Grayling announces Network Rail to use digital signalling

[CdBrux]

What is the difference / similarity with what we know / assume is proposed for TP and what is actually being installed on Thameslink? How much into unknown territory is this going?

 

[Sunset route]

A question for the more technical minded, in level 3 moving Block, how would a stop signal be created so trains can be brought to stand to be cautioned through a section of Line say for bridge strike, persistent trespass, animal incursion user worked crossing misuse? Not a REC call situation.

 

[MarkyT]

A question for the more technical minded, in level 3 moving Block, how would a stop signal be created so trains can be brought to stand to be cautioned through a section of Line say for bridge strike, persistent trespass, animal incursion user worked crossing misuse? Not a REC call situation.

I'd guess there'd be a control 'button' on the screen to 'replace' a 'virtual signal' to stop, so the system would issue no further movement authority to the approaching train for passage beyond the associated marker and the train would slow down and come to a halt at the marker. For L3 moving block I don't know but I think there'll still have to be block markers periodically even if the system allows more than one train in a particular block at a time when in full moving block mode. The blocks wouldn't be important unless you want to set one as a stopping target.

 

[Sunset route]

I'd guess there'd be a control 'button' on the screen to 'replace' a 'virtual signal' to stop, so the system would issue no further movement authority to the approaching train for passage beyond the associated marker and the train would slow down and come to a halt at the marker. For L3 moving block I don't know but I think there'll still have to be block markers periodically even if the system allows more than one train in a particular block at a time when in full moving block mode. The blocks wouldn't be important unless you want to set one as a stopping target.

And there lays a problem knowing when to place a temporary block marker so as not to give the train an adverse speed curve, i.e. Emergency braking. With four aspect signalling I just count back four signals section and pull up the button. watching a train moving on a computer screen with no real points of reference for correct braking will be another problem to add to the list before level 3 becomes reality. I can't see how having marker boards track side with drivers ignoring them in normal running then suddenly obebeying when the signaller in effect replaces a signal/block marker to danger will work. If your running under moving block and the train in rear catches up the train in front and comes to a stand but it's on the approach to a marker board but further back than what drivers do today with signals, how does the driver know what rules are in use. Wait where you are for authority to proceed via the computer, or in fact the signallers is going to caution you through a yet underfund length of section of track because there are no reference points at the beginning or end of section I require you to examine or travel at caution over.

Sorry for the waffle, just wanted to pick people's brains lol

 

[HSTEd]

And there lays a problem knowing when to place a temporary block marker so as not to give the train an adverse speed curve, i.e. Emergency braking. With four aspect signalling I just count back four signals section and pull up the button. watching a train moving on a computer screen with no real points of reference for correct braking will be another problem to add to the list before level 3 becomes reality. I can't see how having marker boards track side with drivers ignoring them in normal running then suddenly obebeying when the signaller in effect replaces a signal/block marker to danger will work. If your running under moving block and the train in rear catches up the train in front and comes to a stand but it's on the approach to a marker board but further back than what drivers do today with signals, how does the driver know what rules are in use. Wait where you are for authority to proceed via the computer, or in fact the signallers is going to caution you through a yet underfund length of section of track because there are no reference points at the beginning or end of section I require you to examine or travel at caution over.

Sorry for the waffle, just wanted to pick people's brains lol

Surely we would not bother with marker boards at all for signalling purposes, and you should simply stop where the computer tells you to stop?

The movement authority that is modified should be communicated to the driver on their panel - and I would imagine the signal system would prevent a caution aspect being set that would cause emergency braking except in situations where it is an emergency?

 

[MarkyT]

I think you've hinted at the scale of work required on rules and regs, procedures, instructions, training etc. for a notional moving block system, let alone the technical challenges! I think these various overlays and hybrids in progress today suggest a mature cautious approach to deployment that reduces project risk and should allow installations to be upgradeable to higher functionality in the future. The one thing missing from the toolbox at the moment is a cost effective ETCS based protection system for secondary and lightly used lines, perhaps representing up to 70% of national route mileage. A limited supervision option would be useful here, based initially on an overlay on conventional signalling to replace of AWS/TPWS and fill in at signals currently not protected, and eventually migrating that to an intermittent 'regional' L3 type solution with virtual fixed blocks, and fixed train detection sensors retained only at junctions and reversing points.

 

[HSTEd]

Well I imagine a transitional ERTMS installation could be designed to be effectively superRETB - allowing practices for that system to be implemented wholesale.

I wonder how much more expensive ERTMS 1LS actually is than TPWS.....

 

[MarkyT]

Well I imagine a transitional ERTMS installation could be designed to be effectively superRETB - allowing practices for that system to be implemented wholesale.

RETB is a bit slow and cumbersome for anything but lightly used isolated corridors, but I appreciate you're talking about an upgraded system that might be better in this respect. I expect the intermittent regional solution will develop for these applications (with or without virtual blocks depending on traffic mix), but I expect it will be as an upgrade to conventional signalling using modular TCB and axle counters. In 2012 NR replaced RETB on the East Suffolk with modular technology, and further rollout of this is planned soon for the Wherry Lines.

I wonder how much more expensive ERTMS 1LS actually is than TPWS.....[/QUOTE]

I don't think it will differ significantly once you have ETCS fitted rolling stock capable of reading the balises. SBB, with Siemens, set up a highly standardised modular retrofitment for all their preexisting signalling, suitable for easy installation and fast commissioning, almost 'between trains'. That's very much like our UK TPWS program rollout in fact, but in Switzerland covering all signals, not just those with the highest risk score.

 

[rebmcr]

All I know is the TMS for my location is running at least 1-2years late and one of the problems is rumours from under the juice rail is correct, is that they are having problems writing the software to deal with multiple junctions where a train standing at junction 1 fouls Junction 2 which in turns fouls Junction 3 type of problem i.e. Looking at multiple conflicting points at the same time and looking again and adapting the plan when the first plan didn't pan out.

This is extremely plausible, as it's almost a textbook example of the sorts of problems that computers really struggle with.

And there lays a problem knowing when to place a temporary block marker so as not to give the train an adverse speed curve, i.e. Emergency braking. With four aspect signalling I just count back four signals section and pull up the button. watching a train moving on a computer screen with no real points of reference for correct braking will be another problem to add to the list before level 3 becomes reality. I can't see how having marker boards track side with drivers ignoring them in normal running then suddenly obebeying when the signaller in effect replaces a signal/block marker to danger will work. If your running under moving block and the train in rear catches up the train in front and comes to a stand but it's on the approach to a marker board but further back than what drivers do today with signals, how does the driver know what rules are in use. Wait where you are for authority to proceed via the computer, or in fact the signallers is going to caution you through a yet underfund length of section of track because there are no reference points at the beginning or end of section I require you to examine or travel at caution over.

As a signaller, all that would be required from you is a position where you want to train to be stopped — it could probably be a kilometre marker, not even a block boundary. The in-cab systems would do the calculation of when a brake application needs to start, and also make sure that the driver is informed in good enough time to avoid any emergency braking. Basically, the 'counting back' would no longer be necessary, because that's ERTMS's job.

Then, 'talking through' would either mean 'tell the driver how far to go' or possibly even 'tell the train to allow the driver to travel X metres'.

 

[Sunset route]

This is extremely plausible, as it's almost a textbook example of the sorts of problems that computers really struggle with.



As a signaller, all that would be required from you is a position where you want to train to be stopped — it could probably be a kilometre marker, not even a block boundary. The in-cab systems would do the calculation of when a brake application needs to start, and also make sure that the driver is informed in good enough time to avoid any emergency braking. Basically, the 'counting back' would no longer be necessary, because that's ERTMS's job.

Then, 'talking through' would either mean 'tell the driver how far to go' or possibly even 'tell the train to allow the driver to travel X metres'.

I know what your trying to say but this is going to be such a new culture of signalling trains in the future, where apart from brand new high speed lines, we are still going to have interactions between members of public using user worked crossings with slow moving vehicles that will still cost too much to interlock with the signalling, that will still need to protecting. So before the train fancy onboard computer does all the calculations based upon when I place a virtual block on the line i need to when it is safe to do so, as a slow moving freight plodding at 40mph can be a lot closer than a regional express doing 90 or 100mph.

That's the next problem I live x I work in y and control trains through z, I've got no idea what the local landmarks are (as I don't now) so I tend you use fixed railway furniture as in signals and points as my point of reference. So with moving block I haven't really got a clue where I need to stop you or how far I need you to go at caution. The only way this is going to work is if every crossing and mean every crossing and every rail over road and rail bridge is easily accessible within the controlling computers drop down lists with a easy way overlaying protection on it.

I'm not saying these are impossible problems, but none the less they are problems that will need solving before we start having trains running on our network that doesn't even use fixed train detection or fixed blocks but solely relies on the trains own radio communication equipment to provide location information to the signaller, who in turn will allow you to cross the line in your slow moving long vehicle at a user worked crossing and just pray that the virtual block was place at the right time that should an approaching train appear it has a nice gentle brakeing curve to bring it to a stand a safe distance from you crossing.

 

[HSTEd]

I know what your trying to say but this is going to be such a new culture of signalling trains in the future, where apart from brand new high speed lines, we are still going to have interactions between members of public using user worked crossings with slow moving vehicles that will still cost too much to interlock with the signalling, that will still need to protecting. So before the train fancy onboard computer does all the calculations based upon when I place a virtual block on the line i need to when it is safe to do so, as a slow moving freight plodding at 40mph can be a lot closer than a regional express doing 90 or 100mph.

But the signaller does not need to know how fast the train is moving to place a safe movement authority.

The signaller places the movement authority - the limits of which are passed to the train, the train then generates its performance curve (being suitably conservative) and displays the speed curve to the driver.

You simply define areas of track that are safe for that train to occupy and the train electronics do the rest.

You simply define a position a metre short of the crossing to be the edge of the block and the train will stop there every time. The position of the crossing will be visible on the signaller's route display - along with all the other fixed infrastructure.
Train positions could then be represented moving along this in a manner comparable to now.

I'm not saying these are impossible problems, but none the less they are problems that will need solving before we start having trains running on our network that doesn't even use fixed train detection or fixed blocks but solely relies on the trains own radio communication equipment to provide location information to the signaller, who in turn will allow you to cross the line in your slow moving long vehicle at a user worked crossing and just pray that the virtual block was place at the right time that should an approaching train appear it has a nice gentle brakeing curve to bring it to a stand a safe distance from you crossing.

If the approaching train is not capable of a service brake application to stop within it's movement authority, outside of an emergency stop order, the entire system has collapsed.

It should not be in a position where it is not capable of stopping within its authority with only its rated service brakeforce.

 

[SpacePhoenix]

Which London termini stations would benefit the most from the implementation of digital signalling?

 

[rebmcr]

So before the train fancy onboard computer does all the calculations based upon when I place a virtual block on the line i need to when it is safe to do so, as a slow moving freight plodding at 40mph can be a lot closer than a regional express doing 90 or 100mph.

The interface should be designed to only offer you options that it already knows are safe.

The only way this is going to work is if every crossing and mean every crossing and every rail over road and rail bridge is easily accessible within the controlling computers drop down lists with a easy way overlaying protection on it.

Yes indeed, or a map with full, certified detail of the kilometre or metre position of all of those features.

I'm not saying these are impossible problems, but none the less they are problems that will need solving before we start

Certainly.

 

[Sunset route]

But the signaller does not need to know how fast the train is moving to place a safe movement authority.

The signaller places the movement authority - the limits of which are passed to the train, the train then generates its performance curve (being suitably conservative) and displays the speed curve to the driver.

You simply define areas of track that are safe for that train to occupy and the train electronics do the rest.

You simply define a position a metre short of the crossing to be the edge of the block and the train will stop there every time. The position of the crossing will be visible on the signaller's route display - along with all the other fixed infrastructure.
Train positions could then be represented moving along this in a manner comparable to now.




If the approaching train is not capable of a service brake application to stop within it's movement authority, outside of an emergency stop order, the entire system has collapsed.

It should not be in a position where it is not capable of stopping within its authority with only its rated service brakeforce.

Moving block in some respects is no different to line worked under the controls of mass automatic signals, the signaller doesn't give authority to the move but the signalling automatically does based on the position of the previous train and what fixed block section it's occupying. With moving block the fixed blocks are gone to be replaced with fixed safe distance to the train in front with both parameters changing all the time. So take the up and down quarry lines on the BML the last authority to move that the signaller does is Earlswood over the fixed interlocking and the next will be Stoats nest junction and from Mark T is saying that will be much the same as there is some sort of need for positive train location before you start swinging points around.

So going back to plan Line over vast miles it looks like that trains will create their own safe authority to move based upon the train in front, but I still maintain at some point for whatever reason the signaller will need to going into the system and place a temporary fixed stop in to protect whatever needs protecting. At the moment I use fixed replacements which I could apply as soon as train passes it or I could of caught it a bit late and now I've applied it when it's just about to leave that signal section a 1000yrds further up the track, that 1000yrds might just make a big difference in moving block with the following train as what was safe with fixed block is not safe with moving block.

 

[Sunset route]

I don't mind being wrong or barking up the wrong tree, I just want to see how the boffins within this forum brain storm around these problems that are taking Europes finest signalling and software engineers a lifetime to dream up and solve. Lol

 

[edwin_m]

I imagine ERTMS L3 will require alternative line blocking facilities. One would be for emergency where if there was a train approaching within the braking curve it would be brought to a stand as soon as possible, and the other would be for more routine blockages and would prevent the signaller applying it if there was a train too close to stop (or perhaps apply it automaticallly once the train had passed through).

I guess also the signaller would see on their display how much track has been reserved for an approaching train (the braking distance plus a safety margin), similar to how a route is shown in white but also applying to sections that work automatically today.

 

[MarkyT]

I'm not convinced we need full moving block as such through most plain line sections, where spacing can be handled perfectly adequately by virtual fixed blocks at similar spacing to typical 3 or 4 aspect signalling. Where moving block will be most useful is at and approaching stations and junctions where it might be overlaid to provide closing up and reduce platform reoccupation time without lots of intermediate block markers being required. In this role it would be used as a kind of 'protected permissive' for admitting a second train into a platform while the one in front is still departing or to allow a tight queue to form approaching a home board perhaps. An advantage of the virtual fixed block concept is it simplifies the calculations and doesn't need so much positional data transfer over the radio which might also be intermittent where that made sense using the stored movement authority feature of regional, which in turn could also help cover any unexpected minor radio dropouts or failures. Furthermore fixed virtual also more closely matches the traditional route setting paradigm, having 'virtual signals' defined at the block markers and to which normal techniques of route setting, replacement etc can be applied. Like L2, fixed virtual ETCS blocks in L3 will decouple board spacing from speed so a line speed increase wouldn't require block respacing as is often the case with lights on sticks.

 

[HSTEd]

So going back to plan Line over vast miles it looks like that trains will create their own safe authority to move based upon the train in front, but I still maintain at some point for whatever reason the signaller will need to going into the system and place a temporary fixed stop in to protect whatever needs protecting.

Not sure why you would engineer it so the train would generate its own movement authorities.
Surely when the route through a junction would set, an authority would be issued for the track beyond the junction - either to the next junction where conflicting routes might exist, or more likely to the position of the next train in front?
As the train end position advanced down the track, authorities would be 'handed back' for the vacated trackwork and every few seconds authorities would be regranted using a predefined set of rules for following a train. [For example: "Previous authority terminated because it reached the end of another authority - check if that authority is still present, if not extend the authority until one of the other constraints is reached"]

And I am not sure why a fixed stop placed with this system is any more difficult to do than in a conventional system?
The route display would have all the relevant landmarks marked on it - and you would simply gilloutine a movement authority (or stop one being issued) at that position, bringing the train to a stop.

Whilst there is a danger that you will force a train into an emergency braking, this is no different that you might do that accidentally by setting a signal to red without giving the train approaching sufficient time to stop.
The computer, knowing the exact position of the train, will however be able to assist the signaller by asking "are you sure" if the signaller attempts to place such a restriction, or by allowing the signaller to place a restriction that only comes into effect when it can do so without causing emergency braking.

This way the only thing the train has to do is stop within the confines of the authority issued by the signal system, and inform the signal control system of its service braking curve to enable emergency braking to be avioded.

 

[edwin_m]

I'm probably rather late with this thought, but I guess by supervising the speed approaching the actual junction location Level 2 would eliminate unnecessary slowing from approach control.

 

[MarkyT]

I'm probably rather late with this thought, but I guess by supervising the speed approaching the actual junction location Level 2 would eliminate unnecessary slowing from approach control.

I did a course over 20 years ago on the Swedish ATP system and that definitely catered for this, able to enforce a precise target speed limit and decelleration curve for each divergence no matter how far from the signal protecting it. It broadcasted speed codes from transponders and loops at and on approach to a signal and included special codes for a speed change deferred for so many metres beyond a signal. I expect ETCS will have similar functionality in its movement authorities.

 

[HSTEd]

The drastic improvement in processing power in recent years certainly helps as it now means a modest workstation can recalculate the authorities for every train several dozen times per second.

 

[DerekC]

I'm not convinced we need full moving block as such through most plain line sections, where spacing can be handled perfectly adequately by virtual fixed blocks at similar spacing to typical 3 or 4 aspect signalling. Where moving block will be most useful is at and approaching stations and junctions where it might be overlaid to provide closing up and reduce platform reoccupation time without lots of intermediate block markers being required. In this role it would be used as a kind of 'protected permissive' for admitting a second train into a platform while the one in front is still departing or to allow a tight queue to form approaching a home board perhaps. An advantage of the virtual fixed block concept is it simplifies the calculations and doesn't need so much positional data transfer over the radio which might also be intermittent where that made sense using the stored movement authority feature of regional, which in turn could also help cover any unexpected minor radio dropouts or failures. Furthermore fixed virtual also more closely matches the traditional route setting paradigm, having 'virtual signals' defined at the block markers and to which normal techniques of route setting, replacement etc can be applied. Like L2, fixed virtual ETCS blocks in L3 will decouple board spacing from speed so a line speed increase wouldn't require block respacing as is often the case with lights on sticks.

A lot of people have argued that Level 3 should initially be deployed as "virtual block" but I still don't understand why. For sure there will need to be markers (and track based detection) at junctions, but on plain track it doesn't make sense to me. To create "virtual blocks" the ETCS RBC has to issue movement authorities to the block markers, rather than to a position set back by the safety margin from the rear of the train in front. But it still has to know where the train in front is so that it can calculate whether the rear of it has cleared the virtual block ahead, so I don't see why it makes the system simpler. There is an operational argument, but in looking at this issue nearly ten years ago the driver reps were less keen on virtual blocks than the signalling engineers. What would probably be needed are decent mile (kilometre?) markers which can be seen and identified clearly from the cab. I understand the point about route setting, but I have a strong feeling that this is going to make the Level 3 system more complex than it needs to be. Systems like SELTRAC happily set routes and monitor movement of trains through junctions without virtual blocks , although they do, as you suggested earlier, have train detection for deadlocking and flank protection purposes.

 

[Roast Veg]

What would probably be needed are decent mile (kilometre?) markers which can be seen and identified clearly from the cab.

Miles and chains in this country I imagine, as silly as it is.

 

[DerekC]

The drastic improvement in processing power in recent years certainly helps as it now means a modest workstation can recalculate the authorities for every train several dozen times per second.

It is now hard to believe that the original SELTRAC system controlled 20 trains quite happily using a minicomputer with 2x64K of memory and a processor running at (I think) 8MHz.

 

[Rational Plan]

It's touching that so much faith is being placed in the Digital Railway, but it relies on Network Rail to implement it. Considering it's history with current resignalling projects, in both time and money I'm not sure why anyone expect to solve any problem. I suspect it soulds like a solution offered to a Ministry as a way spending money on expensive infrastructure.

 

[Emblematic]

Miles and chains in this country I imagine, as silly as it is.

Maybe not, I believe National Rail is implementing ETCS as a fully metric system, so I'd expect locations to be metricated if they need to be referenced. It will probably need some ancient laws to be updated though...

 

[edwin_m]

Maybe not, I believe National Rail is implementing ETCS as a fully metric system, so I'd expect locations to be metricated if they need to be referenced. It will probably need some ancient laws to be updated though...

I understand ETCS on the Cambrian is fully metricated.

 

[Railsigns]

I understand ETCS on the Cambrian is fully metricated.

Not fully. Locations along the line are still measured in miles and chains.

 

[Olaf]

This is extremely plausible, as it's almost a textbook example of the sorts of problems that computers really struggle with.

.

It's not really the software, but rather the verification and validation of the products and final integration. There will also be effort required around the model tuning. This is not rocket science, it is comparable to cases arising in process control in other industries.

Of course, if the vendor is trying to do it on the cheap to keep its margins up, then that adds another dimension.

 

[Olaf]

I know what your trying to say but this is going to be such a new culture of signalling trains in the future ...

Yes. A key problem arising from the adoption of the newer technology is that the approach to operational process has more in common with similar domains in other industries than it has to anything that the operational side of the rail industrial has had experience with to date.