[fm-discuss] FITrate description?

[Steve Hanson]

Hi Lei,

> Hi experts,
>        I read the "Eversholt FaultTree Description Language" 
> manual(version 1.8). Several properties are applicable to fault event, 
> which are listed in table 2-1. All the properties except FITrate are 
> explained in detail underlying the table.

You're right, it does seem to be missing. I'd better fix that.

> Then what does the FITrate property mean? For example, specifically, 
> what is the effect of FITrate=10 in the event declaration of " event 
> fault.power.input_short at acside/pdb, FITrate=10;"? Forgive me for the 
> maybe silly question, I can't find a place explaining it.

Basically it tells you the relative frequency at which faults of a 
particular class are expected to be seen on a particular resource.

It is only used for multi-entry suspect lists to allow calculation of 
relative probabilities for each suspect in the suspect list. So if you 
have two suspects, one with a FITrate of 100 and one with a FITrate of 
300, then the "certainty" fields added to the suspect are 25% and 75% 
respectively (as the latter is deemed 3 times more likely to occur). 
Note again that the FITrate fields are only relative to each other. If 
no FITrate property is present then all suspects in the list are given 
equal probability.

Steve

>
>
> Thanks,
> Lei Chen
>
> _______________________________________________
> fm-discuss mailing list
> fm-discuss at opensolaris.org