The FMEA Mini-Series: Using an FMEA vs. SxO for Prioritizing


On page 43, the standard states that when the severity of the RPN is high, special attention should be given to it. What does everyone consider high? We have been using 100.

Al Dyer

I have found that different suppliers have various "limits" for RPN's but I think the bigger picture would be to have some type of prioritized reduction plan for your highest RPN's.

In a previous life we even had a helpful SQA that gave us RPN numbers of our competitors (no names involved) to benchmark against.

I would like to hear from other forum members if they have any specific supplier limits they could share.


Al Dyer
Mngt. Rep.
[email protected]


"What does everyone consider high"
IMO 1000 or higher. Special attention when severity is 7 or higher.

From the FMEA manual;
"The risk oriority number is the product of severity (S), Occurrence (O), and detection (D). RPN = (S) x(OD) x (D)
This value should be used to rank order the concerns in the process (e.g., in Pareto fashion)
The RPN will be between "1" and "1,000". For higher RPN,s (higher than 1000) the team must undertake efforts to reduce this calculated risk through corrective action(s). In general practice, regardless of the resultant RPN( S x O x D ), special attention should be given when severity (page 35) is high (7,8,9,10)"
Note: expressions in parenthesis is my input.

[This message has been edited by Sam (edited 17 January 2001).]

Kevin Mader

One of THE Original Covers!
I would discourage an arbitrarily established value. Failing to deal with a value below '100' might prove to be determental.

On the other hand, unwisely dispensing resources to countermeasure a low rating could prove equally bad.

Keep in mind that the FMEA is a LIVING DOCUMENT. This means it must be revisited from time to time. As such, you might find that an RPN for an item that was low because of a false expectation proves to be high. The reverse is also true. Still, as the process or system changes, the RPN could toggle back and forth.

So when do you react? Well, what is a manageable risk? Keep in mind that the organization for the most part is the AUTHORITY on the process or design. You make the decision as to what is a high or low risk (please pay attention to regulatory requirement, ie Essential Requirements which govern your actions). Being the authority on the product/process, the experts, and having knowledge of your manageable resources, you pick your own battles.

Another not so scientific method, but perhaps better than arbitrarily established RPNs, is to use the 80/20 principle. Use Pareto analysis to establish your priorities and address the "vital few" as your resources permit. Use Corrective/Preventive Actions. As you periodically do your review the FMEA, update your RPNS,reprioritize, and repeat the process.

Just another opinion.



J.R. Strickland

At a minimum, we require the following...

1. Any RPN >= 100 must have an action plan to reduce.

2. If no RPN's are over 100, the top 3 RPN's must be considered. Why considered and not an action plan?...because if my top RPN is 20, I will probably get more value out of my engineers elsewhere.

3. Any severity of 9 or 10 must be evaluated. Again, the resultant RPN could be low if occurrence and detectability are 1, but because a 9 or 10 can affect safe vehicle operation or impact a govt. regulation, they deserve special attention.

dan dan the qualityman

You have all posed some interesting points of view with respect to the action or lack of action relating to a particular level of RPN. As I see it, the FMEA provides a micro view of your product or system in a prioritized fashion similiar to the Pareto chart which we are all familiar with. The major difference is that the FMEA provides actions and responsibilities and a feedback on the effect of these actions via the re-calculated RPN. Each element on the FMEA should be looked at as an oppurtunity for improvement, obviously with the higher RPN's providing the greatest bang for a buck. Rather than only focusing on RPN's reaching a level you should look at the entire system. Perhaps reducing a number of smaller RPN's may be more timely or economical than placing all of your resources on a single element. Additionally you should not discount the effect of lowering one RPN on other elements in the system. You may see some surprising relationships.


I’m agreeing with Al. In the spirit of continuous improvement, there needs to be some kind of prioritized reduction plan (a.k.a… the pareto chart).

I have facilitated a number of FMEA meetings. I have noticed if there is a arbitrary cut off value, all of the RPN’s magically and conveniently fall below that value. Now, it make take a bit of time and group justification to accomplish this, but by the end of the day they have convinced themselves of it.

If the RPN’s are on a pareto chart, I have found the group to achieve more realistic RPN’s and make the FMEA process more productive.

Anybody else experience this?

[This message has been edited by AJLenarz (edited 15 February 2001).]

[This message has been edited by AJLenarz (edited 15 February 2001).]

Al Dyer


Also agree and have been through it in depth.

Once a limit is set, it seems to be the mind set that human nature will direct you to set your "goals" below the acceptable limit.

Although goals need to be set, once continuous improvement is in place those goals have to be reset downward to continue the improvement process.

I think that internal goals, such as "PPM" must be set lower than any customer expectation.

I believe and have read/heard the mantra "Zero Defects".

Work this entire process into the actual cost of making a good product and we can all see that doing it right in the first place is a good premis, but that finding the faults in the process will lead to real, quantifiable results.



Paul Alexander

FMEA's should not be used own their own. In my experience as an ex-Ford Quality engineer and now a senior engineer at a T-1, the use of FMEA's should be part of a process to make your product as robust to external noises as possible and also to incease the reliability of your product.

The Ford system of :

1. Quality History Analysis (FMA),
2. 'P' Diagram,
3. DFMEA (a. Block Diagram, b. Function Trees, c. DFMEA),
4. DVP,
6. Control Plan

helps supplier and OEM engineers to look at what went wrong in the past and try to fix it for the future. An FMEA must consist of a cross-functional team and also include design, quality and manufacturing representatives.

FMEA's should not be looked at in isolation, but as part of a larger relibility and robustness (systems engineering) program.

Darrell Wenrich

I agree that the prioritized reduction plan is a must. And to set limits prior to the actual composition might be a mistake. The prioritization issue inevitably comes down to resources.
I also agree with Paul that the FMEa needs to be incorportade with other quality functions.
My 2 cents...
Top Bottom