Fault Tree DC Track Circuit

[PJW]

I have been sent the Fault Tree attachment to comment upon.

1. It is good that right side and wrong side failures have been separately considered

2. Need to be careful to distinguish between AND and OR; all seem to have been drawn as ANDs, whereas most at least should have been ORs

3. Conventionally the failures are drawn at the top of the sheet of paper and the various faults in a heirarchy below; this presentation reads left to right- it isn't wrong but probably best to do conventionally.

4. I liked the portion of the tree which leads to NO POWER (although where it said "head transformer" it presumably means "feedset transformer")

5. Other elements should have been broken down in this manner rather than simply all summating all at one node. For example it would have been easy to have separated the faults:
a) leading to high resistance between feed and relay,
and
b) leading to excessive leakage diverting current from track relay
and
c) opposing voltage from neighbouring track imposed due to failed Insulated Rail Joint
any one of which would cause insufficient pick up current.

6. Then for example item 5b could then have been further broken down into the various causes of poor ballast resistance (flooding, contamination, failed rail insulations / pads, point stretcher insulations),

7. Could have considered also the TR failing to pick because it had become open circuit, mechanically jammed. Also it could have high resistance contact or line circuit to TPR in interlocking itself having a fault wheres the actual track circuit on site was working fine.

8. In summary I think that you had a reasonable number of failures of different types, but you should have presented with groupings and sub-groupings to get a more heirarchical structure. The idea is that you can "drill down" from the top event and get more and more levels of detail, with a range of different types of faults then at the next level down a range of different ways in which that particular type of fault can occur from different lower level causes.

Also it would have been good to have demonstrated some combinational faults.
e.g. track set up in wet conditions with too low a drop shunt
AND
ballast resistance gets very high during hot weather
GIVES
train shunts insufficient current to cause TR to drop.

Similarly you could have assumed duplicated rail leads from location to rails. Cable can fail and track carries on working ok, but if this situation is not revealed by routine testing or chance inspectiion and then the other lead fails, then the track will fail.

I note you had as a fault "pipeline crossing railay line"; by itself this is not a FAULT. It is a relevant factor so you need to combine it with some act or omission that turns the result into a fault.

9. Would have been good to have demonstrated that you knew the role of a Fault Tree is to be able to feed in numerical estimates of probabilities of the various fault events at the roots of the tree, to be able to do some probabilistic number crunching to give estimates of probabilities for the various branches of the tree.

10. Overall it would probably have been aound a Pass- certainly have the general idea both of Fault Trees and Track Circuits but not quite there; presents a picture but it is a little "out of focus"

==========================================================================
I assume that you have found the previous discussion of track circuit fault tree but did you download
a) the attachment for the signal lamp out (to look at the presentation) and
b) read the IRSE News article?
These should help

[jenni.joseph9]

Hi,

I tried to have a go at the Track Circuit FTA as per ur comments.

Attaching those sheets and I could nt find a scenario where I can apply AND gate.

Can you please have a look and let me know the comments.

Thanks in advance.

Regards

[PJW]

Hi,

I tried to have a go at the Track Circuit FTA as per ur comments.

Attaching those sheets and I couldn't find a scenario where I can apply AND gate.

Can you please have a look and let me know the comments.

Thanks in advance.

Regards

You can introduce an AND gate when you think of something that will create a failure, but not all by itself.
For example:
a) the track insulations may be quite bad but the track circuit will still work fine UNTIL it rains and the ballast gets wet
b) a double tail track circuit in a non-electrified area has a pair of IRJs approximately opposite each other separating it from the adjacent track. Should one IRJ fail the two tracks will continue to operate properly but just share a common rail (as they would anyway in an electrified area where there is a traction return rail); however should the second of the pair of IRJ fail then at least one (and probably both) of the tracks will fail
c) where the track leads are duplicated, if one gets severed then the track continues to operate almost normally; if however the second one also fails before the first is replaced, then the track fails
d) if there are rusty rails then a lightweight vehicle might be undetected and the track circuit Show Clear When Occupied- a wrong side failure. However a longer heavier train is most unlikely to be lost; indeed the probablility of a lightweight train being detected is significantly increased when it is stationary compared to when it is moving

The things that I liked about your Fault Trees
1. Complete separation of the rightside and wrongside failure scenarios
2. The initial breaking down of the "Track circuit drops" into constituent high level causes: No Power at Feed / Insufficient current for relay to pick / open circuit between feed and relay.

The things that were not so good:
1. Whereas considered what might cause a 110V fuse to blow in a lot of detail, rather glossed over the feedset and relay themselves
2.Not sure what relay contact you are envisaging could have become high resistance that would cause track relay to drop. I guess that you are not assuming cascade track circuits (and if so you should have explained) so the only contact would be of the TR itself that would then cause the TPR to be down- this is a reasonable fault mode to consider but would have to be incorporated higher up your fault tree as it is an alternative fault than the track relay itself being down (yet the interlocking and signaller would see it ias a track circuit failure)
3. It is generally best not to combine too many legs into the same "gate" but to split into two or three at each stage, gradually getting into more detail as progress down the page.
The idea of the Fault Tree is to show the relationship between various faults and the failure which can result and indeed be able to plug in numeric valuesat the foot of the tree to calculate the probability of the failure at the top.
4. I didn't understand why "tail cable severed" fed directly into the top event rather than being counted as a "bad connection"
.
========================================================================
By coincidence Andy Witton produced a Fault tree for use at Derby last weekend for the rightside failure scenario; I'll add here for comparison

[jenni.joseph9]

Hi,

Thanks for your corrections/suggestions and for the attachment. The attachment is really helpful for comparison.

Thanks again.

Regards