Permanent 10 mph through Bristol TM platforms

[D1009]

There are probably some technical solutions, hopefully one will be viable for the current less than ideal arrangement.

It would be interesting to know whether there are any restrictions currently in place on ARS to prevent through freight trains from being routed via a 10mph platform. I shudder to think how long it would take a freightliner to crawl through the station throat at that sort of speed, particularly if it applies right through the platforms.

I would hope the signallers at TVSC are aware of all this. The through roads are not affected, they are both bi directional and there aren't that many freightliners or long car trains compared with the number of passenger trains.

 

[MarkyT]

I also heard of this issue and surmised it may be part of the justification for the 10mph restriction. There are other safeguards in place too. The space between the signals is not adequate to cover the worst case hangover of the longest noses, so there have been special blue platform markings provided in the disputed area and after every arrival station staff are instructed to phone the signaller to confirm the "blue zone" is clear. In practice I suspect the scenario is actually low risk as even worst case overhang is fairly small and an incoming driver would not recklessly drive into a train in front just because the signal was a metre or so beyond the extremity of its nose, typical driving policies requiring a stop at least some 10 metres back in order to maintain good sighting. If we had formal absolute stop lines or markers in British signalling, that could help manage the risk, but I understand investigations are underway on the best way to move the signals slightly further apart. The problem is that stand-back concerns and sighting issues come into play with worst case train lengths.

 

[MarkyT]

Another thought: With axle counters in use, train detection sections could actually overlap to an extent if two sensors were used. That is assuming they are actually used through the platform area, unlike another recent scheme at Nottingham where axle counters were used generally apart from in the platforms at Nottingham station where miscount problems were anticipated in the long multi-section platforms due to wheels routinely stopping over the sensors (Note this problem always fails safe BTW, leaving a section falsely occupied, but clearly that is an operational inconvenience that poses its own risks in working around). Hence track circuits were provided through the platforms there. A solution to the Nottingham problem with axle counters was proposed and may have been implemented already for the similar case at Birmingham New Street, using additional long supervisory sections that can, when going clear, automatically reset an intermediate miscount (an established method from Germany, the traditional home of axle counters).

 

[D1009]

I don't think it unreasonable to expect a driver in a 10 mph speed restriction to stop short of any obstruction.

 

[alxndr]

Another thought: With axle counters in use, train detection sections could actually overlap to an extent if two sensors were used. That is assuming they are actually used through the platform area, unlike another recent scheme at Nottingham where axle counters were used generally apart from in the platforms at Nottingham station where miscount problems were anticipated in the long multi-section platforms due to wheels routinely stopping over the sensors.

This is what I was about to suggest as a solution. Would have thought it would be cheaper than moving the signals, and less clutter on the platform too.

Do we know whether axle counters or track circuits have been used?

 

[lammergeier]

I have an inside source about this. The 10mph restriction was only imposed very late in the day (far too late to incorporate the impact into timetables for example) and was to mitigate an issue which only came to light very late into the installation.

Failure to mitigate this issue would have meant BASRE could not have been commissioned, or at least require severe loss of functionality at Bristol TM.

Basically, with back-to-back mid platform signals, the block joints are positioned directly underneath the signals. The block joint for one direction is the signal replacement joint for the other. So a train arriving across the mid-signal releases the track circuit in rear once the last axle has passed the replacement joint. Problem is, modern rolling stock have increasingly long nose cones, and it was realised that the last axle could clear the replacement joint and stop with the rear nose / couplers still occupying the section behind. The signalling system would not be able to detect this and could then release the East Gantry / West Gantry junction signal to a single yellow without the near-end platform actually being clear. This could be interpreted as a wrong-side failure.

The mitigation is therefore to impose the 10mph and warn drivers to approach the mid-platform signal at extreme caution as there is a slight possibility that another train could be standing foul without occupying the track circuit. Unfortunately speed signage can’t be made signal-aspect specific so it will apply to all trains regardless of movement. Interesting the comments about whether it applies for the whole length of the platform or not - I had heard it would be to the mid-platform signals only. Either way, this restriction is going to be exceptionally difficult to remove, without redesigning the interlocking or moving the mid-platform signals about 8m apart, which would allow the block joints to be positioned in a more appropriate position.

I must admit, I can’t help looking at this situation and just seeing engineering incompetence and loss of knowledge, although I’m sure this isn’t an entirely fair reflection on the people involved.

The thing is this situaion arises at New St and it is part of a drivers route learning to understand that when arriving on a single yellow it is possible (and it happens often) that the nose cone of a voyager or pendo can be foul of the signal you're stopping at. So it's not like this shouldn't have been foreseen earlier, it's not unique.

We have been briefed that it's now 10 in, 10 out UFN although alterations are expected at some point to allow 25mph running to be reinstated. What these alterations will be and how much it will cost I don't know.

This is what I was about to suggest as a solution. Would have thought it would be cheaper than moving the signals, and less clutter on the platform too.

Do we know whether axle counters or track circuits have been used?

Axle Counters.

 

[Tim M]

Yes I think that was in people's minds. It was very much about standardising the trackside equipment as well, from which we get the pared down minimalist nature of some of the the latest conventional schemes, with lightweight, near zero maintenance 'searchlight' LED signals and long block sections usually requiring almost no intermediate equipment due to highly scaleable axle counters, and the lower overall power requirement of the equipment making a more distributed power and backup architecture more practical. The one thing that stands out (literally) today is the disproportionate bulk of some modern signal structures, reaching out above future wires that may never come, when the simple folding straight post integrated signal is not suitable.

The WESTRACE Modular system is indeed built on just a few standard trackside boxes but has shown to be very flexible in application. The original Crewe Shrewsbury installation was followed by the GN-GE Joint Line with a greater number of level crossings and Reading Train Care Depot. It is also used on Crossrail core signalling in conjunction with the Siemens (Germany) train control system.

The comparison with geographical interlocking’s is perhaps less true as the interlocking data is more akin to free wired relay systems. Data is based on templates, e.g. ladder logic for a signal might use rungs (similar to relay circuits) selected from a number of standard template rungs as required by the Control Tables. New templated rungs are relatively easy to design and once incorporated in the template design tool can be used for both new installations and retrospectively if changes are required to existing installations.

Templating was not available for the earliest WESTRACE (Mk 1) installations on the Central Line, in Oslo, Germany, Australia and Spain starting about 25 years ago but added to the overall design suite about 10 years later. Before templates were available each interlocking had rungs ‘free wired’ based on an agreed data design specification. Modular use the WESTRACE Mk2 system but the data design is tha same as Mk1.

 

[edwin_m]

By contrast the core part of SSI data is effectively built from the control tables for each signal, with a data construct giving the points, tracks, etc on each route and the various actions that need to be carried out during the process of setting and releasing it. As far as I know that doesn't have an analogue in relay interlocking design.

 

[D1009]

The thing is this situaion arises at New St and it is part of a drivers route learning to understand that when arriving on a single yellow it is possible (and it happens often) that the nose cone of a voyager or pendo can be foul of the signal you're stopping at. So it's not like this shouldn't have been foreseen earlier, it's not unique.

We have been briefed that it's now 10 in, 10 out UFN although alterations are expected at some point to allow 25mph running to be reinstated. What these alterations will be and how much it will cost I don't know.

That is a bit of a fudge though, surely almost the first principles of signalling are that a distant in the on position or a yellow aspect indicates that the line is clear to the next signal, otherwise you should get a warning signal.

 

[MarkyT]

The thing is this situaion arises at New St and it is part of a drivers route learning to understand that when arriving on a single yellow it is possible (and it happens often) that the nose cone of a voyager or pendo can be foul of the signal you're stopping at. So it's not like this shouldn't have been foreseen earlier, it's not unique.

We have been briefed that it's now 10 in, 10 out UFN although alterations are expected at some point to allow 25mph running to be reinstated. What these alterations will be and how much it will cost I don't know.

That is a bit of a fudge though, surely almost the first principles of signalling are that a distant in the on position or a yellow aspect indicates that the line is clear to the next signal, otherwise you should get a warning signal.

I understand the solution is likely to be moving the signals slightly further apart. I don't know whether that will entail replacement or additional structures, or whether the existing ones can be adapted. The signals at New Street are further apart already apparently.

 

[WelshBluebird]

One thing I have noticed over the last month or so, and this may or may not be related to the slower line speed, is that at least on the local stoppers to Bath, delays of a few minutes are seemingly much more common now than before the work, and those delays tend to show up as "delayed" on the live departure boards / app more frequency than before too (rather than it showing 1 or 2 minutes late). That may just be my perception though!

 

[D1009]

There have certainly been more problems in general since the resignalling took place, some directly related, some indirectly related and some not related. The speed restriction has certainly limited the ability to recover from them.

 

[MarkyT]

There have certainly been more problems in general since the resignalling took place, some directly related, some indirectly related and some not related. The speed restriction has certainly limited the ability to recover from them.

New systems tend to suffer an initial spike in failures. This is due to a small proportion of rogue components that are only revealed faulty once under the load of normal operations. Once these are found and replaced, the system starts to achieve its target reliability. In railway signalling, failures of this type are almost invariably 'right side' in that they cause signals to go red, points to not move to commanded position or fail detection or train detection to falsely show occupied, so while they are tedious they are generally safe. Unsafe components and design mistakes are debugged prior to commissioning by exhaustive testing procedures. The early failure peak is part of a phenomenon known as the 'bathtub curve' where peaks in failure statistics occur at the immediate beginning and towards the end of a system's lifespan.

 

[Dai Corner]

New systems tend to suffer an initial spike in failures. This is due to a small proportion of rogue components that are only revealed faulty once under the load of normal operations. Once these are found and replaced, the system starts to achieve its target reliability. In railway signalling, failures of this type are almost invariably 'right side' in that they cause signals to go red, points to not move to commanded position or fail detection or train detection to falsely show occupied, so while they are tedious they are generally safe. Unsafe components and design mistakes are debugged prior to commissioning by exhaustive testing procedures. The early failure peak is part of a phenomenon known as the 'bathtub curve' where peaks in failure statistics occur at the immediate beginning and towards the end of a system's lifespan.

I'd imagine there was a loss of local knowledge when the work moved to the TVSC too?

 

[lammergeier]

New systems tend to suffer an initial spike in failures. This is due to a small proportion of rogue components that are only revealed faulty once under the load of normal operations. Once these are found and replaced, the system starts to achieve its target reliability. In railway signalling, failures of this type are almost invariably 'right side' in that they cause signals to go red, points to not move to commanded position or fail detection or train detection to falsely show occupied, so while they are tedious they are generally safe. Unsafe components and design mistakes are debugged prior to commissioning by exhaustive testing procedures. The early failure peak is part of a phenomenon known as the 'bathtub curve' where peaks in failure statistics occur at the immediate beginning and towards the end of a system's lifespan.

But that doesn't account for the fact that the 10mph restriction was never part of the initial design. The new set up was intended to be 25mph throughout with the 10mph restriction a very late addition. So if the new set up isn't performing as expected or planned it should be no surprise and, whilst some aspects may improve, full reliability and functionality will probably only be achieved once the design is altered to enable 25mph running to be resumed.

 

[MarkyT]

But that doesn't account for the fact that the 10mph restriction was never part of the initial design. The new set up was intended to be 25mph throughout with the 10mph restriction a very late addition. So if the new set up isn't performing as expected or planned it should be no surprise and, whilst some aspects may improve, full reliability and functionality will probably only be achieved once the design is altered to enable 25mph running to be resumed.

The speed restriction is really a different issue to system reliabilty. It's a particular pain if say one the bathtub's early component casualties takes a platform out of use temporarily and its traffic has to be diverted to an alternative. The restriction adds time to all movements so there's less margin between planned moves to accomodate any such ad hoc diversions, more chance of knock on delays, less opportunity to recover. Let us hope they solve the problem quickly.