fault and failure

[25362]

Kindly describe these terms from the viewpoint of process and plant safety.

 

[JOM]

Defining terms is a funny business. It's all about communication and that means dealing with human perceptions.

That said, I would say "fault" is a condition that is outside the design intent.

I would say "failure" is the event of the fault coming into existence. How does that sit with your ideas?

Dictionaries do not help in cases like this.

John.

J.

 

[25362]

That's precisely the reason for my post.
Would you say they are as it were "cause-and-effect" related events ?

Or would you say that failure implies a loss of performance, while fault doesn't, and it would better mean a loss of some kind of redundancy ?

And, if you please, I'd like to understand the difference between fail-safe and fault-tolerant.

 

[JOM]

>Would you say they are as it were "cause-and-effect" >related events ?

No, not as I described them above. I've said that if an item has a fault, it has also failed. That's two ways of describing the same thing - no cause-and-effect relationship.

Now, that was only my first thoughts about these terms.

I guess sometimes we would see an item with problems as "faulty", but if it still does its job then it hasn't "failed". My TV has a poor image quality, so it has a fault. But it still works, so, to me, it hasn't failed. Someone more demanding might say it has failed.

>Or would you say that failure implies a loss of >performance, while fault doesn't,

Well, that's the reverse of my TV example. Are you thinking of a hidden fault - no apparent symptoms.

What would you say about a pressure relief valve stuck because of corrosion? It will not do its job, but does not interfere with poduction. Has it "failed" or it does it have a "fault"?

>and it would better mean a loss of some kind of >redundancy ?

I guess redundancy compensates for a failure - not always successfully.

>And, if you please, I'd like to understand the difference >between fail-safe and fault-tolerant.

We say a device is fail-safe if, when it "fails", it produces a safe state rather than an unsafe one. That's not very helpful, is it? You can only design fail-safe for specific failures - loss of air to a valve, the valve moves to the "safe" position. What about the PRV jammed by corrosion? That isn't fail-safe.

I guess "fault-tolerant" means the device can still do its primary function even in a deteriorated condition.

Some people will say these fine distinctions things don't matter, but definitions like these can be misapplied.

We use many terms with an assumption that we all know what they mean exactly, and that we all have the same understanding of their meaning.

This is not so. If your plant failure leads to litigation (court action) then you'll find the lawyers will apply a completely independent interpretation of terms.



J.

 

[JOM]

Thinking more about fault and failure -

I would say if a device is not able to perform one of its design functions, then it has a fault.

With "failure" I think we have to consider the degree of failure. There is partial failure and complete failure. That seems important.



J.

 

[25362]

If I interpreted your post correctly, could one say, in general terms, that a fail-safe situation involves a plant shut-down for repairs, while a fault-tolerant event, probably because of redundancy, would enable to rectify the difficulty in-line, without being obliged to stop production ?

 

[JOM]

>that a fail-safe situation involves a plant shut-down for >repairs,

Why do you connect fail-safe with a plant shutdown. I thought fail-safe design related to failure of items during production.

Boom gates at a rail crossing are held up by electrical power. If the power fails the booms fall under gravity. That's fail-safe. In process industries, I suspect it's not always easy to define the "safe" state an item should go to in the case of failure.

>while a fault-tolerant event, probably because of >redundancy, would enable to rectify the difficulty in->line, without being obliged to stop production ?

I think there is sense in that. If a pump fails and there is a backup pump then that can be brought on line. So the system is tolerant to failure.

Can I ask the reason behind your concern with the meanings of tehse terms? Just general interest or some specific task you are working on? I'm interested in the way technical terms are misused by non-technical profession (law, media, academics).



J.

 

[25362]

JOM: I was just trying to revise an "internal" technical manuscript by a friend of mine. Thanks for your time and detailed answers.

 

[rosiethe2]

The NUREG-0492 Fault Tree Handbook defines failures as basic abnormal occurrences, whereas faults are "higher order" events. The handbook has several examples and makes a good case for the differences between faults and failures.