Route Knowledge

[TheKnightWho]

But I would think you would still need to know the routes to some extent (lesser, admittedly). I certainly wouldn't fancy not knowing what was round the next corner! And again the cost of introducing it, retraining everybody etc etc. Why bother?

I don't think anyone's suggesting we'd switch it over right now, even if we could. It'd be far too expensive, and politically very "courageous".

However, it's the likely end result of a smooth transition, which is already being begun with ERTMS. It's the constant transmission and in-cab signalling that are the expensive bits to install, and ERTMS is what will create that in the first place. Although the actual computational software required for an all-ATO network would have to be much more sophisticated, I'm not convinced the trains themselves would actually need a huge amount of work done to them. All that fancy computation would be at the signal controller.

--- old post above --- --- new post below ---

Are you going to Automate the passengers as well?
They are the cause of most small delays with their one door syndrome!
How will automation deal with that?

Because ATO isn't trying to solve that, but is addressing other problems.

Obviously things will never be perfect, but ATO doesn't claim to solve every issue on the railway.

And frankly, if we carried on employing all drivers as guards then an extra pair of hands in the main body of the train would help get people on and off faster!

 

[PermitToTravel]

Someone highlighted one of the big differences between UK signalling/rest of Europe way up thread when they mentioned driving a sim train to somewhere or other. The speed signalling system used abroad (compared to the route signalling system we use here). I would imagine if we had a similar system (drivers being told how fast to drive and then being routed wherever, instead of which route to take and having learnt the speed) then the problems with diverted routes not being known could be minimised. I can also see how ERTMS could be used with benfit in this setup

Definitely agree. Do you know if the current ERTMS trial implements speed signalling in some way or other?

But I would think you would still need to know the routes to some extent (lesser, admittedly). I certainly wouldn't fancy not knowing what was round the next corner! And again the cost of introducing it, retraining everybody etc etc. Why bother?

Yeah, I can't imagine a situation where a train is driven by someone who doesn't know the line at all (which will need to happen in the event of a system failure, and does sometimes have to happen on the DLR at present).

Regarding the cost, the research and development is really being paid for by organisations that need the capacity benefits - LU, Crossrail; and by railways in other countries who have their own reasons for going automatic.

Introducing automatic trains on National Rail today wouldn't make any economic or financial sense whatsoever. As and when, however, new trains need to be built and lines need to be resignalled, why not do so with the technology available at the time?

Are you going to Automate the passengers as well?
They are the cause of most small delays with their one door syndrome!
How will automation deal with that?

"We can't fix one thing so we shouldn't ever fix anything else ever."

 

[muz379]

ATO can ensure consistent uniform train operation, beneficial on very high frequency routes like the Thameslink or ELL cores.
.

Can it though , the number of times delay slips are issued to drivers (who are also meant to drive in a uniform way ) because of differences between top speed , acceleration and braking of the same traction class tells you that even if a computer is driving a train its still going to result in consistancy problems .


Not to mention some of the inconsistencies in the driving exhibited by drivers is for good reason , for example I recall one incident when a driver on a train I was working brought the train to a crawl and then stopped short on aproach to a station because people on the platform where fooling about pretending to push each other over the edge . As much as we can all say yeah it would be their fault if they fell in front of the train and got themselves killed thats not going to be a nice experience for anybody else on that platform or the passengers and driver on the train . Not to mention emergency services who have to come and deal with that

 

[PermitToTravel]

I've been sniped by TheKnightWho, but, err, what he said. ^

 

[PermitToTravel]

Can it though , the number of times delay slips are issued to drivers (who are also meant to drive in a uniform way ) because of differences between top speed , acceleration and braking of the same traction class tells you that even if a computer is driving a train its still going to result in consistancy problems .


Not to mention some of the inconsistencies in the driving exhibited by drivers is for good reason , for example I recall one incident when a driver on a train I was working brought the train to a crawl and then stopped short on aproach to a station because people on the platform where fooling about pretending to push each other over the edge . As much as we can all say yeah it would be their fault if they fell in front of the train and got themselves killed thats not going to be a nice experience for anybody else on that platform or the passengers and driver on the train . Not to mention emergency services who have to come and deal with that

The ideal from a capacity standpoint is to have trains running at linespeed until it's time to brake, at which point they do so at the highest comfortable rate (obviously less a margin for safety and in case greater braking is needed). If the trains can meter adhesion in real time and calculate accordingly (I believe that at present they can't) then they'll all come to the same conclusion.

People aren't as likely to pretend to push each other over the edge if there's no driver to see them do it as I imagine they're feinting with the driver a lot of the time, not with their mate.

Nonetheless, drunk people close to the edge is a safety issue. On the DLR it's dealt with by, on Friday/Saturday nights / whenever else it's felt necessary, the guard either sitting up front ready with the emergency brake, or even driving into platforms so as to do so suitably slowly.

 

[bb21]

And frankly, if we carried on employing all drivers as guards then an extra pair of hands in the main body of the train would help get people on and off faster!

That's a very idealistic view of human behaviour.

 

[TheKnightWho]

Can it though , the number of times delay slips are issued to drivers (who are also meant to drive in a uniform way ) because of differences between top speed , acceleration and braking of the same traction class tells you that even if a computer is driving a train its still going to result in consistancy problems .


Not to mention some of the inconsistencies in the driving exhibited by drivers is for good reason , for example I recall one incident when a driver on a train I was working brought the train to a crawl and then stopped short on aproach to a station because people on the platform where fooling about pretending to push each other over the edge . As much as we can all say yeah it would be their fault if they fell in front of the train and got themselves killed thats not going to be a nice experience for anybody else on that platform or the passengers and driver on the train . Not to mention emergency services who have to come and deal with that

There are a few issues with ATO that I can think of - stuff like that is obviously a worry. I'm not sure how you'd account for people that are acting unsafely or who look like they might want to jump in front of a train. Perhaps better CCTV monitoring or more staff platform presence could work? The latter would certainly help improve image, too.

However, in terms of inconsistency of driving, those delays you're talking about are exactly what ATO wants to solve. Computers can make trains accelerate, brake or whatever at extremely consistent rates, and can even account for mechanical inconsistencies in the trains and track themselves without too much work. It's exactly why they need it to get 32tph in some places.
--- old post above --- --- new post below ---

That's a very idealistic view of human behaviour.

People are already used for that sort of thing in this country and Japan (although I'm not a big fan of literally crushing people on).

An extra pair of hands for when people dither or misbehave can be extremely useful.

 

[bb21]

People are already used for that sort of thing in this country and Japan (although I'm not a big fan of literally crushing people on).

An extra pair of hands for when people dither or misbehave can be extremely useful.

You clearly haven't been on my train a few weeks back when the guard tried in vain to get people to board from all available doors, and that is far from an isolated incident. That is just one of many examples I can give.

It might possibly be useful when some people "misbehave", but that constitutes a very small minority.

 

[muz379]

The ideal from a capacity standpoint is to have trains running at linespeed until it's time to brake, at which point they do so at the highest comfortable rate (obviously less a margin for safety and in case greater braking is needed). If the trains can meter adhesion in real time and calculate accordingly (I believe that at present they can't) then they'll all come to the same conclusion.

People aren't as likely to pretend to push each other over the edge if there's no driver to see them do it as I imagine they're feinting with the driver a lot of the time, not with their mate.

Nonetheless, drunk people close to the edge is a safety issue. On the DLR it's dealt with by, on Friday/Saturday nights / whenever else it's felt necessary, the guard either sitting up front ready with the emergency brake, or even driving into platforms so as to do so suitably slowly.

your missing the point completely though . If An ideal 156 takes 5 minutes to get up to linespeed currently with a driver driving . But then the same driver driving another 156 it takes him 5 minutes 50 seconds to get up to the same linespeed using the same driving technique over the same stretch of track in the same conditions how is the computer going to be able to remove that inadequacy between the two different trains of the same class ? It happens and it happens more often than you think . Again what if Linespeed is 75mph , the ideal 156 can obviously achieve that . What if the computer is driving a 156 that can only actually achieve 73.5MPH ? again another inconsistancy .

The issues that cause this are more often than not mechanical . nothing to do with inconsistancies in driving techniques .

Same with brakes . obviously a set of new blocks is going to achieve a much better brake rate (once bedded in ) than ones that are just about near replacement . Unless you tighten up the tolerances for stuff like brakes which will end up costing more in replacement and units out of traffic I dont see how you are going to get rid of these inconsistencies

 

[TheKnightWho]

You clearly haven't been on my train a few weeks back when the guard tried in vain to get people to board from all available doors, and that is far from an isolated incident. That is just one of many examples I can give.

It might possibly be useful when some people "misbehave", but that constitutes a very small minority.

Of course it wouldn't be perfect. However, that doesn't mean it can't work at times. Better than people losing their jobs, right?
--- old post above --- --- new post below ---

your missing the point completely though . If An ideal 156 takes 5 minutes to get up to linespeed currently with a driver driving . But then the same driver driving another 156 it takes him 5 minutes 50 seconds to get up to the same linespeed using the same driving technique over the same stretch of track in the same conditions how is the computer going to be able to remove that inadequacy between the two different trains of the same class ? It happens and it happens more often than you think . Again what if Linespeed is 75mph , the ideal 156 can obviously achieve that . What if the computer is driving a 156 that can only actually achieve 73.5MPH ? again another inconsistancy .

The issues that cause this are more often than not mechanical . nothing to do with inconsistancies in driving techniques .

Same with brakes . obviously a set of new blocks is going to achieve a much better brake rate (once bedded in ) than ones that are just about near replacement . Unless you tighten up the tolerances for stuff like brakes which will end up costing more in replacement and units out of traffic I dont see how you are going to get rid of these inconsistencies

Because if it's monitoring in real time it can take into account mechanical differences between the trains, and can therefore account for it. This is how a system with driverless cars would work, too, and they're already getting reasonably good, although not perfect yet. So if a train has poorer acceleration than another, the system would account for this by applying more power than it otherwise would etc. After all, varying passenger loadings and fuel use will always cause variations like this, and so it's never assumed that a train's weight or performance will be uniform, even at different points on the same journey.

The issue you raise is way more of a problem when it comes to freight trains, which have much more variations in their loadings (although tend to remain of relatively uniform weight throughout a particular journey).

ATO should (and would) never be installed until it can perform at least as well as drivers.

 

[W230]

However, in terms of inconsistency of driving, those delays you're talking about are exactly what ATO wants to solve. Computers can make trains accelerate, brake or whatever at extremely consistent rates, and can even account for mechanical inconsistencies in the trains and track themselves without too much work. It's exactly why they need it to get 32tph in some places.

Spot on. For example through the TL core I will wait outside a station on a single yellow (due to a mid platform red) until the signal clears to two yellows. I'll also drive below linespeed when on restricted signals between stations in the core to give the train currently in the platform a chance to leave and save me stopping outside if possible.

But as I understand it, ATO will know what is happening further ahead and see the bigger picture of trains starting to move/will know a signal is going to clear where the driver can't physically see them moving and has to presume the signal will stay at danger, so the loss of capacity with defensive driving polices will no longer happen. Everything will in effect be driven at maximum linespeed with maximum (comfortable) braking.

 

[455driver]

Spot on. For example through the TL core I will wait outside a station on a single yellow (due to a mid platform red) until the signal clears to two yellows. I'll also drive below linespeed when on restricted signals between stations in the core to give the train currently in the platform a chance to leave and save me stopping outside if possible.

But as I understand it, ATO will know what is happening further ahead and see the bigger picture of trains starting to move/will know a signal is going to clear where the driver can't physically see them moving and has to presume the signal will stay at danger, so the loss of capacity with defensive driving polices will no longer happen. Everything will in effect be driven at maximum linespeed with maximum (comfortable) braking.

Just like the Jubilee line isnt!

 

[TheKnightWho]

Spot on. For example through the TL core I will wait outside a station on a single yellow (due to a mid platform red) until the signal clears to two yellows. I'll also drive below linespeed when on restricted signals between stations in the core to give the train currently in the platform a chance to leave and save me stopping outside if possible.

But as I understand it, ATO will know what is happening further ahead and see the bigger picture of trains starting to move/will know a signal is going to clear where the driver can't physically see them moving and has to presume the signal will stay at danger, so the loss of capacity with defensive driving polices will no longer happen. Everything will in effect be driven at maximum linespeed with maximum (comfortable) braking.

And that's the crux of it. Even on lines which aren't even remotely approaching capacity, defensive driving can still cause delays even when trains are 2 blocks apart - ATO could drive trains even 1 block apart as a current driver would only be able to on a green signal.

That being said, one danger I can anticipate of this (and with driverless cars) is that if something extremely unexpected happens and a train crashes, it could cause crashes to be more severe as the trains behind will have less of a chance to slow down as they'll be travelling faster.

So, say, where a train would today be either going much more slowly or would be 2 or 3 blocks behind, with ATO it could be going full linespeed and be only a block behind. Although under normal circumstances this would be perfectly safe as the computer in the second train knows exactly what the first train is doing, it could still cause issues if something totally unpredictable happens.

Therefore, it wouldn't totally be able to get rid of defensive driving. Presumably it would never allow trains to be at a distance less than the second train's stopping distance, which would still mean driving below linespeed at times.

--- old post above --- --- new post below ---

Just like the Jubilee line isnt!

Look, I get that you see ATO as a threat to people's jobs, but trying to snipe it at every opportunity by pointing out small issues, teething problems and things it doesn't even purport to solve isn't helping.

 

[bb21]

Of course it wouldn't be perfect. However, that doesn't mean it can't work at times. Better than people losing their jobs, right?

I am confused as to your point.

What is the point of these roles? Do these problems already exist on the network and is it worth spending money trying to solve them? If so, why not employ these people now if it benefits the overall running of the railway?

If not, why employ people to carry out these roles after some drivers lose their original jobs? Is it simply a measure to avoid redundancies?

 

[PermitToTravel]

your missing the point completely though . If An ideal 156 takes 5 minutes to get up to linespeed currently with a driver driving . But then the same driver driving another 156 it takes him 5 minutes 50 seconds to get up to the same linespeed using the same driving technique over the same stretch of track in the same conditions how is the computer going to be able to remove that inadequacy between the two different trains of the same class ? It happens and it happens more often than you think . Again what if Linespeed is 75mph , the ideal 156 can obviously achieve that . What if the computer is driving a 156 that can only actually achieve 73.5MPH ? again another inconsistancy .

The issues that cause this are more often than not mechanical . nothing to do with inconsistancies in driving techniques .

Same with brakes . obviously a set of new blocks is going to achieve a much better brake rate (once bedded in ) than ones that are just about near replacement . Unless you tighten up the tolerances for stuff like brakes which will end up costing more in replacement and units out of traffic I dont see how you are going to get rid of these inconsistencies

Computers are actually better than humans at detecting small discrepancies and adjusting - it's a wholly solved problem that's necessary whenever any machine autonomously does something mechanical. The first 156 in your example would just get driven harder, and the second one would lose as much time as it would if human driven.

 

[455driver]

Look, I get that you see ATO as a threat to people's jobs, but trying to snipe it at every opportunity by pointing out small issues, teething problems and things it doesn't even purport to solve isn't helping.

I have yet to travel on an ATO train that could in any way be described as comfortable, its either full acceleration or full braking which isnt exactly conversant with a smooth journey!

 

[PermitToTravel]

Spot on. For example through the TL core I will wait outside a station on a single yellow (due to a mid platform red) until the signal clears to two yellows. I'll also drive below linespeed when on restricted signals between stations in the core to give the train currently in the platform a chance to leave and save me stopping outside if possible.

But as I understand it, ATO will know what is happening further ahead and see the bigger picture of trains starting to move/will know a signal is going to clear where the driver can't physically see them moving and has to presume the signal will stay at danger, so the loss of capacity with defensive driving polices will no longer happen. Everything will in effect be driven at maximum linespeed with maximum (comfortable) braking.

Indeed. Honestly it surprises me that TOC maintain their defensive driving policies on routes like the Thameslink core - drivers are obviously professional enough to stop where they want to otherwise trains would crunch buffers all the time, so I don't see why they can't be allowed to drive up to reds and stop in the same manner, particularly where there's a generous overlap. LU, for whom capacity is far more important, clearly recognise this

 

[SWTCommuter]

This is how a system with driverless cars would work, too

You are Zoe and I claim my five pounds

 

[Domh245]

The selctrac S40 system, which is the most commonly used system in the UK isn't known for it's smoothness - it has a binary driving style, which gives a very uncomfortable ride, especially when it can't handle coasting! However I find the westinghouse system on the victoria line sufficiently comfortable. It is highly unlikely that any mainline ATO will use a binary driving style, but will instead be a bit more of a polished product.

 

[PermitToTravel]

I have yet to travel on an ATO train that could in any way be described as comfortable, its either full acceleration or full braking which isnt exactly conversant with a smooth journey!

Indeed, anything used on the big railway would need to drive to the timetable. This isn't a difficult problem at all, but it's something that has to be considered.

It's worth noting that full acceleration and full braking is the quickest way, which is what's needed on LU when you have a queue of trains that need to get out of each other's way.

That said, I might be wrong here, but don't ATO trains on one of the LU lines ease off if they're running early?

 

[TheKnightWho]

The selctrac S40 system, which is the most commonly used system in the UK isn't known for it's smoothness - it has a binary driving style, which gives a very uncomfortable ride, especially when it can't handle coasting! However I find the westinghouse system on the victoria line sufficiently comfortable. It is highly unlikely that any mainline ATO will use a binary driving style, but will instead be a bit more of a polished product.

This. I think it's important to remember that by the time ATO would even be remotely considered for mainline use that the technology will have improved enormously.

Another important thing to remember is that with the required measuring of track for adhesion and trains for mechanical performance a lot of potential serious problems could be spotted far more quickly and solved at little expense.

It's all very well to say that a driverless train couldn't spot a cracked rail on another line or know what to do when the engine fails, but it could also very well not ever let the track or trains get to that state, as faults would be reported to maintenance staff and fixed long before they caused issues.
--- old post above --- --- new post below ---

You are Zoe and I claim my five pounds

I'm afraid that one's gone over my head

 

[bramling]

The selctrac S40 system, which is the most commonly used system in the UK isn't known for it's smoothness - it has a binary driving style, which gives a very uncomfortable ride, especially when it can't handle coasting! However I find the westinghouse system on the victoria line sufficiently comfortable. It is highly unlikely that any mainline ATO will use a binary driving style, but will instead be a bit more of a polished product.

Seltrac also has what is essentially a major design flaw -- if the train slides (for example due to wet rails), the train will lost communication with the system, and have to go through a long re-entry process. This issue is avoided with use of extremely gentle brake rates in the open air (LUL). It is slightly less of an issue on DLR as their network is much more compact, much is elevated or in tunnel, and the nature of the surroundings means there are few tree-lined cuttings, therefore control room adjustments of brake rates is more effective.

Gentle brake rates would simply not work on the mainline, as the amount of time lost would be too much, both due to the amount of open-air mileage (virtually everything!) and the relatively higher speeds meaning that the earlier braking would impose a proportionally greater impact on timekeeping.

The Victoria Line system doesn't have this handicap as far as I know, the Central Line definitely doesn't. But inability for ATO to adjust its brake rate to prevailing conditions will always be a major challenge for designers. The technology definitely allows for brake rates to be altered from the control room, but you can't apply a brake rate that fits the needs of all trains right across the system, or predict that it's about to rain! You could allow the train driver to use their judgement and select a brake rate, but if you're doing that then why bother with ATO?

 

[PermitToTravel]

I believe the issue is that trains on those lines communicate with the signalling system via the track, which isn't handy if the train slides (and it's said that this does result in ATO giving up on the Central line a lot, but doesn't impose a lower brake rate there - the driver just jumps in and drives if necessary).

The system that was to be fitted to the Circle/Metropolitan/Hammersmith and City/District lines before Bombardier jumped out communicates by wifi, which is far more resilient to this sort of problem. I imagine on the big railway it would use GSM-R, which has support already built-in for carrying signalling and other safety-critical communications. Doesn't the Cambrian work this way?

 

[Crossover]

Were they route conducted over the last bit on this occasion?

Conducted I think. When I did it I think we picked someone up at Newark and we paused briefly on the return North at Ally Pally, presumably to drop someone off

I love how you actually ignore what I said and just zoned in on drivers losing jobs.

Don't worry - the particular poster in question has a habit Of doing such

I'm afraid that one's gone over my head

Zoe is a previous forum member

 

[TheKnightWho]

I am confused as to your point.

What is the point of these roles? Do these problems already exist on the network and is it worth spending money trying to solve them? If so, why not employ these people now if it benefits the overall running of the railway?

If not, why employ people to carry out these roles after some drivers lose their original jobs? Is it simply a measure to avoid redundancies?

Yes. I was being a bit tongue-in-cheek when I said it. A far better system would be passive retirement, where drivers are simply not replaced. Where they work on a line that has ATO implemented they could do as a DLR driver currently does, or they could be put onto another role that isn't being automated.
--- old post above --- --- new post below ---

Zoe is a previous forum member

Ah ok! No idea if it's a good thing or not that someone thinks I was them

 

[PermitToTravel]

Ah ok! No idea if it's a good thing or not that someone thinks I was them

Zoe had a bit of a reputation for posting about driverless cars a lot

 

[muz379]

Because if it's monitoring in real time it can take into account mechanical differences between the trains, and can therefore account for it. This is how a system with driverless cars would work, too, and they're already getting reasonably good, although not perfect yet. So if a train has poorer acceleration than another, the system would account for this by applying more power than it otherwise would etc. After all, varying passenger loadings and fuel use will always cause variations like this, and so it's never assumed that a train's weight or performance will be uniform, even at different points on the same journey.

The issue you raise is way more of a problem when it comes to freight trains, which have much more variations in their loadings (although tend to remain of relatively uniform weight throughout a particular journey).

ATO should (and would) never be installed until it can perform at least as well as drivers.

But then your basically saying we need ATO to get to 32tph in some places . but then because of inconsistencies in passenger loadings and train weight and unit performance there will be discrepancies which will threaten your 32tph .

Don't get me wrong trains driven by drivers aren't any more likely or able to achieve 32tph either , but then nobody is making such grand claims of that mode of operation .

ATO will eventually offer advantages to the railway of course and as I have acknowledged it will eventually happen no matter what the resistance . That eventuality is however a long way off and I dont think for many it is welcome you yourself have even pointed that out . What I do however wish to see is honestly in the benefits it will bring about and not just adding things in like capacity and consistency of operation without actually thinking about the issues seriously .

32tph with passenger operation is IMO not going to be possible in a consistent and reliable fashion on a timetabled service on anything other than a computer model with perfect passengers and uniform unit performance .

Units do and will continue to offer discrepancies no matter what their actual mode of operation (take the number of trains I work that cant actually achieve their maximum speed only a figure close enough to be in tolerance even when "flat out")There will always be the train that no matter how hard it is driven cannot achieve the acceleration or top speed of the model train of that class .

 

[21C101]

Indeed. Honestly it surprises me that TOC maintain their defensive driving policies on routes like the Thameslink core - drivers are obviously professional enough to stop where they want to otherwise trains would crunch buffers all the time, so I don't see why they can't be allowed to drive up to reds and stop in the same manner, particularly where there's a generous overlap. LU, for whom capacity is far more important, clearly recognise this

Indeed, defensive driving is worth a topic with its own thread. My recollection that it came in after a political panic about SPADS and resulted in trains creeping towards red signals like a zookeeper approaching a pit of Crocodiles (because Drivers have been ordered to drive like that) and made approach control (where a signal is held at red until the train is almost upon it when on diverging routes) cause significant delays.

In intercity trains it probably dosen't make much difference, but on high density suburban routes it is I'm sure one reason why trains are often slower now than they were in the '50s, and I'm sure its one reason why there are so often delays at places like London Bridge.

Taking Thameslink Core, the funny thing is that go down the escalators at Blackfriars or over the Footbridge at Farringdon and you see, despite a two aspect signalling system with no distants (yellow signals are repeaters on LUL), up to 27 manually driven heavy rail trains an hour each way (more than Thameslink will have with ATO) rushing into a platform towards a red light and stopping just before it, just like on BR in the '70s and '80s.

The question is, if its safe for LUL trains to approach red signals like this, why isn't it safe for TOC trains running at less than 60mph, especially now that TPWS is ubiquitous and gives similar protection to the LUL tripcocks?

Indeed, is defensive driving, in extremis, making Drivers uneconomic compared with ATO by making ATO far more efficient in terms of train throughput; and necessary on things like Thameslink Core to get 24 TPH?

Similarly, I'm pretty sure also that after two minutes, an LUL train can start off and go through a red proceeding at sight to the next green signal (not at all signals but most). Why do we have to have complicated systems with flashing shunt signals on Thameslink to allow this, again making manual drivers more "expensive" as a complicated proceed at sight permission system has to be paid for?

 

[Skoodle]

On the East London Line core route, the signalling is technically compatible for 24tph, but currently struggling with 16tph. Southbound is fine, however Northbound from Surrey Quays gets congested. Due to the smaller signal sections and AWS magnets being closer to signals, our defensive driving policy is to be doing 10mph over the magnets. This obviously makes the situation worse.

Get to Surrey Quays arriving on a red. 10mph to signal before Canada Water, it clears, then again 10mph in to the platform. You can imagine the slow progress of trains behind.

Using route knowledge though, I can tell by how quickly the signal clears exactly how the train ahead is progressing. If it's going very slowly I will just continue cautiously as I had already been. If it's moving quite quickly, I will hold back a few extra seconds. Green at Rotherhithe, I can travel linespeed to Wapping. Two successive greens at Wapping, I can travel linespeed to Shadwell. At Shadwell, there's a reflection of the second signal from there further in the tunnel. Wait an extra ten seconds for it to turn green and I know that I'll have clear signals all the way to Haggerston.

If I'm late and I can see from the CIS screens on platform that the train behind is close, I won't do the above, as it will just make things worse, however I don't see the point in chasing reds all the way to Dalston!

ATO *could* be implemented on the core route, however with too many other lines we integrate in I don't believe it would be practical.

 

[Hyphen]

It's a completely closed system. There would be no more risk of "hackers" taking control of trains than there is of them taking control of the current signalling system - absolutely minimal.

You haven't thought this answer through at all. The signalling system is not closed; by its very nature it can't be. A closed system is one which sits alone in (presumably) a guarded building, with no connections to the 'outside world'. Depending on your definition of the word, the signalling system gets "hacked" on a regular basis.

Cable thefts are an all-too common occurrence on the railway (although one which feels like it's going down in my experience), and if the cables to the signals are the ones stolen, this is a primitive Denial of Service attack (even if it wasn't necessarily the motive of those doing it). Of course, this doesn't represent taking control of the signalling system, but this has happened in the past. The most famous incident being The Great Train Robbery, where the gang covered over the green aspect and lit the red aspect up with a battery. And that was possible in the 60s!

You state in a later post that GSM-R is most likely to be used for any sort of ATO communication - I agree that it's handily placed and designed for this sort of thing. But GSM-R is a radio system - it communicates wirelessly. For any 'hacker' interested in this sort of thing, that only makes the whole game easier! No longer do you need to physically turn up at the signal or splice a cable (and risk being seen) - all you'd need is to rock up at a car park near the line and stick an antenna on the roof. GSM (not specifically -R) itself is not infallible. The early GSM (2G) encryption was broken many years ago, and researchers did prove the ability to listen into calls made on live base stations. More recent editions of GSM (3G/LTE) did provide much more complex crypto between handset and base, but I don't know if these were written into the GSM-R spec. And I couldn't tell you what they are using. But it doesn't mean the more complex ones are unbreakable either.

Of course, there are then other attack vectors such as modern signalling centres - from press pictures I've seen, the workstations in there seem to have Internet access, and probably no "air gap" onto the actual underlying signalling system (even if there was, some way of remotely controlling the workstation's mouse would do it :P). There has been a lot of news in recent years about very expensive SCADA systems being vulnerable to this sort of thing. In reality this isn't much different.

So with that in mind, do you still think any ATO control system will be "closed" and carry minimal risk? Just because it doesn't happen much (likelihood) doesn't mean the severity of any incident would also be low (quite the opposite, I'd have said). And choosing to ignore a risk because the likelihood is low is akin to security by obscurity. And a lot of people have learned the hard way with that over the years.