to dissect/analyze your entire process and "root out" potential ways the process can fail. Then, based on the risk assessment, controls are put in place to help mitigate the risk to your organization (as well as your customer's). As Jennifer has mentioned in another thread, if you truly believe that inspection adds no value (and I don't want to go there in this thread) then why are you doing it? I agree that if processes were set up and entropy was not a possibility, there should be no need for inspection (but now let's get in the real world).

Once you start performing FMEA on inspection, testing and handling you will find that you grow the scope of the FMEA to levels that become difficult to manage.

Please understand that I'm not trying to say that my way is the only way, but I believe there are Failure Modes in the Inspection, Testing, Material Handling, and Storage processing steps and they are addressed in our PFMEAs (therefore the Control Plan as well). If you are still using spreadsheets for your PFMEA documentation, I agree they can become difficult to manage, but that sounds like an excuse and, IMHO, shortcuts the PFMEA methodology.

Here is my take on the documentation aspect - In its "purest" form, the Control Plan would account for most to all controls you have in place for managing your process from receipt of materials through shipment of product. If an item is "worthy" of being addressed in the Control Plan, it should also be addressed in the PFMEA. (By that I mean, if something is so insignificant that it is not addressed in the PFMEA, why would you have that something documented in the Control Plan?)

I agree, but we should keep things in order. The PFMEA comes first. If inspection is a separate operation or process, with its own potential failure modes, it should be addressed in the PFMEA. (Keep in mind that PFMEA is a process, and not a document. The document is the record of the PFMEA process). Not everything that is considered in the PFMEA process will find its way to the control plan, but if something (such as an inspection operation) is on the PFMEA document and not on the control plan, you should be able to see why it's not on the control plan by looking at the PFMEA output.

The purpose of the PFMEA methodology is to dissect/analyze your entire process and "root out" potential ways the process can fail. Then, based on the risk assessment, controls are put in place to help mitigate the risk to your organization (as well as your customer's). As Jennifer has mentioned in another thread, if you truly believe that inspection adds no value (and I don't want to go there in this thread) then why are you doing it? I agree that if processes were set up and entropy was not a possibility, there should be no need for inspection (but now let's get in the real world).


Please understand that I'm not trying to say that my way is the only way, but I believe there are Failure Modes in the Inspection, Testing, Material Handling, and Storage processing steps and they are addressed in our PFMEAs (therefore the Control Plan as well). If you are still using spreadsheets for your PFMEA documentation, I agree they can become difficult to manage, but that sounds like an excuse and, IMHO, shortcuts the PFMEA methodology.

When you do FMEA on testing / inspection you are really evaulation how good your detection is (ie the failure mode "Accept a defective part") I'd be curious as to what you would use for detection. What is the detection of your detection? I'd like to see some examples of your full FMEA line.

I think trying to do FMEA on handling processes can be very difficult. Material handling would have to start with raw material coming in the door and all material handling processes until the part reachs the customers door. The number of failure modes and different effects that could occur in most cases is almost infinite and cannot be properly evaluated. I'd also be curious on how you write a control plan item for a material handling process step.

Since over 90% of companies don't do FMEA properly I'd prefer to focus on the basics.

The decision to include inspection as a process step in a PFMEA will largely be a judgement call. I would take some of the following into consideration when making this call.

1) Does the inspection serve a strictly monitoring role such as an audit or SPC sampling? This probably should be left as a Control entry for the process step with which it is associated.

2) Does the inspection play a vital role for an incapable process or a critical to quality or product safety (e.g., 100% Hipot test)? This should probably be included as its own process step.

Some example failure modes for inspection would include:

Operation is bypassed.
Good product is rejected.
Bad product is accepted


If the inspection is automated, you can further evaluate it the same as any other process (e.g., test probe does not fully seat, part not seated in fixture, etc.).

When you do FMEA on testing / inspection you are really evaulation how good your detection is (ie the failure mode "Accept a defective part") I'd be curious as to what you would use for detection. What is the detection of your detection? I'd like to see some examples of your full FMEA line.

I think trying to do FMEA on handling processes can be very difficult. Material handling would have to start with raw material coming in the door and all material handling processes until the part reachs the customers door. The number of failure modes and different effects that could occur in most cases is almost infinite and cannot be properly evaluated. I'd also be curious on how you write a control plan item for a material handling process step.

Since over 90% of companies don't do FMEA properly I'd prefer to focus on the basics.

I certainly agree with the >90% figure, but I think that just because you find that a round peg (the potential failure modes of an inspection process) doesn't fit in a square hole (the PFMEA "detection" factor) that you should just forget the whole thing. Don't throw the baby out with the bath water. (That's two clichés in one paragraph now :rolleyes: .) The fact is that every process has potential failure modes that should be addressed whether or not they ever wind up on a PFMEA document or control plan.
Even if you do decide to include it in the document, if it doesn't make sense to use a "detection" factor and calculate an RPN, then just don't do it. You still have the record that the potential issues were considered.

My opinion is similar the opinion from tac123, but do this failure sense?
Than you have a never ending story. You have a detection for the detection for the detection. You should analyse, what kind of failure could happen on the operation inspection, for example damage the test sample or is it possible contaminate the test sample ? What can you do to prevent this?
colorblindness is not the right cause, because the inspectors have to got periodic see tests, isn't it?

The focus should be the product!

Jim123