How to develop a Control Plan from a PFMEA (Process FMEA)

Kales Veggie

People: The Vital Few
#1
Today I received a question about how to develop a control plan from a PFMEA. The APQP manual was the first thing that came to mind, then I researched other threads for a "cookbook" or "guidance document" or other hands-on literature but have not found what I am looking for.

Can you share how you develop a control plan from a PFMEA?
 
#2
Re: Develop Control plan from PFMEA: how to?

Here's how I do it. This method has passed 1st, 2nd and 3rd party audits (and I have received compliments on ease of translation).

Start with your process flow diagram, here's a simple example:
1 - Incoming inspection
2 - Assembly
3 - Final Test
4 - Shipping

Notice the numbering!

Create your PFMEA (simplified for space) from the flow diagram (e.g. Process Function/Requirements - Potential Failure Mode - Current Process Controls. I didn't fill in all the columns here for simplicity):
1.1 - Incoming inspection - wrong parts - match parts against Mfg. and Mfg. number
2.1 - Assembly - incorrect assembly - use drawing XYZ
2.2 - Assembly - incorrect torque - note torque values on page 3 of drawing XYZ - mark torqued items with paint pen
3.1 - Final test - outputs out of specification - test per instruction 123.
3.2 - Final test - Key Control Characteristic - Voltage Out - section 1 on 123
3.3 - Final test - KCC - Current Out - Section 2 on 123
4.1 - Shipping - damaged packaging and or product- design and test packaging per requirements - see results in APQP file
4.2 - shipping - incorrect packaging - training to requirements
4.3 - shipping - incorrect packaging - product/dock audits per audit schedule

Create your Control Plan from the PFMEA, matching those items that will be ongoing (note that Items 4.1 and 4.2 were resolved at PFMEA stage and the ongoing requirements are in section 4.3)
Number - Process Name - Control Method - Reaction Plan (columns left out for simplicity)
1.1 - Incoming inspection - Incoming Inspection Procedure - Supply Corrective Action Procedure
2.1 - Assembly - Drawing XYZ - Reject and Rework, Control of Non-Conforming Product Procedure
2.2 - Assembly - Torque values on page 3 of drawing XYZ, mark torqued items with paint pen - Reject and Rework, Control of Non-Conforming Product Procedure
3.1 - Final test - test per instruction 123 - Reject and Rework, Control of Non-Conforming Product Procedure
3.2 - Final test, KCC, Voltage Out - Section 1 on 123 - Reject and Rework, Control of Non-Conforming Product Procedure
3.3 - Final test, KCC, Current Out - Section 2 on 123 - Reject and Rework, Control of Non-Conforming Product Procedure
4.3 - Shipping - Customer packaging requirement E16 & Our Packaging Instruction, product/dock audits per audit schedule - Reject and Rework, Control of Non-Conforming Product Procedure & Improvement Procedure for any audit findings or observations.

I hope that makes sense. The important thing here is to have some way to follow "Process A" from Process Flow to PFMEA to Control Plan. Obviously, this has been incredibly simplified and there is a lot of missing data from both PFMEA and Control Plan that you will need to address.

If this is WAY too simple, PM me and I'll send you a redacted version of a PFMEA -> Control Plan from our latest product.
 
Last edited:

Caster

An Early Cover
#3
Re: Develop Control plan from PFMEA: how to?

Nicely done Icy

I was taught that the Flow - FMEA - Control Plan is a loop

First list your process steps (Flow)

Identify risks in each process step (FMEA)

Put in place controls to manage those risks (Control Plan)

Of course in the lean mean automotive scene of the 00s we all pretty much only have time to quickly copy the last control plan, change the part number and dimensions, find a decent looking FMEA on the Cove, change the header and send all it in with the PPAP package...and hope the customer has contracted out the review.

A real shame, cause when it's done right it works!
 
#4
Re: Develop Control plan from PFMEA: how to?

Wow! That's a lot of TKS and Karma!

Caster posted a much more concise version. This really is a flow from Process Diagram to PFMEA to Control Plan.

Let me re-emphasize this point: Whether you use numbers, letters, color coding or green beans, use some kind of symbols to connect (and organize) each process/item from one document to the next. Organization is the foundation of control.
 
N

nativeamrc

Guest
#5
Re: Develop Control plan from PFMEA: how to?

Here's how I do it. This method has passed 1st, 2nd and 3rd party audits (and I have received compliments on ease of translation).

Start with your process flow diagram, here's a simple example:
1 - Incoming inspection
2 - Assembly
3 - Final Test
4 - Shipping

Notice the numbering!

Create your PFMEA (simplified for space) from the flow diagram (e.g. Process Function/Requirements - Potential Failure Mode - Current Process Controls. I didn't fill in all the columns here for simplicity):
1.1 - Incoming inspection - wrong parts - match parts against Mfg. and Mfg. number
2.1 - Assembly - incorrect assembly - use drawing XYZ
2.2 - Assembly - incorrect torque - note torque values on page 3 of drawing XYZ - mark torqued items with paint pen
3.1 - Final test - outputs out of specification - test per instruction 123.
3.2 - Final test - Key Control Characteristic - Voltage Out - section 1 on 123
3.3 - Final test - KCC - Current Out - Section 2 on 123
4.1 - Shipping - damaged packaging and or product- design and test packaging per requirements - see results in APQP file
4.2 - shipping - incorrect packaging - training to requirements
4.3 - shipping - incorrect packaging - product/dock audits per audit schedule

Create your Control Plan from the PFMEA, matching those items that will be ongoing (note that Items 4.1 and 4.2 were resolved at PFMEA stage and the ongoing requirements are in section 4.3)
Number - Process Name - Control Method - Reaction Plan (columns left out for simplicity)
1.1 - Incoming inspection - Incoming Inspection Procedure - Supply Corrective Action Procedure
2.1 - Assembly - Drawing XYZ - Reject and Rework, Control of Non-Conforming Product Procedure
2.2 - Assembly - Torque values on page 3 of drawing XYZ, mark torqued items with paint pen - Reject and Rework, Control of Non-Conforming Product Procedure
3.1 - Final test - test per instruction 123 - Reject and Rework, Control of Non-Conforming Product Procedure
3.2 - Final test, KCC, Voltage Out - Section 1 on 123 - Reject and Rework, Control of Non-Conforming Product Procedure
3.3 - Final test, KCC, Current Out - Section 2 on 123 - Reject and Rework, Control of Non-Conforming Product Procedure
4.3 - Shipping - Customer packaging requirement E16 & Our Packaging Instruction, product/dock audits per audit schedule - Reject and Rework, Control of Non-Conforming Product Procedure & Improvement Procedure for any audit findings or observations.

I hope that makes sense. The important thing here is to have some way to follow "Process A" from Process Flow to PFMEA to Control Plan. Obviously, this has been incredibly simplified and there is a lot of missing data from both PFMEA and Control Plan that you will need to address.

If this is WAY too simple, PM me and I'll send you a redacted version of a PFMEA -> Control Plan from our latest product.
Hey Icy Mountain:

Your knowledge is admired. Would you please send me a copy of a redacted plan I can look at? Thanks!
 
#7
Here they are. Just in case you are trying to get into a few more details, here's the key to all the abbreviations. Abbreviations are your friend on these documents. They should be concise guidelines to your control. Do not write level 3 test procedures here!
PM = Procedures Manual (our "Quality Manual")
PR = Product Realization
BOM = Bill of Materials
WI = Work Instruction (a.k.a. Level 3 procedure)
SAR = Supplier Action Request
PCB = Printed Circuit Board
SMT = Surface Mount Technology
KCC = Key Control Characteristic = § (the generic symbol for KCC)

Please note that you absolutely must cover the KCCs in your PFMEA and Control Plan and they should be marked with § (or your customer's required symbol). Furthermore, if you have lower level procedures that specify how to measure they should be marked there and anyplace you record them §.

Note the connection between the # column on the PFMEA (e.g. 01-01) with the Part/Process Number column (e.g. 1) on the Control Plan. The only characteristics on the Control Plan that I have broken out specifically are the KCC's. They are all listed as one item on the PFMEA (07-01) so they are listed on the the Control Plan in Part/Process Number 7 but all have the same characteristic number 1. If you really want to get technical, these should be listed on the PFMEA as 07-01, 07-02 and 07-03 then listed on the Control Plan as Part/Process Number 7 and characteristics 1, 2, 3. That level of detail is probably overkill. Make sure that you are matching the Process/Function Requirements Number to the Part/Process Number and you should have enough for most folks to connect everything together.
 

Attachments

Caster

An Early Cover
#8
Icy

Thanks for taking the effort to post such a detailed example.

Have you ever been called out on the reaction plan part of the CP?

It says adjust....there MAY be a missing link to go back to last good check, and retest all suspect product.

We do the same, very little concrete advise in the RP column.

I am in the process of developing operator friendly, detailed and specific RP for each characteristic.

Interestingly, this was driven by operator request, never had a customer or registrar complain.
 
#9
In my experience, an auditor digging down to this level has not found anything significant in my quality system and is now desperately trying to find something, anything, to save him from and empty finding list. Quite simply, this system meets the shalls and I have enough evidence in low scrap, low rework, and low warranty to prove the system effective (YMMV). I even have a few instances where the correct and feedback system generated a trend that ended up in our Continual Improvement system for correction. We don't need to generate a corrective action and an 8D for every single non-conformance: such a requirement would be extremely wasteful and inefficient given our company's operating philosophy.

The "correct and feedback" system is built into our company culture. Since we are a high mix, low volume manufacturer with very little automated assembly and test equipment, we rely on 100% testing and each subsequent operation depending on the correctness of the previous operation. For example, if you do not install a part correctly, the next station cannot install their part. If a product fails voltage output, the rest of the batch is not "suspect" because you are testing the voltage output of every unit. This correct and feedback is recorded but the "quality" group does not get involved unless the rework times or incidents rise to a significant level. Usually, returning the rejected piece to the station where the problem was introduced (for rework) is enough to stop it. We may have a couple more pieces in the pipeline but it is vanishingly rare for us to have a large pile of product, all with the same defect.
 

Stijloor

Super Moderator
Staff member
Super Moderator
#10
In my experience, an auditor digging down to this level has not found anything significant in my quality system and is now desperately trying to find something, anything, to save him from and empty finding list. Quite simply, this system meets the shalls and I have enough evidence in low scrap, low rework, and low warranty to prove the system effective (YMMV). I even have a few instances where the correct and feedback system generated a trend that ended up in our Continual Improvement system for correction. We don't need to generate a corrective action and an 8D for every single non-conformance: such a requirement would be extremely wasteful and inefficient given our company's operating philosophy.

The "correct and feedback" system is built into our company culture. Since we are a high mix, low volume manufacturer with very little automated assembly and test equipment, we rely on 100% testing and each subsequent operation depending on the correctness of the previous operation. For example, if you do not install a part correctly, the next station cannot install their part. If a product fails voltage output, the rest of the batch is not "suspect" because you are testing the voltage output of every unit. This correct and feedback is recorded but the "quality" group does not get involved unless the rework times or incidents rise to a significant level. Usually, returning the rejected piece to the station where the problem was introduced (for rework) is enough to stop it. We may have a couple more pieces in the pipeline but it is vanishingly rare for us to have a large pile of product, all with the same defect.
Icy,

Congratulations! This is an excellent description of an effective quality system. Nothing "high-tech", straightforward timely internal customer feedback and immediate response. It works. In my opinion, some quality systems are unnecessarily complex; either to make an impression on someone or to create job-saving busy work. It seems that you have covered all the bases. I looked at the previous posts. Again, great work. :applause: :applause:

Stijloor.
 
Top