ETCS Level 2 Rollout

[HSTEd]

It should also be noted that the trend towards ECP brakes in the US (whenever the switch finally starts for general traffic rather than block loads) means that trains are increasingly being provided with power lines that run the length of the train.

 

[Dave1987]

Always amuses me reading threads like this TBH. There are a lot of perfect world 'works good on paper' theories on here. What I would like to know is where is the fail safe in an ETCS signalling system. If the whole system relies on the train reporting its position to the signalling system and other trains around it how to you protect against a train having complete power loss or some problem with the antenna so the train stop reporting its position. I don't see where the fail safe is like you have with TCB signalling.

 

[Railsigns]

What I would like to know is where is the fail safe in an ETCS signalling system. If the whole system relies on the train reporting its position to the signalling system and other trains around it how to you protect against a train having complete power loss or some problem with the antenna so the train stop reporting its position. I don't see where the fail safe is like you have with TCB signalling.

ETCS Level 2 relies on continuous train detection by track circuits or axle counters, just the same as TCB does.

 

[HSTEd]

Always amuses me reading threads like this TBH. There are a lot of perfect world 'works good on paper' theories on here. What I would like to know is where is the fail safe in an ETCS signalling system. If the whole system relies on the train reporting its position to the signalling system and other trains around it how to you protect against a train having complete power loss or some problem with the antenna so the train stop reporting its position. I don't see where the fail safe is like you have with TCB signalling.

If the train stopd reporting it is likely to be in the same movement authority area it was reported in.
It would require multiple failures for it to be otherwise

 

[Sunset route]

So apart from the ins and outs of the mechanics of how it works, does anyone have else have or hold any theories informed or other, that after Thameslink core where the next route is liking to get ETCS?

 

[Dave1987]

There is basically little difference in the fundamental way in which this level 2 thing works compared to normal TCB with three and four aspect signalling. So the ultimate fail safes are still there.

 

[HSTEd]

Meanwhile regional is sort of like a super-RETB.

 

[TheKnightWho]

There is basically little difference in the fundamental way in which this level 2 thing works compared to normal TCB with three and four aspect signalling. So the ultimate fail safes are still there.

Well if a train just stops reporting, expect the system to send out stop signals to any trains behind it. This would be the case for level 3 too.

 

[edwin_m]

On Level 3 there must be a mechanism to apply emergency brakes after a very short period if the train loses contact with the system. This ensures that the train will stop within the movement authority it was most recently given, and other trains will be prevented from entering that section of track until the errant train is found and the system reset in some way. With Level 3 not being defined yet the details of how this happens probably haven't been decided but it must be something along those lines.

This is obviously a major reliability/performance issue if it happens too often, but not directly a safety issue. Like all signalling systems failure is on the side of safety but can cause a lot of disruiption.

 

[LAX54]

Well if a train just stops reporting, expect the system to send out stop signals to any trains behind it. This would be the case for level 3 too.

If its anything like most modern technology, and not only railways, if one bit fails, it all fails ! so on NR everything will come to a stand for miles around

 

[Dave1987]

Still not convinced as to where the ultimate fail safes are. The way ETCS is being reported is that the train broadcasts its position to other trains to provide the block section so you get the moving block idea. If the system relies on the train reporting it's position constantly either you need that positional reporting to be extremely reliable and I'm still not convinced as to where the ultimate fail safe is.

 

[Dave1987]

If its anything like most modern technology, and not only railways, if one bit fails, it all fails ! so on NR everything will come to a stand for miles around

Indeed, reading MR mag it seems they are currently concerned with the enormous amount of data transfer required and that that data transfer needs to be very very reliable.

 

[edwin_m]

Still not convinced as to where the ultimate fail safes are. The way ETCS is being reported is that the train broadcasts its position to other trains to provide the block section so you get the moving block idea. If the system relies on the train reporting it's position constantly either you need that positional reporting to be extremely reliable and I'm still not convinced as to where the ultimate fail safe is.

Yes, although I think the trains report position to a central point and that then allocates movement authorities (reserved sections of track ahead) for each train.

The protocol will be something like this. Each train reports its position every few seconds (and yes accurate positioning is important, derived from odometry between fixed balises but need to take account of wheel slip/slide). The central system needs to hear from each train every few seconds and if it doesn't it orders that train to stop. Similarly each train will expect to hear from the central system every few minutes and will stop if it doesn't get an update.

 

[Dave1987]

Yes, although I think the trains report position to a central point and that then allocates movement authorities (reserved sections of track ahead) for each train.

The protocol will be something like this. Each train reports its position every few seconds (and yes accurate positioning is important, derived from odometry between fixed balises but need to take account of wheel slip/slide). The central system needs to hear from each train every few seconds and if it doesn't it orders that train to stop. Similarly each train will expect to hear from the central system every few minutes and will stop if it doesn't get an update.

Still sounds like ideal world thinking. Wheel slip/slide is not predictable, and if the system relies on regular intense positional reporting from trains it sounds like an incredibly complex system that could fail very easily.

 

[bramling]

Yes and Yes but the first was LU deciding that the cost of the PPP procured scheme was too expensive and cancelling it, then the second was discovering that going for cheaper scheme was going to meet the requirements then going to a 3rd supplier for roughly the same cost as the first.

Being cynical, the cancellation of the Westinghouse contract appeared more like LU got excited at taking back control of Metronet and this was their chance to flex their muscle. A shame, as the Westinghouse product is rather superior to the Thales one.

 

[bramling]

Still sounds like ideal world thinking. Wheel slip/slide is not predictable, and if the system relies on regular intense positional reporting from trains it sounds like an incredibly complex system that could fail very easily.

Which is exactly what happens with the Seltrac system used on the Jubilee and Northern lines. Quite simply it struggles to cope with wheelslide.

 

[LAX54]

Still not convinced as to where the ultimate fail safes are. The way ETCS is being reported is that the train broadcasts its position to other trains to provide the block section so you get the moving block idea. If the system relies on the train reporting it's position constantly either you need that positional reporting to be extremely reliable and I'm still not convinced as to where the ultimate fail safe is.

Of course there is still a Signaller in overall charge of things.

 

[HSTEd]

Still sounds like ideal world thinking. Wheel slip/slide is not predictable, and if the system relies on regular intense positional reporting from trains it sounds like an incredibly complex system that could fail very easily.

All signalling systems are enormously complex, unless it is something like a train staff.
That is just the way this field works.
One wire coming lose or breaking in a traditional system can collapse the entire thing or even cause a wrong side failure.

You can put balises every few hundred metres, or less, to keep reseting the odometers.
After all balises are just giant RFID tags, they are cheap and require no power or comms systems.

 

[snowball]

Can't a train use Doppler radar to measure its ground speed and so correct for wheel slip/slide?

 

[Tim M]

Can't a train use Doppler radar to measure its ground speed and so correct for wheel slip/slide?

ETCS does. Also used on the Westinghouse system on the Victoria Line.

 

[alxndr]

All signalling systems are enormously complex, unless it is something like a train staff.
That is just the way this field works.
One wire coming lose or breaking in a traditional system can collapse the entire thing or even cause a wrong side failure.

I think the point being made is that traditional signalling is designed in such a way that most failures will be right side, or protected wrong side, unless impracticable to design it that way.

For example, a broken track lead, breakdown of an IRJ, or power supply failure will cause a track circuit to show occupied. An axle counter smashed off a rail, or a cable cut, will cause it to show occupied. A black signal will put others back. A failed electromagnet will result in a driver receiving a horn regardless of signal aspect. A failed crossing barrier should end up down, not up. Go through any diagram and imagine a contact failing, you'll probably find some sort of contingency made for that.

I have a huge amount of respect for the designers of this stuff, especially interlockings. They're almost beautiful in their intricacy and completeness.

I'm not saying signalling is infallible, it's not. But it generally does a very good job of making sure things don't go catastrophically wrong if it can be helped, or two+ faults need to happen in order for a wrong side failure.

I'm sure this is something that's been thought of, and I just don't know enough about it to see the failsafes, but it does seem less reassuring.

 

[HSTEd]

It is best to imagine the system like a sort of electronic token.
The 'token' in this case is a electronically transmitted authority to occupy a certain piece of track, as designated by odometry, radar and balises. The signalling system will only release the authority to another train once the previous train has transmitted a signal surrendering it. Otherwise it will assume that the train remains within the authority area granted.

 

[edwin_m]

I think the point being made is that traditional signalling is designed in such a way that most failures will be right side, or protected wrong side, unless impracticable to design it that way.

For example, a broken track lead, breakdown of an IRJ, or power supply failure will cause a track circuit to show occupied. An axle counter smashed off a rail, or a cable cut, will cause it to show occupied. A black signal will put others back. A failed electromagnet will result in a driver receiving a horn regardless of signal aspect. A failed crossing barrier should end up down, not up. Go through any diagram and imagine a contact failing, you'll probably find some sort of contingency made for that.

I have a huge amount of respect for the designers of this stuff, especially interlockings. They're almost beautiful in their intricacy and completeness.

I'm not saying signalling is infallible, it's not. But it generally does a very good job of making sure things don't go catastrophically wrong if it can be helped, or two+ faults need to happen in order for a wrong side failure.

I'm sure this is something that's been thought of, and I just don't know enough about it to see the failsafes, but it does seem less reassuring.

I tried to explain it a few posts back. In principle it's no different from how a Solid State interlocking ensures communication with its trackside equipment. If the trackside module doesn't hear from the interlocking every second or so then it will lock all affected points in existing positions and put all its signals to red. If the interlocking doesn't hear from the module it will assume all tracks are occupied and points undetected etc. The details of the communication are however rather different as it's over radio to trains rather than over cable to modules, with a greater risk of "dropouts".

 

[bahnause]

Can't a train use Doppler radar to measure its ground speed and so correct for wheel slip/slide?

We use Doppler Radars on our equipment, but it struggles with different surface conditions (especially snow).

 

[DerekC]

So apart from the ins and outs of the mechanics of how it works, does anyone have else have or hold any theories informed or other, that after Thameslink core where the next route is liking to get ETCS?

Yes, this thread has got lost in the minutiae of how ETCS works and opinions on whether it is really fail-safe or not.

The logical place to go is ECML, if ATP is being life extended for GWML. The Class 700s are fitted, the Class 800s are fitted and for goodness sake I hope that the new 7xx trains for Moorgate will be fitted too! The plan that was in place pre-Waboso was for KX to about Barnet to be fitted initially with ETCS as an overlay over conventional signalling, to allow training, familarisation with the system and reliability bugs to be ironed out, then take the plunge and roll ETCS Level 2 only (no visual signals) out northwards. This made sense to me. If we can't get to the point of eliminating the visual signals then ETCS in its current form isn't worth doing and we should invent something better.

 

[glbotu]

Yes, this thread has got lost in the minutiae of how ETCS works and opinions on whether it is really fail-safe or not.

The logical place to go is ECML, if ATP is being life extended for GWML. The Class 700s are fitted, the Class 800s are fitted and for goodness sake I hope that the new 7xx trains for Moorgate will be fitted too! The plan that was in place pre-Waboso was for KX to about Barnet to be fitted initially with ETCS as an overlay over conventional signalling, to allow training, familarisation with the system and reliability bugs to be ironed out, then take the plunge and roll ETCS Level 2 only (no visual signals) out northwards. This made sense to me. If we can't get to the point of eliminating the visual signals then ETCS in its current form isn't worth doing and we should invent something better.

But surely the real point of ETCS is increasing capacity, not removing poles with lights on top. I mean, that's a nice, cost-saving side-effect, but it's not the primary goal. It certainly makes sense to keep some poles with lights on top, so that we don't have to go to the vast expense of requiring literally every train that needs to run up certain bits of mainline with expensive ETCS equipment. You can do it eventually (like say, 30/40 years time) when all stock is fitted, because it's a requirement of new stock, but while things like the 387s and 365s are needed for capacity, let's also save money by not retrofitting a bunch of trains.

 

[MarkyT]

But surely the real point of ETCS is increasing capacity, not removing poles with lights on top. I mean, that's a nice, cost-saving side-effect, but it's not the primary goal. It certainly makes sense to keep some poles with lights on top, so that we don't have to go to the vast expense of requiring literally every train that needs to run up certain bits of mainline with expensive ETCS equipment. You can do it eventually (like say, 30/40 years time) when all stock is fitted, because it's a requirement of new stock, but while things like the 387s and 365s are needed for capacity, let's also save money by not retrofitting a bunch of trains.

That's exactly the approach in the Thameslink core, adding capacity for the new ETCS Siemens trains which will form the vast majority of the service. Retaining signals and AW/TPWS as well should allow a Siemens train to proceed through the core whose ETCS or ATO subsystems are faulty or disabled in some way. Even in the peak, a single train having to do this will not effect capacity significantly, with following trains under ETCS being able to follow just as closely behind as they would if the faulty train was still under ETCS (a benefit of retaining conventional train detection in the enhanced system). This is a much better outcome than the faulty train being able to go no further and blocking the line at St Pancras or Blackfriars! Older non fitted trains, track machinery etc can be planned to pass through outside peak hours, as on the trackside both modes are switched on and available at all times to appropriately equipped rolling stock.

 

[DerekC]

But surely the real point of ETCS is increasing capacity, not removing poles with lights on top. I mean, that's a nice, cost-saving side-effect, but it's not the primary goal. It certainly makes sense to keep some poles with lights on top, so that we don't have to go to the vast expense of requiring literally every train that needs to run up certain bits of mainline with expensive ETCS equipment. You can do it eventually (like say, 30/40 years time) when all stock is fitted, because it's a requirement of new stock, but while things like the 387s and 365s are needed for capacity, let's also save money by not retrofitting a bunch of trains.

The idea that ETCS automatically increases capacity is wrong. It only does so if the number of blocks is increased. Without lineside signals you can do that easily - you can think of it as a multiple aspect system with an unlimited number of aspects. If you keep lineside signals you either have to lose the capacity benefit by keeping the same blocks for both systems or adding intermediate blocks which only work for fitted trains. The latter is what Thameslink has done in its core. Either way, the cost goes up and in the latter case the cost (and complexity) goes up even more. The Thameslink approach is OK for short sections of route where the capacity benefit can really be used, but it doesn't wash its face for multiple use main lines where capacity is usually constrained by train service mix, infrastructure layout and so on. To argue the whole thing on the basis of train fitment cost misses the point. There are 5000 or so trains (i.e. units and locos) in the UK and about 28000 signals. The infrastructure savings (if you get rid of visual signals) will outweigh the train fitment costs - and even more so if you can make this a step to Level 3. To think that ETCS Level 2 overlay - an expensive add-on with very little benefit - will be funded so that in 30-40 years time there will be a cost saving is unrealistic.

The depressing thing is that all these calculations have been gone through in 2003 for the SRA, in 2007 or so for Network Rail and now it seems we are going to do them all over again - with the same result, that ERTMS rollout stands still. What we really need is a serious Level 3 pilot. That has been clear for at least ten years, but all we get is talk, and ever changing plans which are never acted on!

Sorry - end of rant!!

 

[Dave1987]

But Derek from what I've read this Level 3 thing is just mere fantasy at the moment. A sort 'wouldn't it be nice to have'. And this is what I mean by lots and lots of looking good on paper but in the real world doesn't work. They still don't know what the actual capacity gain from Level 2 is even though they keep banding out this 40% figure even though the people who know what they are talking about say it will be closer to 10-15%. You need systems that actually work in the real world and not just on paper and in an engineers head.

 

[Roast Veg]

Question about Thameslink ECTS - in the case where a 700 is worked into a section behind another, but after a colour light signal, will the traditional signal show clear or danger? It's fairly inconsequential so long as the in cab display is correct, but will the interworking between systems change the colour lights or not, does anybody know?