Are track faults really track faults?

[ozbob]

This is something that has been getting me quite confused of late.  I have been a user of railways for all my life.  I have constantly used trains from my earliest memories (3 years or so, hey how can you forget bouncing around in the cab of an R class at 70mph, anyway ..) and regularly rode the Melbourne suburban system almost daily.  I struggle to recall any delays due to 'track faults'.  In fact I cannot remember much in the way of delays (I am sure there must have been some ) but it is something I have no memory of.  I recall once being in a steam hauled passenger service that failed north of Melbourne  and being pushed by a suburban train, but not much else.  I subscribe to the Metro twitter feed --> http://twitter.com/#!/metrotrains  it is almost a constant stream of 'track failures'.  Here in SEQ 'track failures' happen with regularity. Are the so called 'track failures' actual track failures or are they just breakdowns in electrical circuits and so forth?  It seems that the track is not actually failing but with the CTC etc. a minor issue with loss of a circuit continuity is a 'track failure' and all stops until the issue is checked.

Anyone know what is actually going on?  There are enough problems with 'police incidents', ill pax, errant behaviours at level crossings, other signal, power and points failures let alone the constant background of track failures.  Electronic systems that are meant to be better are actually failing at a much higher rate than failures were evident in the days of signal boxes and mechanical interlocking it seems to me.

Is there some way of building a redundancy into the circuits? It seems an obvious thing to do.  Anyone able to shed some light on 'track failures' please?

[Derwan]

I can't shed any light on it, but "track fault" and "signal failure" appear to be interchangeable.  Sometimes an eRail update will say one and the TransLink website will say the other.

[ButFli]

Is there some way of building a redundancy into the circuits? It seems an obvious thing to do.  Anyone able to shed some light on 'track failures' please?

I think the problem with built-in redundancy is that signalling systems are designed to be fail-safe. Any hint of a fault and every affected signal is set to danger. It would be entirely possible to build a system that kept going after a failure but there is a risk that two signals could be set to proceed, causing a crash.

There might have been fewer delays caused by signal faults in the old days but there were certainly more crashes caused by them.

[ozbob]

Thanks ButFli.  I agree signals are designed to fail safe.  Signal failures seem to be relatively rare compared to 'track failures'.   I am trying to establish what actually are the track failures?  Signal failures and point failures seem to be reported as such.  What is the bulk of the track failures?

[BrizCommuter]

Thanks ButFli.  I agree signals are designed to fail safe.  Signal failures seem to be relatively rare compared to 'track failures'.   I am trying to establish what actually are the track failures?  Signal failures and point failures seem to be reported as such.  What is the bulk of the track failures?

On London Underground it is usually either a broken rail, or track circuit failure (metal bridging gap between adjacent track circuits).

[ozbob]

Thanks,  I expect it is something similar here as well.  I think disruptions at joins may disrupt track circuits (only a guess).  See what is posted ..

[mufreight]

Track faults can be any of a multitude of problems, ranging from a broken rail/failed weld on a joint through to loss of continuity of the track circuts. signal failures can be as a consequence of a track failure but usualy refer to an actual failure of a signal which does not respond to the comands from control and give the appropriate indication, causes anything from a relay failure, loss of power on.

[Golliwog]

Just out of interest, what is QR's policy on an actual broken rail? I remember being told when I did a course on railways that a single break in a rail isn't a massive problem and you can usually still run trains over it, as the adjacent sleepers will hold the rail in position. Once a break is detected, will QR stop services until the break is fixed, or will they continue to run services over it and have it fixed ASAP?

[HappyTrainGuy]

I can't say 100% but I think it depends on how badly its broken and where it is (On a curve/straight part of track/sleeper type). They'd reroute trains to another line if available and do an inspection where they could butt weld it with a speed/weight restriction, adding a joint bar right up to cancelling all movements for replacement asap.

[Golliwog]

Cool, fair enough.

[mufreight]

Just out of interest, what is QR's policy on an actual broken rail? I remember being told when I did a course on railways that a single break in a rail isn't a massive problem and you can usually still run trains over it, as the adjacent sleepers will hold the rail in position. Once a break is detected, will QR stop services until the break is fixed, or will they continue to run services over it and have it fixed ASAP?

With the axle loads of the electrics a broken rail now means a stop at least untill a clamp type rail joint can be installed, then services can be resumed subject to a speed restriction until such time as the rail can be replaced or welded.
With most of the present signaling equipment a broken rail generaly means a signal failure due to the loss of continuity with the track circuts.

[p858snake]

does the "track fault" keyword have any effect on the "on time" stats?

[colinw]

I think the problem with built-in redundancy is that signalling systems are designed to be fail-safe.

Yes, that's how they are engineered. I work in signalling & ATP (currently working on Communications Based Train Control (CBTC) for a Chinese metro).  Our systems have a huge amount of redundancy in them (often two-out-of-three voting), and are engineered to avoid a "wrong side failure" at all costs. (A "wrong side failure" could be an unsafe output state in an ATP system, in a signalling system it could be an incorrectly displayed proceed aspect, whatever is deemed unsafe for the application being developed).  E.g. in every ATP system I have worked on the DEFAULT state of the system is to assert a request for application of the emergency brake, the mantra being "a stationary train is a safe train".  On ANY kind of error or inconsistency, the first instinct of the system is to go to brake application.  The system effectively works as an "emergency brake hold off", where any system failure leads back to the default state of brake requested.

One effect of this kind of architecture is that systems tend to cascade fail from fairly minor errors through to "everything stop, NOW!" quite rapidly.  I have lost count of the hours I have spent tracking down obscure corner cases which can cause such behaviour to emerge due to relatively minor faults.  Particularly in the multiply redundant systems these kind of errors can be really hard to reason about, often involving timing effects (redundant system A saw something 20 milliseconds before its peers B & C did, and thus took a different processing path).

I think it is true to say that the modern generation if digital control & communications systems are wonderful when they are working, but a real pain when they fail, and often time consuming & costly to rectify.  I suspect that in the digital age the railway is much less robust & tolerant of localised failures than it was in the days of mechanical interlocking, local control panels and signal cabins.  Certainly many of the "old timers" here express opinions about that being the case, and the history of Melbourne & Sydney in the 1920s both managing headways that would be unthinkable today, with very high ontime running, tends to support that.

When I look at current & future projects, one major trend is towards increasingly hard line specification of system "availability" to try and deal with these problems.  I don't know about our competitors, but the company I am working for is increasingly looking toward technology crossover from avionics, where system availability & robustness is paramount (because on a plane you can't just bung on the brake and say "we're good").  Of course, this all comes at a price, and quite a high one.

[ozbob]

I suspect that in the digital age the railway is much less robust & tolerant of localised failures than it was in the days of mechanical interlocking, local control panels and signal cabins.  Certainly many of the "old timers" here express opinions about that being the case, and the history of Melbourne & Sydney in the 1920s both managing headways that would be unthinkable today, with very high ontime running, tends to support that.

When I look at current & future projects, one major trend is towards increasingly hard line specification of system "availability" to try and deal with these problems.

Thanks Colin, for that informed post.  Yes, Flinders St in its heyday was amazing in terms of trains and pax numbers.  Manual signalling, time-keeping and so forth, and achieved much better throughput than today by a long shot.  It was also facilitated by the door configuration on the Tait sets.


http://www.brownfam.com.au/ROLL29/ROLL29_files/image001.jpg

 It is encouraging to read that there is a trend 'one major trend is towards increasingly hard line specification of system "availability" to try and deal with these problems ..'

I am sure it must be frustrating for the operators (train crew, control and repair staff) as well as pax.  I do wonder though by just saying there is a 'track fault' that some folks think the track is really bad.  When in actual reality the track itself is generally fine, it is circuits etc.  Might be better to say there is a 'circuit failure' when that is the case.

=============

From Film Australia

More than a million train passengers a day go through the gates at Flinders Street Station, Melbourne. Claimed to be the largest one station traffic in the world. Control of passengers and freight trains, in and out of Flinders Street and on time, is a masterpiece of precision planning. Made by National Film Board 1953.

[Derwan]

Multiple "track faults" this week.

Could the colder weather be causing it?

[ozbob]

Nope, the planets are in a straight alignment ...  

You know, everytime I look at a picture of a Tait set I get 'tears in my eyes' ....   what a fantastic train .... I did so enjoy riding in them, particularly in the heat waves with all the doors open as we flew along the 5'3" ...

[Golliwog]

Multiple "track faults" this week.

Could the colder weather be causing it?

It could, but I didn't think the temp. has changed that drastically. I know other railways change the tension in their rails depending on the season to allow for expansion/contraction to occur without causing any issues. I'm not sure what QR does.