Risk and Corrective actions - Currently no FMEA's - Car systems
- Thread starter marcusja2002
- Start date
indubioush
Ah ha!
What type of organization do you work for?
That's a good idea.
I don't see the relationship. An FMEA might be helpful to understand the underlying causes of the nonconformity but not, I don't believe, in determining the risk.
Generally, risk is the product of severity and likelihood (probability). So keep it simple. A 5x5 matrix (severity x probability) is pretty typical. You would need to define your levels of both severity and probability (based on the answer to the question @indubioush posed) and then use that as a guide to drive your subsequent actions.
marcusja2002
Involved In Discussions
indubioush
Manufacturing / Assembly
yodon
You actually see the relationship better then you realize. The FMEA will develop an RPN number which consists of Severity, Likelihood of Occurance and Detectability. I'm used to having a detailed FMEA to guide every possible failure and if a new failure shows up, the FMEA is updated with eventual controls put in place through the corrective action process. The FMEA is supposed to be a living document.
Problem is that is a very alien idea to my current company and would be an elephantine task to push through at time time. That is why I'd like to develop a CAR system without the need for the FMEA's and then use the CAR system to help develop them slowly.
Then over time with new development introduce fresh FMEA's. Chicken vs Egg.
The problem is our product has 1000's of parts with multiple failure modes, so I'm having a difficult time visualizing just a Severity and Probability chart.
This actually was an ad that popped us as I was typing. Connecting this to injurying or the ability to function is where I have the tough time still seeing.
Manufacturing / Assembly
yodon
You actually see the relationship better then you realize. The FMEA will develop an RPN number which consists of Severity, Likelihood of Occurance and Detectability. I'm used to having a detailed FMEA to guide every possible failure and if a new failure shows up, the FMEA is updated with eventual controls put in place through the corrective action process. The FMEA is supposed to be a living document.
Problem is that is a very alien idea to my current company and would be an elephantine task to push through at time time. That is why I'd like to develop a CAR system without the need for the FMEA's and then use the CAR system to help develop them slowly.
Then over time with new development introduce fresh FMEA's. Chicken vs Egg.
The problem is our product has 1000's of parts with multiple failure modes, so I'm having a difficult time visualizing just a Severity and Probability chart.
This actually was an ad that popped us as I was typing. Connecting this to injurying or the ability to function is where I have the tough time still seeing.
Tagin
Trusted Information Resource
If I understand correctly, are you essentially looking for some kind of risk model for the CAR system to feed into, in a manner similar to how you describe CARs feeding into an FMEA model in your previous experience?
marcusja2002
Involved In Discussions
Yes
Using Yodons Idea, I believe if I can get Management to set these limits I might be able to get this system moving.
Set a defect limit 3,5,10 etc. and a Severity which may not be a number but a type of action, shutting down the line, prevention of specific functions, etc.
If I have that list created I believe I can make this work.
Attachments
Just be careful about getting too complex.
I tend to go for simplicity. Is the NC systemic? If yes, then CAPA. If not, then fix it.
If you have a systemic, low-severity NC, it would appear that you might not consider it eligible for a CAPA.
If you have a process that is only done every other quarter and it consistently produces NCs, it might be excluded based on question 3.
Not saying this won't work for you; just making sure you're thinking it through and considering the effects.
I tend to go for simplicity. Is the NC systemic? If yes, then CAPA. If not, then fix it.
If you have a systemic, low-severity NC, it would appear that you might not consider it eligible for a CAPA.
If you have a process that is only done every other quarter and it consistently produces NCs, it might be excluded based on question 3.
Not saying this won't work for you; just making sure you're thinking it through and considering the effects.
Tagin
Trusted Information Resource
I agree w/Yodon. That kind of algorithmic decision tree is restrictive if used as a rule. It might be useful as an aid, but I wouldn't be constrained by it.
- Management's option for an 8D doesn't even show up unless all the other preceding decisions pass.
- The defect qty in most cases needs to be related to the volume processed in a timeframe. In other cases, the defect qty might be unique by specific product.
You might be better off to have something that simply says something like:
"The quality person/team will consider defect qty (or percentage), defect trends, severity, customer and management input in determining if an 8D should be initiated.
If within the scope of an existing 8D, then the issue will be handled under that 8D." etc., etc.
- Management's option for an 8D doesn't even show up unless all the other preceding decisions pass.
- The defect qty in most cases needs to be related to the volume processed in a timeframe. In other cases, the defect qty might be unique by specific product.
You might be better off to have something that simply says something like:
"The quality person/team will consider defect qty (or percentage), defect trends, severity, customer and management input in determining if an 8D should be initiated.
If within the scope of an existing 8D, then the issue will be handled under that 8D." etc., etc.
Bill Levinson
Industrial Statistician and Trainer
Re: "Problem is that is a very alien idea to my current company and would be an elephantine task to push through at time time. That is why I'd like to develop a CAR system without the need for the FMEA's and then use the CAR system to help develop them slowly. "
The new AIAG FMEA manuals are excellent, and improve greatly on the approach from 15 or 20 years ago. They provide much better guidance on the Occurrence and Detection rating.
I have however been looking at a simpler option for organizations that are not required to do formal FMEAs and do not want to use them. The Army uses DD2977 (Deliberate Risk Assessment Worksheet) which is in the public domain as a publication of the U.S. Government. It uses only a Severity rating and a combined probability rating (which seems to merge Detection and Occurrence) and a risk matrix very similar to what you included above. The Army's Risk Management Process also raises the point that the risk of something happening includes not only its probability (the FMEA occurrence rating) but also the frequency with which we are exposed to the risk.
This can still be connected to a control plan for each potential failure mode, but the assessment process is simpler than for traditional FMEA.
The new AIAG FMEA manuals are excellent, and improve greatly on the approach from 15 or 20 years ago. They provide much better guidance on the Occurrence and Detection rating.
I have however been looking at a simpler option for organizations that are not required to do formal FMEAs and do not want to use them. The Army uses DD2977 (Deliberate Risk Assessment Worksheet) which is in the public domain as a publication of the U.S. Government. It uses only a Severity rating and a combined probability rating (which seems to merge Detection and Occurrence) and a risk matrix very similar to what you included above. The Army's Risk Management Process also raises the point that the risk of something happening includes not only its probability (the FMEA occurrence rating) but also the frequency with which we are exposed to the risk.
This can still be connected to a control plan for each potential failure mode, but the assessment process is simpler than for traditional FMEA.
Similar threads
