Failure Rate Data for 61508 / 61511 1

[southerton]

Does anyone know of any good sources of generic failure rate data on the web e.g. Thermocouple, DP transmitter, Relay etc.

I am trying to calculate SIL levels for a safety system and would like to try out a few manual calcs before I commit to buying a software package with built in database.

 

[shobi]

I think your best bet would be to ask your manufacturer. For most of their product Emerson (Rosemount) have a 3rd party evaluation on their website.

 

[sdl]

southerton:

There are many sources of failure data. The manufacturers are one source. I would recommend Faradip as that is a good source. Others include OREDA, exida, AIChE.

I have evaluated many different SIL calculation packages and have decided that it is best to go with any package with an industry compiled database (i.e. Faradip).

sdl

 

[jsummerfield]

Operating companies experience the failures more than the manufacturers. Perhaps a furnace flame detector manufacturer predicts 600 years between failures - because they shipped 600 units and only one filed a warranty claim that year.

John

 

[RobV]

jsummerfield,
basically you are correct, but if the vendor has more than 10000 logged operational years, and only 1 warranty call ... then statisically the accuracy of the prediction would increase!
I'd recommend to talk to the vendors with ample operational years of experience that have taken the trouble of monitoring their products performance.

 

[sdl]

RobV:

Not all failures are reported back to the manufacturer and this can lead to optimistic assumptions. The best failure data is one from an operator who has logged the failures, maintenance, etc. that can be accumulated and compiled. In all cases, the more conservative approach is best unless you have great confidence in the data.

In your example, they may have had 1 warranty call, but there may be 10 or 20 problems that people either fix themselves or have "worked around". In some cases total failures are replaced after the warranty period, etc.

Jsummerfield is absolutely correct. Manufacturers will usually provide more optomistic numbers.

If no other data is available (operator or industry databases) then we have no choice but to use the manufacturer's data (again, with caution).

sdl

 

[RobV]

sdl:

Fully agreed sdl. southerton is referring to SIL so he is basically adressing safety. If components that make part of the SIL loop fail, I believe that they HAVE to be reported to the manufacturer as the operators log will have to show the cause of the failure!
In Germany, reporting of failures has even been mandated by law for their overpressure system. This means that the manufacturers that have supplied these systems in the past have ALL reports of ALL failures. If not, operators may have operated outside the law ...
This gives to my opinion, an accurate database and failure rates.

But all in all, I agree, if you can't find certified data, always be conservative.