(27-08-2010, 10:25 AM)jenni.joseph9 Wrote: Hi,
I tried the fault tree for points and came out with the attached output.
Can anyone please check the attachment and let me know the suggestion/corrections, if any, i need to implement in that?
Awaiting the comments.
Regards.
"The fault tree for points" is somewhat vague; you haven't said what particular failures you are considering.
Your top event is "points fail to operate" which I would interpret as failure to move at all, and would not include the failure mode of the points moving though without actually achieving detection at end of stroke. It certainly doesn't cover the invalid achieving of detection when the switch rails are not in the correct position, nor the mechanical collapse that might cause a derailment as a result of the switch rails being held firmly in position. So I will judge on this basis yet I warn that depending on the question actually set then it might be a very limited answer.
You have basically copied across the NO POWER and BAD CONNECTION sections from other fault trees. Hence these areas are fundamentally ok, but probably not what I'd have given great priority to in the context of points. The fact that you have written "head transformer" and refer to "650/110T" definitely doesn't suggest to me that you have given it a lot of thought or perhaps you don't have much knowledge of points. Either way it fails to impress me- I just don't get the feeling that you really understand points and how they can fail (which is what you are trying to demonstrate when answering the question).
If we now look at the third leg, the heading looks a jumble of different elements-
points not set properly,
points not locked,
points not detected.
These are actually good, but they needed to be kept separate and indeed these should have separately fed into the generic
points failure top event on the same level as the
points fail to operate .
By lumping "chalk and cheese" together you can't hope to show how the various faults can lead separately or in combination to the particular failure mode.
To take an example, "rail creep" was a good thing to consider but your diagram does not show whether you claim it will result in loss of detection or loss of locking; since this is a fundamental reason for producing a fault tree then (if you excuse the pun) you appear to "have missed the point".
In a Fault Tree the specific faults for which a failure rate can be estimated / supplied from historic data are at the bottom and shown in the circles. Where there is a "gate" which gathers a number of "strands" into something common that can then be considered as an input to a higher level of the tree then there can be a description written above it if there is not room in the "gate" itself. For example one of the causes of a point failing to operate is NO POWER and that itself can arise from several lower level causes such as 650V DISTRIBUTION LOST or NO OUTPUT FROM LOCAL TRANSFORMER or BLOWN FUSE IN SPECIFIC CIRCUIT. A lower level of the tree can be added to consider the things that could result in each of these, and so on and so on until it is not worth going down any further for the particular top event being considered.
You however have put sentences of text into these boxes, there are some good ideas here but instead of being in this box they should often have been shown as separate feeds into a "gate" that you haven't depicted. In the case of "stretcher bar" you have lumped together dissimilar faults- the breakage of the stretcher bar, the deterioration of the insulations within the stretcher bar. The fact that both involve the stretcher bar is irrelevant; you should be considering the various ways in which the points could cause the track circuit over them to fail and as a result be deadlocked by the interlocking and this could certainly include this insulation failure as one of severlal causes (others might include a discarded fizzy drinks can shorting out between rails of opposite polarity in close proximity, the burring over of the rails with a broken end plate at an IRJ within the S&C). You would include the fracture of the stretcher bar completely separately- the failure rate is different, the effect is different; indeed the failure of any one stretcher isn't likely to immediately cause a failure and so here is an opportunity to incorporate som ANDs as well as just ORs in the combinational logic, whereas it only takes one losss of insulation to cause the track failure.
In summary, there is a big difference in the consequences from faults which might result in:
a) points not being able to throw, but trains can be signalled over them normally in one lie,
b) points not giving detection at all and thus holding relevant signals at danger, necessitating hand-signalling but at least being safe,
c) points mechanically jamming midstroke such that an operator is unable to operate manually making it impossible to pass any traffic at all,
d) giving detection when the switch rails aren't quite correctly set, perhaps resulting in derailment
e) points with both switch trails against respective stock rails leading almost certainly to derailment and quite likely also a collision.
You may well understand all this; however unfortunately your answer did not get that message across; it is your responsibility to do this.
