Why is the FMEA RPN = S*O*D?

Q

qq130230416

Dear all
i have a question, why RPN is S*O*D? what is meaning for this formula?:thanks:
 

Statistical Steven

Statistician
Leader
Super Moderator
Dear all
i have a question, why RPN is S*O*D? what is meaning for this formula?:thanks:

S = Severity
O = Occurrence
D = Detection

Severity is how "bad" the injury would be if there was a failure (1=none, 10=death)

Occurrence is how likely is the defect to occur (1=unlikely, 10=often)

Detection is how likely are your controls to detecting the defect should it occur (1 = Always detected, 10 = never detected)

You rate each failure mode on a rating of 1 to 10 for each of the three. Multiply the ratings to get the RPN for the failure mode.

A flaw with this rankin is that a failure mode with very low severity, high occurrence with low detection would score an RPN of 100, while a failure mode with high severity, low occurence and low detection would also score a 100. I favor a rating system that weights the severity more than detection.

:2cents:
 

antoine.dias

Quite Involved in Discussions
A flaw with this rankin is that a failure mode with very low severity, high occurrence with low detection would score an RPN of 100, while a failure mode with high severity, low occurence and low detection would also score a 100. I favor a rating system that weights the severity more than detection.

In the new version of the FMEA manual (AIAG) they present SOD as an alternative way to the RPN ranking
Example :
S= 9 / O= 4 / D= 5 would mean RPN= 180 but SOD = 945
S= 4 / O= 9 / D= 5 would mean RPN= 180 but SOD = 495
S= 5 / O= 4 / d= 9 would mean RPN =180 but SOD = 549

In the past years SOD is applied in several companies I visit, just for the reason you stated.

best regards,

Antoine
 

Statistical Steven

Statistician
Leader
Super Moderator
In the new version of the FMEA manual (AIAG) they present SOD as an alternative way to the RPN ranking
Example :
S= 9 / O= 4 / D= 5 would mean RPN= 180 but SOD = 945
S= 4 / O= 9 / D= 5 would mean RPN= 180 but SOD = 495
S= 5 / O= 4 / d= 9 would mean RPN =180 but SOD = 549

In the past years SOD is applied in several companies I visit, just for the reason you stated.

best regards,

Antoine

My only concern is how does the AIAG deal with a situation of the following?
S= 9 / O= 1 / D= 1 with RPN= 9 but SOD = 911
S= 8 / O= 8 / D= 8 with RPN=512 but SOD = 888
 
Q

qq130230416

S = Severity
O = Occurrence
D = Detection

Severity is how "bad" the injury would be if there was a failure (1=none, 10=death)

Occurrence is how likely is the defect to occur (1=unlikely, 10=often)

Detection is how likely are your controls to detecting the defect should it occur (1 = Always detected, 10 = never detected)

You rate each failure mode on a rating of 1 to 10 for each of the three. Multiply the ratings to get the RPN for the failure mode.

A flaw with this rankin is that a failure mode with very low severity, high occurrence with low detection would score an RPN of 100, while a failure mode with high severity, low occurence and low detection would also score a 100. I favor a rating system that weights the severity more than detection.

:2cents:
thank you so much.
i am sorry. my english is poor so i can not express clearly.
i mean why not RPN≠S*O or S*D? what does S*O*D mean in statistics?
:thanx:
 

Jim Wynne

Leader
Admin
thank you so much.
i am sorry. my english is poor so i can not express clearly.
i mean why not RPN≠S*O or S*D? what does S*O*D mean in statistics?
:thanx:

They mean nothing in statistics--they are relative values, not absolute, and the scaling is mostly arbitrary. In other words, the idea is that there is a consistent rating scale, and it should be usable in terms of deciding which are the most important things to improve or be aware of.
 

Bev D

Heretical Statistician
Leader
Super Moderator
Jim has it correct - these are really arbitrary rankings. They have no statistical validity or mathematical validity or science validity. There was a desire to try to quantify ratings rather than use a qualitative ranking.

Some people do use S*O
I haven't heard of anyone using S*D

A 'recent' development is to use a matrix of Severity by Occurrence that isn't mathematically derived.

Another development primarily from the medical device community is to use just the severity and allow the testing to determine the occurrence rate.
 

julsbear

Involved In Discussions
Jim has it correct - these are really arbitrary rankings. They have no statistical validity or mathematical validity or science validity. There was a desire to try to quantify ratings ratehr than use a qualitative ranking.

Some people do use S*O
I haven't heard of anyone using S*D

A 'recent' development is to use a matrix of Severity by Occurrence that isn't mathematically derived.

Another development primarily from the medical device community is to use just the severity and allow teh testign to determine the occurence rate.
Bev:

Do you have any links or references on the matrix of Severity by Occurrence?
 
Top Bottom