Moving vs Fixed Block signalling

[Deleted]

There's a reference to this question on one of the Victoria line threads. Since I'm a mere customer my knowledge of this is limited to a newspaper article I read a long time ago, about the WCML I think, which left me with the impression that whilst "moving block" would increase line capacity there were safety concerns.

What I don't understand is how "moving block" can help much on an urban transit system like the London Underground since presumably trains spend a significant amount of their time at station stops (by comparison with a line like the WCML, anyway). How much extra capacity could it inject?

My apologies if this has all been thrashed out before.

[Deleted]

It allows trains to get closer together without having to slow down. Trains being put closer together means more trains can run which means more capacity with means happy punters!

Sam

[Deleted]

Moving block signalling systems can allow trains to run closer together, which in turn improves the station roll out run in time (RORIT) - the time it takes for a train to depart a station and for the next train to arrive. This in turn increases the lines capacity. Lets looks at some RORITs:-
Central Line's fixed block signalling with ATO - 60secs RORIT with 130m trains
Northern Line's fixed block signalling with manual driving - 60secs RORIT with 103m trains.
District Line's old multiple home fixed block signalling - theoretical 52sec RORIT with ?100-130m trains.
Paris Line 14's moving block signalling - 40secs RORIT with 90m trains.
Madrid's Line 6 moving block signalling (being installed) - planned 40sec RORIT with 110m trains.

As you can see, moving block signalling can improve the RORIT by nearly 20secs. Taking into account the Victoria Line's newer 133m trains, the RORIT with moving block would be approx 45secs. With fixed block it will likely be 60secs (although I havn't heard the official figures). This extra 15secs, could increase the line's capacity from 30tph to 34tph, or 33tph to 38tph!

I don't know what safety concerns you mention? Moving block signalling has an immaculate safety record. There is always a safe braking distance (taking into account a certain % of braking failure) + safety distance (based on positioning inacuracies), between each train.

[Deleted]

The entire purpose of moving block is to remove the primary weakness of fixed block signalling. That weakness is the fact that the blocks are indeed fixed; a train moving slowly occupies more blocks than a train moving quickly, and thus a slow train will occupy more blocks than it should and will thus hold signals in rear at danger, which limits capacity by preventing trains standing at said signals from moving up into the next block, which might be physically unoccupied but cannot be entered due to signalling overlaps or other fixed-block signalling constraints.

Moving block signalling operates on the principle of resizable signalling blocks - each train gets a virtual signalling block assigned to it. This block's size is determined by several factors: train speed, line gradients, line speed, location, and so on. When trains on such a line are moving rapidly, the blocks increase in size sufficiently to prevent a following train from approaching too closely without being stopped first by the signalling. However, when trains are moving slowly, a large block is no longer needed as a following train would be physically incapable of reaching a high enough speed sufficient to hit a train in front without being stopped by the signalling. Therefore, each train's block is progressively shrunk as the line speed decreases, allowing trains to get closer together and therefore decreasing the headways between trains. The "moving" aspect is a reference to the fact that each train gets its own virtual block, which follows the train as it traverses the line; therefore, every train occupies its own properly sized block, resized as per line conditions, and line capacity is increased as a result.

Naturally, there are other constraints and requirements that are taken into account by the intelligence controlling the sizing of the virtual blocks, but the basic resizing of the blocks in response to the above criteria is the core of moving block signalling.

fx: stampede of AETs and infracos and signalmen...

[Deleted]

They use a similar system on the DLR from what I have heard. Isn't it the same system that will be installed on the Jubilee?

[tubeprune]

May I offer a few thoughts since I have been very closely involved with "Run Out - Run In" (RORITs) in the recent past.

First, RORITs are not the same for each station. The composition of each RORIT is dependent on train speed on the approach and exit to each station and therefore on the overlaps of the signals provided. If you have a RORIT of 60s and a dwell of 45 seconds at a busy central London station like Oxford Circus, you get a headway of 105s. This is just over 34 trains per hour (tph). However, not all stations will allow a 60s RORIT and the headway is always restricted by the worst station.

"Innocentabroad" is right. The dwell times will be a serious factor for headways but there are other issues. The Piccadilly Line is currently signalled for 28.5 tph but the timetable only allows 24 tph. Various pinch points, including dwell times at busy stations, the terminal arrangements at the east end, and the variability of manual driving all restrict the throughput. More modern signalling could improve things but, beware of the "moving block" title.

As members have pointed out above, the theory behind moving block signalling is that it gets rid of the restrictions of fixed block signalling and it allows trains to get closer together. Put another way, it operates like cars running on a motorway. You assume all cars are moving at the same speed in the same lane so, if the car in front slows down, you have room to slow down too, even though you are quite close.

If you want to do this on a railway, you need to transmit electrical signals between every train on the line and a control centre. You must do this continuously and vitally so that it is always safe. Every train has to tell the control centre where it is, which direction it's going and how fast it's going. It must update the info every few milliseconds. If you lose the computer, you lose the railway, just as we saw on the Central Line a couple of weeks ago.

Now, the signalling system to go on the Jubilee and Northern Lines is Alcatel S40, they tells us. This is like the Docklands system. Alcatel sells it as "moving block" but it isn't really. It's very small fixed block, set by the 25m cable crossovers which are used as position locators. It does let the trains operate more closely and it will offer a more frequent service, as long as the dwell times are kept down.

So far, no one has been able to deliver a proper moving block system. It requires a very good transmission system and a recovery and fall back strategy which no one has yet been able to offer. That's why it was abandoned for the Jubilee and WCML.

I will be happy to expand on these issues or answer questions if you would like to know more.

[Deleted]

Many many thanks for all your replies.

What little understanding of the current system I have is that it could be described as "overlapping fixed block" with all the trains alternately in what might be thought of as odd and even numbered blocks. If I understand Tubeprune properly, the S40 system simply increases the number of overlaps from pairs to triples (or perhaps some higher number than that) and dispenses with visual signals - or tries to, until the H&S reps get on the case

The 10% capacity increase quoted was roughly in line with what I expected it would be. In the case of the Victoria line, I would imagine that loadings would increase by that much within, oh, a week of the system being introduced.

[Deleted]

Absolute block is still the best! ;D

Sam

[Deleted]

tubeprune said:

May I offer a few thoughts since I have been very closely involved with "Run Out - Run In" (RORITs) in the recent past.

First, RORITs are not the same for each station. The composition of each RORIT is dependent on train speed on the approach and exit to each station and therefore on the overlaps of the signals provided. If you have a RORIT of 60s and a dwell of 45 seconds at a busy central London station like Oxford Circus, you get a headway of 105s. This is just over 34 trains per hour (tph). However, not all stations will allow a 60s RORIT and the headway is always restricted by the worst station.

I should have mentioned the stations at which the measurements I took were taken. The Central Line's measured RORITs are very consistant to within +/-2secs of 60secs at all central area stations where I happened to have been able to measure the RORIT. The Northern Line's RORIT was measured SB at Clapham Common with 3 multiple homes. Paris Line 14's RORIT was measured at both termini, with Line 14 using a Matra/Siemens moving block signalling system. The figures for the Madrid Line 6 were in a press release by Bombardier about their moving block signalling system.

I would be interested to know the planned RORITs for the Victoria/SSL upgrades and Jubilee/Northern upgrades? I've heard the 60sec figure banded about for the former, which is consistant with the Central Line's fixed block signalling system.

"Innocentabroad" is right. The dwell times will be a serious factor for headways but there are other issues. The Piccadilly Line is currently signalled for 28.5 tph but the timetable only allows 24 tph. Various pinch points, including dwell times at busy stations, the terminal arrangements at the east end, and the variability of manual driving all restrict the throughput. More modern signalling could improve things but, beware of the "moving block" title.

I've read some interesting papers, where operating margins of at least 20secs on top of the RORIT and dwell time are usually used by metro systems. The longer the RORIT and dwell, the longer the operating margin should be. However, there are a handful of metro systems, which can operate with less operating margin.

Now, the signalling system to go on the Jubilee and Northern Lines is Alcatel S40, they tells us. This is like the Docklands system. Alcatel sells it as "moving block" but it isn't really. It's very small fixed block, set by the 25m cable crossovers which are used as position locators. It does let the trains operate more closely and it will offer a more frequent service, as long as the dwell times are kept down.

So far, no one has been able to deliver a proper moving block system. It requires a very good transmission system and a recovery and fall back strategy which no one has yet been able to offer. That's why it was abandoned for the Jubilee and WCML.

I will be happy to expand on these issues or answer questions if you would like to know more.

I don't want to question you too much, as I'm sure you know more on this subject than me, but was under the impression that Seltrac S30 was a fixed block distance to go system, and Seltrac S40 was a true moving block system?

Alcatel, Siemens, Bombardier, and Alstom have all installed moving block signalling systems recently with two way continuous radio transmission between train and wayside. All use different methods for locating train position.

[Phil]

Moving block must be 100% fail-safe if fixed signals are to disappear - and that's difficult to achieve.

I remember the early days of the Cl 158s on Swindon-Gloucesters. At times they would disappear off the signaller's diagram, even though it was impossible (it was the new disc brakes not wiping the wheel treads clean). Regrettably (Sod's law?) it was always on the single-line stretch between Swindon and Kemble (and that's been protected solely by Track Circuit Block. Hmmm.......).

With fixed block it is no problem, assuming the signaller is awake. All Signals to red until train reappears/driver telephones from red stick. Delays but no danger.

But if a train were to disappear off the moving block system (and weren't to be noticed) there is no way to stop it apart from cab-to-shore. And if it had truly disappeared the train behind would have no indication at all.

Now obviously this would not be allowed to happen: the fail-safe aspect would have to be proved in all situations before being rolled out, but I am simply trying to show how much more difficult it is to get 100% fail-safe with moving block.

[Deleted]

rbruce1314 said:

Moving block must be 100% fail-safe if fixed signals are to disappear - and that's difficult to achieve.
...
But if a train were to disappear off the moving block system
...
Now obviously this would not be allowed to happen: the fail-safe aspect would have to be proved in all situations before being rolled out, but I am simply trying to show how much more difficult it is to get 100% fail-safe with moving block.

But how difficult should it be? Surely the software can be written so that the default situation is that the train is where it was last seen. In this case, if contact with train A is lost, the train B behind it assumes that train A is still where it was last seen, and therefore cannot be endangered. The software should allow for physical/mechanical mishaps, shouldn't it?

[agoodcuppa]

alantan said:

But how difficult should it be?

As with so many things the basic theory is easy. Making it work in the real world is, however, a different matter.

Unless of course, if you've got the skills...... ;D

[Deleted]

Not me, it's all I can do to get our simple banking software to do what it says on the tin, and no lives are ever in jeopardy. ;D

[tubeprune]

A whole lot of questions to answer already! I'll try to answer as many as I can.

First, stephenk: Please don't think I am an expert of signalling technology and I'm sorry, I don't know the original RORITs as calculated by LU for the various lines. The formula they use is a little odd. I woud prefer to see an approach where the dwell was completely separate from the calculation. RORI should comprise the distance between the sighting point of the home signal and its clearance point at the starting signal overlap. The train equipment response times and passenger loading/unloading times should be separate.

S30 uses what Alcatel refers to as "virtual block". These are based on a fixed block divided into smaller sections. It is used as an overlay for older systems. S40 is the same except they make the fixed blocks fit their own design.

As for the companies which make signalling systems, they all have to use some form of fixed reference - a "tag", a "balise", a "waveguide", call it whatever you like. We cannot get away from the fact that they all need some sort of marker system to divide the route into sections. Perhaps we can ignore the misnomer "moving block" and simply describe the technology.

The margin on LU was originally fixed at 25s. They have relaxed this in some locations.

Phil: If a train disappears from a transmission based signalling system it will automatically stop. The system then has to re-establish communication with the train and all others in the area before movement can been authorised. The delay caused by this "re-boot" is one of the reasons why many operators require a back-up system as well. One may ask whether the expense of providing the two systems makes the new transmission based system worthwhile. Of course, the extra throughput does offer a big incentive.

[Deleted]

rbruce1314 said:

Moving block must be 100% fail-safe if fixed signals are to disappear - and that's difficult to achieve.

Whilst moving block is very complicated stuff, it has an excellent safety record. As far as I'm aware there has never been a fatal accident on a metro with moving block signalling in operation. There are now over 20 metro systems with moving block signalling, all running trains safely.

On a side note, the rubber tyred transit system under construction at Heathrow T5 will use Bombardier's version of moving block signalling.

[Phil]

stephenk said:

Whilst moving block is very complicated stuff, it has an excellent safety record. As far as I'm aware there has never been a fatal accident on a metro with moving block signalling in operation.

Good news, so in the UK it will all be down to the cost/benefit ratio.....

(shortsighted refusal to spend money up front has been the failing of almost all sides of UKplc for years now)

[Deleted]

rbruce1314 said:

Good news, so in the UK it will all be down to the cost/benefit ratio.....

(shortsighted refusal to spend money up front has been the failing of almost all sides of UKplc for years now)

According to the manufacturers, communication based train control (CBTC) signalling should be cheaper to install and maintain than conventional signalling. This is because there is less wayside equipment, particularly in the more modern radio based systems. Also, track circuits are usually not required (although a few metros still retain them for back-up or broken rail detection).

[tubeprune]

Published prices for LU resignalling show:
Vic/SSL at £4.25million per km
JNP at £3.16million per km
One has to assume that both include on-train and control centre equipment. The Alcatel system is more developed so it is bound to be a bit cheaper. Both systems retain a back-up element in case the computers fall over.

[tube10]

How can I tell if an undergroud line is using fixed or moving block signals and which one has it?

[Tom]

We don't have any moving block at the present time.

[jimini]

The only place you'll find it in Blighty is on the DLR, tube10.

[Deleted]

.....But in a few years time there will be more moving block systems in the UK:-
Heathrow T5 AGT - 2008
Jubilee Line - 2009
Northern Line - 2011
Piccadilly Line - 2016?