DFMEA detection actions

[stm55]

I thought I had a good understanding of DFMEA, however I recently got my hands on some ASQ training, and I am seeing something that I don't agree with/understand.

They start by saying a primary assumption of DFMEA is that the part is produced to specification (which I agree with). However, the ASQ training shows a single line item example of DFMEA where it is essentially:
Monitoring Bracket fails to support motor ---> leading to bracket breaks and motor slides out of harness ---> caused by material out of spec.... Detection controls are "Random manual spot check material inspection", and as a result of a high RPN, a possible new Detection action is "Select new supplier that can certify all received brackets are in line with spec"

To me this whole line is essentially for PFMEA. It does not get into the design related causes of the failure, it is strictly related to a supplier not being able to meet the print. To use the same example, I would think it'd be more appropriate if the cause were something like "Material design inadequate to withstand application", and the detection controls would be something like specific testing (design testing, not routine incoming inspcetion) done to determine whether the correct material specifications were chosen...

Based on this, I'm thinking it likely that this is just an error with the training, and they meant to say this was for PFMEA? (this is not uncommon that these sorts of trainings are not well-reviewed in my experience-- I saw some RAPS training content that was blatantly contradictory/clearly not what was intended. Any input is much appreciated!

 

[Tidge]

In my professional opinion... there is a spectrum of understanding when it comes to analyzing failure modes, especially for examples as described above.

I don't like this failure mode (cause = "material out of spec") in a Design FMEA because the failure mode "Monitoring Bracket fails to support motor" isn't with the design, because the material spec was chosen to be appropriate to provide the necessary support. However...

...it probably is the case that in the Design FMEA, the specification for the material would be listed as Verification of implementation (VoI) and some specific test report is listed as the Verification of Effectiveness (VoE) for the design. Once that line exists, there happens to be a convenient place to list some "process controls"... if there is not a specific process that otherwise includes such a control.

I really don't like "pick a new supplier" as a control, because (as noted) the material itself should have been established to prevent the failure mode to reduce the risk commensurate with the evidence about the material. I also don't think there is value in generating FMEA for quality processes like supplier selection. If there is a legitimate chance that the supplier controls are insufficient, new controls have to be added somewhere, so the DFMEA might end up being the best place for them.

 

[stm55]

In my professional opinion... there is a spectrum of understanding when it comes to analyzing failure modes, especially for examples as described above.

I don't like this failure mode (cause = "material out of spec") in a Design FMEA because the failure mode "Monitoring Bracket fails to support motor" isn't with the design, because the material spec was chosen to be appropriate to provide the necessary support. However...

...it probably is the case that in the Design FMEA, the specification for the material would be listed as Verification of implementation (VoI) and some specific test report is listed as the Verification of Effectiveness (VoE) for the design. Once that line exists, there happens to be a convenient place to list some "process controls"... if there is not a specific process that otherwise includes such a control.

I really don't like "pick a new supplier" as a control, because (as noted) the material itself should have been established to prevent the failure mode to reduce the risk commensurate with the evidence about the material. I also don't think there is value in generating FMEA for quality processes like supplier selection. If there is a legitimate chance that the supplier controls are insufficient, new controls have to be added somewhere, so the DFMEA might end up being the best place for them.

Thanks for your thorough response. When you say a "spectrum of understanding", I assume you are saying that not everybody has the same understanding of general FMEA concepts? There are different "shades" or correctness, but (at least from my research/experience) there are quite a few self proclaimed "FMEA experts" out there who probably would not agree with each other?

Also your 3rd paragraph about the convenient spot to put process controls... if there's not a more appropriate spot for them---
Are you basically saying that a PFMEA is probably a better spot for something like that if you have one that addresses the specific failure mode, but if you don't a specific DFMEA scenario may also allow for it? If so, where would you list process controls in a DFMEA? In my understanding, in its most simplified explanation a DFMEA is to determine ways the product's design goes wrong/was designed incorrectly. Most process controls are theoretically not relevant (there's probably some exceptions that can be argued, but I don't think "supplier providing cert to drawing" is one of them).

 

[Miner]

I thought I had a good understanding of DFMEA, however I recently got my hands on some ASQ training, and I am seeing something that I don't agree with/understand.

They start by saying a primary assumption of DFMEA is that the part is produced to specification (which I agree with). However, the ASQ training shows a single line item example of DFMEA where it is essentially:
Monitoring Bracket fails to support motor ---> leading to bracket breaks and motor slides out of harness ---> caused by material out of spec.... Detection controls are "Random manual spot check material inspection", and as a result of a high RPN, a possible new Detection action is "Select new supplier that can certify all received brackets are in line with spec"

To me this whole line is essentially for PFMEA. It does not get into the design related causes of the failure, it is strictly related to a supplier not being able to meet the print. To use the same example, I would think it'd be more appropriate if the cause were something like "Material design inadequate to withstand application", and the detection controls would be something like specific testing (design testing, not routine incoming inspcetion) done to determine whether the correct material specifications were chosen...

Based on this, I'm thinking it likely that this is just an error with the training, and they meant to say this was for PFMEA? (this is not uncommon that these sorts of trainings are not well-reviewed in my experience-- I saw some RAPS training content that was blatantly contradictory/clearly not what was intended. Any input is much appreciated!

You are 100% correct in thinking this. It is a very common tendency for the design people to blame all failure causes on manufacturing/suppliers rather than consider that they might be the cause. In addition, they will often cite a manufacturing test as the design control. I placed one restriction on both causes and on controls in a Design FMEA: Both causes and controls MUST EXIST prior to release to production. That took all arguments away from them.

In the ASQ example you cited, the cause should be that the wrong material was specified, or the operating stresses were unknown/underestimated, and the design control(s) should be something like a Finite Element Analysis (FEA), or worst-case strength testing. Note that in ASQ's example both occur after release to production and my example occurs prior to release.

 

[Bev D]

Miner nailed it! (As usual)
DFMEA is tough for design engineers to understand. The purpose is to ensure that the Design is correct; it guides development activity such as experiments to determine specifications and then verify & validate the design (these are the important controls).

I would also add that simply saying the cause is “wrong specification” is not sufficient. ( and a lazy way out)
In your example of a material that ‘breaks’, the engineer must specify what material properties and characteristics are important. What type of material? What thickness? Hardness? Ductility?
You might want to take a look at this process ‘map’ I used to demonstrate and specify the ‘design process’ from FMEA on to release.

 

[Tidge]

FWIW, for my initial response in this discussion: I am assuming that the specification was correct (supported by plenty of OE during design verification), but that due to some defect in the supply chain the sourced material did not meet the specification.

Thanks for your thorough response. When you say a "spectrum of understanding", I assume you are saying that not everybody has the same understanding of general FMEA concepts? There are different "shades" or correctness, but (at least from my research/experience) there are quite a few self proclaimed "FMEA experts" out there who probably would not agree with each other?

I have worked with ASQ question banks for their BoK: writing questions, reviewing materials submitted by others, and reviewing comments made by others on materials I have submitted. A few fundamental things about my experiences:

1. ASQ makes updates to materials "when they feel like it", the process is opaque enough that I can't say what goes into that decision.
2. The folks who make updates to materials are self-selected volunteers, sometimes chosen "at random".
3. The folks who do the reviews are (mostly) self-selected volunteers.

This is not to say that if there is someone in the ASQ hierarchy that is a "recognized FMEA expert" that such a person wouldn't keep controls of "their" materials with an iron fist. I get the sense that ASQ leadership is something like a local Moose lodge.

My "spectrum" comment is meant to imply that there will be items present in official ASQ materials that are (a) more wrong than right, and (b) weirdly specific to circumstances that are unknowable to almost everyone. The latter is sometimes due to "knowledge" from a very specific industry, sometimes it is due to the peculiar need for most (all?) ASQ materials to require some sort of "published" source. It is entirely possible that this (poor) example was found in a published source and that is how it got added to the training... as well as some question bank. I could not get wrong or weirdly specific questions removed from a question bank, despite finding multiple sources which pointed out the incorrect (or highly ambiguous) question and answer... because the "wrong" answer had been published somewhere.

 

[dramman]

I thought I had a good understanding of DFMEA, however I recently got my hands on some ASQ training, and I am seeing something that I don't agree with/understand.

They start by saying a primary assumption of DFMEA is that the part is produced to specification (which I agree with). However, the ASQ training shows a single line item example of DFMEA where it is essentially:
Monitoring Bracket fails to support motor ---> leading to bracket breaks and motor slides out of harness ---> caused by material out of spec.... Detection controls are "Random manual spot check material inspection", and as a result of a high RPN, a possible new Detection action is "Select new supplier that can certify all received brackets are in line with spec"

To me this whole line is essentially for PFMEA. It does not get into the design related causes of the failure, it is strictly related to a supplier not being able to meet the print. To use the same example, I would think it'd be more appropriate if the cause were something like "Material design inadequate to withstand application", and the detection controls would be something like specific testing (design testing, not routine incoming inspcetion) done to determine whether the correct material specifications were chosen...

Based on this, I'm thinking it likely that this is just an error with the training, and they meant to say this was for PFMEA? (this is not uncommon that these sorts of trainings are not well-reviewed in my experience-- I saw some RAPS training content that was blatantly contradictory/clearly not what was intended. Any input is much appreciated!

you are correct in your understanding

 

[Kim from Korea]

I thought I had a good understanding of DFMEA, however I recently got my hands on some ASQ training, and I am seeing something that I don't agree with/understand.

They start by saying a primary assumption of DFMEA is that the part is produced to specification (which I agree with). However, the ASQ training shows a single line item example of DFMEA where it is essentially:
Monitoring Bracket fails to support motor ---> leading to bracket breaks and motor slides out of harness ---> caused by material out of spec.... Detection controls are "Random manual spot check material inspection", and as a result of a high RPN, a possible new Detection action is "Select new supplier that can certify all received brackets are in line with spec"

To me this whole line is essentially for PFMEA. It does not get into the design related causes of the failure, it is strictly related to a supplier not being able to meet the print. To use the same example, I would think it'd be more appropriate if the cause were something like "Material design inadequate to withstand application", and the detection controls would be something like specific testing (design testing, not routine incoming inspcetion) done to determine whether the correct material specifications were chosen...

Based on this, I'm thinking it likely that this is just an error with the training, and they meant to say this was for PFMEA? (this is not uncommon that these sorts of trainings are not well-reviewed in my experience-- I saw some RAPS training content that was blatantly contradictory/clearly not what was intended. Any input is much appreciated!

I think you are correct and I think the same.

Firstly, 'material out of spec' should not be considered in both DFMEA and PFMEA.
In PFMEA we consider all incoming material has no defects. PFMEA focuses on only your manufacturing process.

Secondly, even if 'material out of spec' can be used in PFMEA, this ambiguous expression is not appropriate in FMEA.
Material spec value could be higher than upper spec limit and also could be lower than lower spec limit.
Those 2 cases will have totally different failure effect.

Thirdly, 'Select new supplier' cannot be detection measure.
Detection measures are usually tests to recognize failures at final/semi product.

All I mentioned above are described in AIAG-VDA FMEA handbook.
I hope my reply helps you.

Thank you.

 

[Ed Panek]

Ive seen many times a part is respecified after learning more about the part variance after mass production and lot to lot variance occurs within specification.

I once worked at a company that made film printers. We built the printers and sourced the film from a film company. During design, we used several lots of the suppliers film to V&V the design of the printer. We later received complaints about certain lots of film jamming in the printer. I was the manager of film at the time (Supplier wise) and I investigated. There were hidden specification the film supplier had we were unaware of that did impact our ability to utilize their film consistently. We had to add a new part of our film specification to make sure only some material was accepted.

Obviously adding "We learn a new specification of film manufacturing leading to design failure" is not actionable and was not part of our FMEA process.

 

[Bev D]

The purpose of FMEA is to iteratively understand and improve your product designs and manufacturing processes. Detection in a design FMEA is (primarily) in the design reviews, verification and validation testing. Too often the design and development phase ignores the variation inherent in supplied materials (specified by custom design or from the data sheets - and hidden allowances as Ed pointed out - of off the shelf parts. This later comes to bite you in manufacturing or in the Customer’s and User’s hands.

I’m not necessarily advocating that this should be in a process FMEA but that we need to ensure that everyone does there job without resorting to finger pointing at other functions…