Detection Rating in Design FMEAs (DFMEA)

[Geoff Lee]

Detection Rating d-FMEA

I am implementing FMEA in design and development of appliances. I find Occurence ratings to be fairly self-explanatory. Also, Severity ratings seem to be described in understandable terms; degrees of performance degradation, operability, dissatisfaction of user, hazard etc. These enable a group to fairly readily reach consensus on ratings.

However, most Detection rating scales are less helpful. They tend to have a few distinctive descriptions at the extremes of the scale, for example 1="can be corrected prior to prototype", 10="None" etc. However, for the main part, they seem to have a sliding scale from, for example, 8="lowest effectiveness" to 2="highest effectiveness". Such sliding scales do not lead to ready consensus amongst an FMEA team. Nor, as a leader, can I provide effective guidance on where a particular method might lie on such a scale.

It seems to me that a Detection method should have several characteristics:
1 ability to detect whether the failure mode does occur
2 verify what causes it
3 determine its level of occurence (i.e. provide a degree of confidence in the Occurence rating)

So, I would hope that a set of Detection scale descriptions could be generated which might be framed in these, or similar, terms. For example, we might have:
1="Proven detection of failure mode from all causes, with accurate measure of occurence"
5="Detects failure mode in most cases, provides good guidance on occurence"
...

Does anyone know of any such Detection scales or have any comments on this issue?

Thanks,
Geoff

 

[chalapathi]

I am implementing FMEA in design and development of appliances. I find Occurence ratings to be fairly self-explanatory. Also, Severity ratings seem to be described in understandable terms; degrees of performance degradation, operability, dissatisfaction of user, hazard etc. These enable a group to fairly readily reach consensus on ratings.

However, most Detection rating scales are less helpful. They tend to have a few distinctive descriptions at the extremes of the scale, for example 1="can be corrected prior to prototype", 10="None" etc. However, for the main part, they seem to have a sliding scale from, for example, 8="lowest effectiveness" to 2="highest effectiveness". Such sliding scales do not lead to ready consensus amongst an FMEA team. Nor, as a leader, can I provide effective guidance on where a particular method might lie on such a scale.

It seems to me that a Detection method should have several characteristics:
1 ability to detect whether the failure mode does occur
2 verify what causes it
3 determine its level of occurence (i.e. provide a degree of confidence in the Occurence rating)

So, I would hope that a set of Detection scale descriptions could be generated which might be framed in these, or similar, terms. For example, we might have:
1="Proven detection of failure mode from all causes, with accurate measure of occurence"
5="Detects failure mode in most cases, provides good guidance on occurence"
...

Does anyone know of any such Detection scales or have any comments on this issue?

Thanks,
Geoff

Basically, all design validations and verifications come under detection methods. Detection method is one which will catch the error/mistake of the designer. The Detection rating (1-10) is about how effective the detection method with the current detection methods. For example, if the designer has selected a wrong material for a component. It is lead to reliablity failure in the long-run. How effective are the validation trial to detect this mistake. In the prototype stage you may be running for let us say 50 hours. It can catch some limited mistakes. May be this wrong selection of material may fail after 100 hours. There is a remote chance of detecting this mistake.

Detection rating will be less, if the effectiveness is very high and 10 if the mistake can not be detected in the validation stage.

This is my experience about the detection rating. You can develop guidelines based on this.

 

[Michael Walmsley]

The AIAG table for Detection is meant as a guideline only.

What you need to do is tailor the table to suit your processes.

See the attached as an example.