Hi all. After our successful registration to ISO 9001:2000, my next major task is to work on FMEA review and RPN reduction. This will be the first time my company has actually reviewed FMEA's under circumstances other than a corrective action that requires it. Many control plans will also be updated as a result of the across the board review. The goal, as stated by the QM, is to reduce RPN's. He started looking at the broad picture last week, and ran into this question. "How do we identify that a particular RPN is too high?" The question is subjective of course, and depends on what we could do to improve detection, or reduce occurence. The root of his question, I think, is how do we label a particular RPN for an element of our process as excessive? We manufacture bearing accessories - sleeves, spacers, bearing races, thrust races, etc. We supply to industries of every description, mostly automotive of course, but also marine, aerospace, and manufacturing. No particular customer is pressing for FMEA review or RPN reduction - we are doing it so it doesn't become an issue. Typical severity for us is a 4, occurrence is 3-5, and detection is 6-9. RPN's range from 72-192 with 128-144 being typical. We do a sample plan of product verification at the production cell, checking most features every 15 minutes. Some less critical features are checked a couple of times per shift, or at set-up only. Some critical features are checked 100%. We use a few quality gates, and some machines use an error proofing feature we call "fixed configuration". This is an air detection mechanism built into the work-holding that verfies the part is loaded flush to the locating face of the work holding to reduce parts being turned out of parallel.

My question, to answer the QM's question, is: Is there any "rule of thumb" as to what makes an RPN too high? Is it preferable to be under 100? Obviously, in some cases that won't be possible, but is there a desirable RPN in general?

On machined parts we do not use a fixed level of RPN to review FMEA's and I personally am very suspicious of set targets as it tends to effect the scoring i.e. things just happen to be scored just under this level :notme: .

We are trying to use FMEA as a continous improvement tool not just a compliance item.

We operate review on a number of "Top Items" basis. For example top 3 highest overall items , items with severity over 7, 3 lowest cost to give good benifit/cost items and last but not least, internal Non compliance pereto analysis.

It does not mean we institute projects on all of these, but it makes a good start to focus the efforts

All this goes out of the window if your customer has a CSR for a level but then you are back to my first point.

I agree with Martin, use the severity and RPN to prioritize your actions.

I have direct experience of a customer who wants an action plan for any RPN over 100 and will reject PPAPs if the magic figure is exceeded. Consequently he gets lots of 90ish RPNs on his submissions and the whole exercise becomes pointless.

:2cents:
In my experience the Severity it what is looked at regarding high RPN's. A lot of times the RPN is not even looked at or considered if the Severity is high (8 or above). If you can reduce Severity, it will be an automatic reduction of RPN. Another way to keep RPN's low or at least prove that you have addressd the issue is to Pareto the Top 5 Highest RPN's, work to reduce those, then pick the next top 5, put's you in a constant continual improvement mode and the Customer's and Auditor's love it.
Hope this helps..Good Luck

I agree with the previous posts. Don't let your people get in the habit of using a threshold. RPNs vary form team to team and a risk of 100 for one team may not mean the same level of risk for another team. This is especially true when all teams are not using the same rating scales. As a general rule I suggest always attacking the top 20%.

Additionally I suggest you look for those natural break points where you can see the teams level of concern has significantly dropped. In many cases you will see a step like function happening with the RPNs. Those are what I call the natural break points. You can just keep stepping down as your priorities and resources allow. If you take the top 20% path, there will always be the next 20% to work on. (Continuous Improvement)

One last thing. When you look at reducing risk, you should also look at the criticality matrix, (Severity x Occurrence) and work on these issues. This will force you to take preventative action, since the detection number is not included in the analysis.

I hope this helps.

On machined parts we do not use a fixed level of RPN to review FMEA's and I personally am very suspicious of set targets as it tends to effect the scoring i.e. things just happen to be scored just under this level :notme: .

We are trying to use FMEA as a continous improvement tool not just a compliance item.

We operate review on a number of "Top Items" basis. For example top 3 highest overall items , items with severity over 7, 3 lowest cost to give good benifit/cost items and last but not least, internal Non compliance pereto analysis.

It does not mean we institute projects on all of these, but it makes a good start to focus the efforts

All this goes out of the window if your customer has a CSR for a level but then you are back to my first point.


An excellent approach, in my opinion. :applause:

The intent of a FMEA is to separate the trivial many from the importnat few. If we do not tackle any additional improvement, then the FMEA provided little benefit to the process.

Thanks for the applause. I will be here all week.

Martin,

I think using severity as a driver for review is circumventing the intent of a FMEA. You are correct in using RPN as FMEA views the health of the entire process with severity as part of the equation. Unless otherwise dictated by a customer the RPN is the "process view" and a simple rating method. When companies try to add subjective views in rating processes for improvement it can too cumbersome and lead to mistakes.

Al...

Martin,

I think using severity as a driver for review is circumventing the intent of a FMEA. You are correct in using RPN as FMEA views the health of the entire process with severity as part of the equation. Unless otherwise dictated by a customer the RPN is the "process view" and a simple rating method. When companies try to add subjective views in rating processes for improvement it can too cumbersome and lead to mistakes.

Al...

Al
I hear what you say, but I was trying to suggest to rstocum, the original questioner, that fixed rpn targets as the end in themself, are not necessary the only way when "annual" review is being undertaken.

FMEA in my opinion, right or wrong, is yhat its nature is very subjective and company/product/process specific. I accept what appears to work here may not be applicable elsewhere and he will have to find his own way forward.

We are trying, but not always suceeding, in looking at a broader picture in considering projects on the high overall rpn's, on high risk areas when failure occures i.e the high severity items and finally some quck easy fix/low cost items just to get into the way of winning some progress.

We went down this route in an effort to break a culture that had started to pay a certain degree of lipservice to FMEA's numbers and not considering value added benifits to the customer and ourselves. I accept we may have gone to far away from the ideal but some improvements are being seen.

Martin,

I think using severity as a driver for review is circumventing the intent of a FMEA.

Al...

I hate to agree with someone twice in one thread but I'm on Martin's side here again.
The AIAG PFMEA manual states that " In general practice, regardless of the resultant RPN, special attention should be given when severity is high".

I know several organizations that use Severity x Occurrence to prioritize their activities.

I like that Martin is looking at the high severity. One should always look for opportunities to reduce the severity by eliminating the FM or through the use of a compensating provision. However you can not always reduce the severity, nor is it always practical to do so. One can not understand the risk without taking into account the occurrence of the Causes. Pro-active / preventative controls that reduce the frequency of causes are often less expensive and more effective at reducing the risk. If I have a Severity of 8 and an occurrence of 1 associated with it, but I have a Severity of 7 and an occurrence of 4 associated with it. I will miss reducing the greater risk opportunity if I focus on Severity alone.

rstocum - You've gotten some pretty sage advice already and I'm not going to chime in in this particular thread. Please scroll to the bottom of the page and there are other threads listed which discuss this very question - at length. If, after reviewing those, you still have some confusion and/or questions or inputs, you could either resurrect one of those threads or come back here.

The bottom line is that you need to adopt a methodology that works for your organization. If that methodology works, you need to have the backbone to tell your customer to take a hike (sorry but I didn't want to sugarcoat it:notme: ). Common sense needs to prevail in FMEA reviews - not a customer induced "bogey".

Thank you all. I was hoping to hear guidance not to go with a threshold number. Pareto charting the highest severity x occurrence number seems a good way to start. Checking the other similat threads got me onto correlating those with associated quality cost too. The QM will like that.

Thanks again!

Rich Stocum