ETCS Discussion

[MarkyT]

Mod Note: This thread has been split from the discussion on the Dovey Junction loop.

Local RUG Newsletter has pictures

see http://sarpa.info/news.html

Gareth, I like your article on the 'Dodgital railway'. ETCS cannot in itself provide extra capacity on the rural railway. It was the additional and extended passing loops that did that on Cambrian.

Personally I think we should be looking at ETCS Level 1 LS (limited supervision) for perhaps up to 70% of the UK railway system. This is the method being used to introduce ETCS, network wide, in Switzerland during 2017 by completing the replacement of all their legacy track and vehicle transponders with ETCS standard equipment that replicates the functionality of the historic cab protection and warning systems.

It should be noted that the ETCS balise equipment is immensely more reliable and robust than our flimsy TPWS grids, although our legacy AWS magnets are pretty bulletproof, although expensive. The technique allows older rolling stock to to be far more easily modified to run on lines so equipped without the modern 'tight loop' traction and braking control systems necessary for a more sophisticated 'ATP like' standard Level 2 to function.

The genius of the SBB approach is in influencing the technical committees to incorporate the LS mode into the international specs so all modern fully fitted traction from all over Europe will in the future also be able to run over the entire LS fitted Swiss network as well as 'conventional' Level 2 lines. There IS Level 2 in Switzerland, but it is reserved for the ultra busy and high speed main lines dedicated to Intercity traffic and heavy international freight where its capacity benefits can be truly justified. The older traction, with only LS functionality emulating the legacy protecting systems, has no access to these lines of course. SBB worked very closely in cooperation with DB and SNCF in getting LS into the specs and I suspect these administrations also see a future in this technique for much of the secondary rail system.

Widely used French and Belgian balise-based systems are already very similar. The UK has not entirely ruled out LS but a report by RSSB dismisses it as a widespread method of improving safety, although acknowledged it could be a cost effective technique to apply ETCS in areas where the significantly higher costs of full Level 2, with all it's radio coverage, detailed infrastructure mapping and data maintenance difficulties were difficult to justify.

I think it might be justified to enhance safety by equipping all the intermediate plain line headway signals that were not fitted with speedtraps and trainstops as part of the TPWS project. Only signals protecting junctions were so equipped as well as bufferstops and certain speed restrictions. Rolling out LS in many areas (maybe as much as 70% of the UK network) instead of committing to network wide Level 2 could be an effective method of decoupling rollout from the wider signalling renewals and control centralisation programme with all its risks and funding challenges.

I think Digital Railway as a project had lost its way at one point but under the new leadership of David Waboso it's finding it's feet again. Sensibly, they've abandoned the Wherry lines L2 project which threatened to be an expensive repeat of the Cambrian, i.e L2 applied inappropriately to another rural route. Quick and inexpensive modular resignalling will be applied instead. LS would be a very simple and inexpensive bolt-on to that, and because of SBBs sterling work would be completely compatible with the new Stadler trains on order for Anglia.

 

[Gareth Marston]

Gareth, I like your article on the 'Dodgital railway'. ETCS cannot in itself provide extra capacity on the rural railway. It was the additional and extended passing loops that did that on Cambrian.

Personally I think we should be looking at ETCS Level 1 LS (limited supervision) for perhaps up to 70% of the UK railway system. This is the method being used to introduce ETCS, network wide, in Switzerland during 2017 by completing the replacement of all their legacy track and vehicle transponders with ETCS standard equipment that replicates the functionality of the historic cab protection and warning systems.

It should be noted that the ETCS balise equipment is immensely more reliable and robust than our flimsy TPWS grids, although our legacy AWS magnets are pretty bulletproof, although expensive. The technique allows older rolling stock to to be far more easily modified to run on lines so equipped without the modern 'tight loop' traction and braking control systems necessary for a more sophisticated 'ATP like' standard Level 2 to function.

The genius of the SBB approach is in influencing the technical committees to incorporate the LS mode into the international specs so all modern fully fitted traction from all over Europe will in the future also be able to run over the entire LS fitted Swiss network as well as 'conventional' Level 2 lines. There IS Level 2 in Switzerland, but it is reserved for the ultra busy and high speed main lines dedicated to Intercity traffic and heavy international freight where its capacity benefits can be truly justified. The older traction, with only LS functionality emulating the legacy protecting systems, has no access to these lines of course. SBB worked very closely in cooperation with DB and SNCF in getting LS into the specs and I suspect these administrations also see a future in this technique for much of the secondary rail system.

Widely used French and Belgian balise-based systems are already very similar. The UK has not entirely ruled out LS but a report by RSSB dismisses it as a widespread method of improving safety, although acknowledged it could be a cost effective technique to apply ETCS in areas where the significantly higher costs of full Level 2, with all it's radio coverage, detailed infrastructure mapping and data maintenance difficulties were difficult to justify.

I think it might be justified to enhance safety by equipping all the intermediate plain line headway signals that were not fitted with speedtraps and trainstops as part of the TPWS project. Only signals protecting junctions were so equipped as well as bufferstops and certain speed restrictions. Rolling out LS in many areas (maybe as much as 70% of the UK network) instead of committing to network wide Level 2 could be an effective method of decoupling rollout from the wider signalling renewals and control centralisation programme with all its risks and funding challenges.

I think Digital Railway as a project had lost its way at one point but under the new leadership of David Waboso it's finding it's feet again. Sensibly, they've abandoned the Wherry lines L2 project which threatened to be an expensive repeat of the Cambrian, i.e L2 applied inappropriately to another rural route. Quick and inexpensive modular resignalling will be applied instead. LS would be a very simple and inexpensive bolt-on to that, and because of SBBs sterling work would be completely compatible with the new Stadler trains on order for Anglia.

Problems today with ETCS and yesterday. Same driver had movement authority denied automatically by computer as unit was approaching Newtown from Caersws on approach to points into loop. Signalers at Mach couldn't explain why. 1G35 actually reversed for a bit between 1039 & 1040/1041 markers yesterday.

 

[MarkyT]

The advent of TPWS has reduced the safety risk of SPADs by about 70% compared to previously. Therefore the safety "budget" available to spend on replacing it by something better is only 30% of what it was then - and the case for TPWS was pretty marginal once the traditional rounds of gold-plating had taken place.

So ERTMS isn't going to be justifiable on safety grounds alone - it needs to reduce costs compared to a modular signalling alternative, or provide useful other benefits such as extra capacity (somewhere it is needed!) or higher speeds.

Agreed that is exactly what the RSSB report said. Importantly TPWS equipment is all still relatively new. The question is what could replace it in the longer term. The assumption today is that ETCS, L2 or otherwise, will eventually replace the the entire signalling system including current protection systems. I argue that fully specced L2 is overkill for most of the railway network and has been proved to be significantly more expensive and time consuming to engineer than conventional resignalling. Perhaps some despecced reqional variant will be deployed like HSTeds Swedish example, but the pilot line there was not an unqualified success as the train operator involved was not able to convert the rolling stock economically and withdrew passenger services entirely. So much for cost savings through technology!

ERTMS Regional provides the benefits but also handles the signalling system.

Which seems to be perfect for this scenario. ERTMS 1LS still requires an entire classic signalling system be installed.

The signalling system is very much still there in 'Regional' as applied in Sweden. It just has LESS trackside equipment. In this respect it is not dissimilar to modern conventional signalling applied to simple rural and secondary routes. Perhaps most noticeable in Regional, signals are replaced by block markers as in L2, but today modern lightweight LED signals in themselves are no longer the major capital expense or maintenance liability of old, as long as they're simple examples not requiring major structures. The latest Unipart Dormans units, for example, are almost entirely maintenance free with a self cleaning finish to the lens and an extraordinary lamp life.

As to train detection, the Swedish installation retains conventional technology at junctions and passing loops for 'deadlocking'. This is a prerequisite of power operated points to prevent any possibility of them being commanded to move under a train. This is also the case now on remaining RETB lines in Scotland, where the obsolete train operated hydropneumatic points have been replaced by conventional point machines. For long plain line block sections the modern technique of axle counters is ideal as there is no intermediate equipment between the extremities. By comparison, typical late 20th century colour light signalling would have required separate track circuits every kilometer or so through the section, each with its own relays and feed sets housed in equipment cabinets located throughout the section and linked together with cabling to carry a summating repeat circuit for the entire logical block. Although in the signalbox it would be indicated as one continuous track circuit, there would be a large amount of equipment distributed along the trackside, all requiring periodic maintenance and testing. With axle counter sensors are only provided at the extremities of the logical block, so all this intermediate equipment and its associated maintenance is simply no longer required.

In regional, movement authority is by radio, but crucially unlike L2 the system only requires radio coverage at passing loops, junctions, and any intermediate plain line block boundaries, rather like RETB only requires gauranteed coverage at its token exchange points.

Regional has a limited population of trackside equipment and to avoid the high cost of continuous trackside cabling it communicates between the centralised interlocking and the trackside object controllers via the same radio system it uses to convey movement authority to the train. Using radio is not a unique selling point however and 'conventional' digital signalling can also use radio comms for this theoretically. Not surprising when 'reqional' is in fact engineered entirely from the same signalling components as any other system and has a similar architecture. The fact it has no trackside signals is a fairly minor point in overall costs today. In fact a modern 'modular' installation would be a very good interim measure to eventually convert to a regional ERTMS variant using the same underlying systems.

 

[HSTEd]

ERTMS does have other advantages that are not present in conventional resignalling however - for example it eliminates all concerns about sighting of signals, which removes a constraint on route speed and also eliminates the need for endless banner signals all over the place. Which would be nice if we could get that benefit without enormous cost.

 

[MarkyT]

ERTMS does have other advantages that are not present in conventional resignalling however - for example it eliminates all concerns about sighting of signals, which removes a constraint on route speed and also eliminates the need for endless banner signals all over the place. Which would be nice if we could get that benefit without enormous cost.

Short range sighting on final approach to the block markers is still a concern, but I agree with your point generally and banners are expensive. They are most prevalent in areas of high speed running, on main lines and particularly fairly twisty ones like WCML. They exist on less important lines today clearly as well, usually where bridges, tight curves or platform canopies obscure a signal approach. They're more likely in hilly areas than in the Norfolk Broads I suggest. Much better to indicate in the cab when you're approaching a block marker, or a signal for that matter (another 'LS' feature that could alternatively be communicated via a balise perhaps). A banner on its own structure could easily equal the cost of the signal it is repeating.

 

[HSTEd]

Short range sighting on final approach to the block markers is still a concern, but I agree with your point generally and banners are expensive. They are most prevalent in areas of high speed running, on main lines and particularly fairly twisty ones like WCML. They exist on less important lines today clearly as well, usually where bridges, tight curves or platform canopies obscure a signal approach. They're more likely in hilly areas than in the Norfolk Broads I suggest. Much better to indicate in the cab when you're approaching a block marker, or a signal for that matter (another 'LS' feature that could alternatively be communicated via a balise perhaps). A banner on its own structure could easily equal the cost of the signal it is repeating.

The really interesting part of this work, and Regional especially, is that it provides for the possibility of secondary routes being able to obtain speeds more normally anticipated on the main line without expensive multi-aspect signal systems, banner repeaters and similar.

Indeed with tilting trains you could eventually get to the point where any branch could have significantly improved speeds, limited only by track curvature and level crossings.

 

[MarkyT]

Indeed with tilting trains you could eventually get to the point where any branch could have significantly improved speeds, limited only by track curvature and level crossings.

Ideally it would become a simpler trade off between the increasing track maintenance and renewal costs of higher speeds versus the commercial benefits of journey time and stock utilisation. On single lines it's more difficult however as passing loops may be in the wrong place for a faster regular interval service. In any case signalling on rural routes is rarely the big issue per se. More likely it's the slow turnouts at intermediate passing loops (where most if not all trains stop) and the dreaded foot and farm crossings. If it affects signalling at all, raising the speed usually means 'moving the distant' out to the new braking distance without any other substantial changes to interlocking. The long blocks of most rural railways mean there simply aren't that many signals to move. The spacing between many older signals designed for the 'steam age' decades ago already contains some margin for speed increase for better braked more modern passenger vehicles. In many cases that has already been exploited for lightweight trains such as sprinters.

 

[HSTEd]

What benefits does ETCS Level 2 have over Regional however?
Regional appears to be able to operate using logical block boundaries that can be made almost arbitrarily short (balises are cheap after all).
In situations with short blocks the infrastructure costs might approach Level 2 but Regional seems likely to be cheaper in most situations.

 

[MarkyT]

What benefits does ETCS Level 2 have over Regional however?
Regional appears to be able to operate using logical block boundaries that can be made almost arbitrarily short (balises are cheap after all).
In situations with short blocks the infrastructure costs might approach Level 2 but Regional seems likely to be cheaper in most situations.

To build very high capacity with a 'regional' system you'd need the continuous radio coverage of L2, so no saving there. Train length verification is the other challenge. Sweden got away with it for their fixed virtual blocks on their very rural line because they were able to assume a fairly short maximum train length without it affecting capacity materially for the application.

For high capacity scenarios if you had variable train length, maximum capacity would demand accurate knowledge of the position of the rear of each train. If you assumed everything was 12 car for instance, that could affect capacity if you actually operated a mixed length fleet, and 12 would also become an absolute limit, so no freight or engineering trains exceeding this without special arrangements. The safety issue is if a 12 car misreported its length as a 4 car then following traffic could approach much too too close behind and a rear end collision would be more likely.

Whereas balise based (or wiggly wire) train detection can work well on a metro with fixed train length, I think we're a long way away from abandoning conventional fixed train detection methods on the main line for following train separation. I also believe we'll always use track circuits or axle counters for deadlocking junction areas and directional locking over single lines, just as TfL does on it's Seltrac equipped lines.

 

[QueensCurve]

I think Digital Railway as a project had lost its way at one point but under the new leadership of David Waboso it's finding it's feet again. Sensibly, they've abandoned the Wherry lines L2 project which threatened to be an expensive repeat of the Cambrian, i.e L2 applied inappropriately to another rural route. Quick and inexpensive modular resignalling will be applied instead. LS would be a very simple and inexpensive bolt-on to that, and because of SBBs sterling work would be completely compatible with the new Stadler trains on order for Anglia.

Thank you for your well informed, interesting and informative post.
--- old post above --- --- new post below ---

As to train detection, the Swedish installation retains conventional technology at junctions and passing loops for 'deadlocking'. This is a prerequisite of power operated points to prevent any possibility of them being commanded to move under a train. This is also the case now on remaining RETB lines in Scotland, where the obsolete train operated hydropneumatic points have been replaced by conventional point machines.

I wasn't aware of the fact that the train operated hydropneumatic points had been replaced with conventional power operated machines on RETB lines.

This presumably requires a power supply and long line control to the point machines. Does it not negate the advantage of RETB.

An update on this would be very interesting.

 

[Railsigns]

I wasn't aware of the fact that the train operated hydropneumatic points had been replaced with conventional power operated machines on RETB lines.

They haven't been, not even a single set.

 

[najaB]

I wasn't aware of the fact that the train operated hydropneumatic points had been replaced with conventional power operated machines on RETB lines.

They haven't been, not even a single set.

Interesting, I've seen it claimed in several places that they have been.

 

[Railsigns]

Interesting, I've seen it claimed in several places that they have been.

Where, out of interest?

And, yes, I am 100% certain that they haven't been.

 

[najaB]

Where, out of interest?

And, yes, I am 100% certain that they haven't been.

I'll need to have a hunt, it was just something I've seen while generally Internet browsing. And I'd trust you over 99% of random Internet sites any day of the week.

 

[MarkyT]

They haven't been, not even a single set.

OK thanks for that info. My bad . I know replacement was being talked about nearly a decade ago when I was still in the industry. I just assumed it must have happened by now along with the radio updates, TPWS additions etc. just as it had on the NSTR lines. I guess the ex-NSTR examples provided a good source of spares for Scotland.
--- old post above --- --- new post below ---

I wasn't aware of the fact that the train operated hydropneumatic points had been replaced with conventional power operated machines on RETB lines.

This presumably requires a power supply and long line control to the point machines. Does it not negate the advantage of RETB.

Subject to Railsigns clarification about the Scottish RETB lines, power was always required for the facing point indicators. There is also a plethora of TPWS related equipment now at each loop. The power supplies are obtained locally and backed up by batteries for resilience. Control of the TPWS grids and indicators is achieved by monitoring and responding to the token activity on the radio system, rather than having a dedicated control link out to each loop site. See here:
http://www.park-signalling.co.uk/downloads/RETB_Product_Information_Sheet_(April 2016).pdf

A notional point machine installation could use the same radio monitoring method to set a 'virtual route' locally for the appropriate movement. At remote loops on the NSTR lines the point machine commands and TPWS grid controls together with their visual indications are derived from the conventional physical token withdrawal activity at each site. My 'point' is that with all this ancillary equipment (indicators, point machines, train detection) required at each loop today there is little or no cost advantage in using these exotic systems on rural lines over a well engineered modular TCB installation using LED signals and axle counters.

 

[Roast Veg]

But what about in the longer term, where widespread ECTS adoption brings cost of maintenance and spares down? At present yes, taking on a signalling system at all in these areas is a cost outlay because of the necessary power consumption.

 

[MarkyT]

But what about in the longer term, where widespread ECTS adoption brings cost of maintenance and spares down? At present yes, taking on a signalling system at all in these areas is a cost outlay because of the necessary power consumption.

Using axle counters and LEDs dramatically reduces power consumption over track circuits and filament lamps. Significant economy also arises from not having to have lots of intermediate equipment between signals. Individual track circuits can typically cover a section of track about 1km long, so to continuously cover a (say) 10km single line section you'd need 10 separate track circuits spread out along the route with power and cabling requirements throughout. Axle counters already solve that with sensors only required at the extremities of the logical block and nothing between except digital communications (which could be via radio). The biggest power consumer in any system is always going to be point machines. Whatever control system you have you cannot do away with these unless you come up with a replacement train operated device as used on the RETB lines, and these are severely speed limited so would preclude any fast passing and overtaking. In any case the load of a small number of distributed point machines used moderately in a rural area is not excessive, and is easily managed with typical domestic rated supplies that can be arranged locally and backed up with batteries, and, where suitable, solar or wind generation to keep things topped up.

Unfortunately Cambrian L2 demonstrated that the additional requirements of continuous 'signalling quality' GSM-R radio represented a major cost and as reported by Gareth a continuing reliability challenge. There is also a major ongoing data collection management and dissemination issue with conventional L2 that simply doesn't exist with conventional signalling. A precise digital map of the railway has to be surveyed and maintained as the onboard systems continuously require this to calculate their safe speed envelope. Management systems have to be in place to ensure all temporary speed restrictions other short term and permanent changes are taken care of safely.

ETCS L2 has conventional train detection, point machines and needs an underlying signalling system of interlocking and control centre connected by a system of trackside cabling. OK it lacks signals but on a rural line this is no longer a major individual cost item in itself and is far outweighed by the costs of the radio system and data management. That's before you even start factoring in the costs and difficulties of rolling stock modification and the challenges and restrictions of dedicated fleets for new islands of this technology, although new fleets will have ETCS capability built in from new.

I am not dismissive of ETCS/ERTMS at all, it's just that the level 2 implementation we have in UK today is simply far too expensive for much of the rural and secondary network and will likely remain so in the future. We need a cheaper regional variant, and that will likely be an easy low riak upgrade from the kind of modular conventional signalling I have also been talking about, which is always specified today to be 'ETCS ready', and is easily made so by having that common underlying ecosystem of centralised computer based interlocking and control.

 

[Roast Veg]

All very good points, and well thought out. How long would you say it will take before the tide turns on this (either L2 or new-regional replacing RETB)? Will it be once the rolling stock is ready for it alone or do you think the GSM-R costs need to come down as well? Also do you think we should move towards activating the ERTMS equipment on board trains whist retaining lineside signals, or is the risk of inconsistency too great?

 

[HSTEd]

ETCS Regional has the advantage over 'conventional' L2 that intermediate blocks for flighted trains become far more practical - as all you need is an intermediate balise in an area with radio coverage suitable to obtain the movement authority.

This could prove useful in improving the effective capacity of single track lines by allowing multiple trains to follow each other through a block without having to wait for it to clear. (For example a freight coudl be put into a section first and a few minutes later a passenger train could enter the section, making several stops but timed so it catches up at the next passing loop, where it could stack behind the freight train).

 

[MarkyT]

How long would you say it will take before the tide turns on this (either L2 or new-regional replacing RETB)? Will it be once the rolling stock is ready for it alone or do you think the GSM-R costs need to come down as well? Also do you think we should move towards activating the ERTMS equipment on board trains whist retaining lineside signals, or is the risk of inconsistency too great?

Difficult to say exactly how long. RETB has been remarkably long-lived in Scotland already, demonstrating an ability to accomodate multiple upgrades of interlockings, radio systems and addition of TPWS, all without fundamentally changing the underlying principles of operation. The trackside and control infrastructure is now very much a 'Trigger's Broom'! In the absence of any proven affordable alternative system available off the shelf I am sure it can be kept going as long as neccessary on the new frequencies. Once there is exclusively native ETCS capable stock running in the Highlands, then a change may be justified, but I understand GSM-R may stilll be a stumbling block. Coverage on Cambrian required many more GSM base station cells than the previous RETB radio equipment, so in the even more extensive and remote Scottish Highlands, it will probably remain unaffordable as available currently. The Scottish RETB radio infrastructure is now all fairly new having been renewed recently to comply with radio frequency reallocations. It must have at least another decade or so of life left before it's next renewal fund could help contribute towards an alternative system. Of course radio data comms is a fast developing general purpose field and many innovations may occur in that time frame.

As to activating trainborne ETCS equipment alongside signals and legacy warnings and interventions, new train manufacturers are doing this already anyway, using the ETCS onboard computer to emulate AWS/TPWS in the cab and at traction/braking interfaces and a 'specific transmission module' (STM) to interface with the legacy track transponders themselves. This avoids yet another set of independent equipment to accomodate on board to cover the transition to ETCS. An LS implementation would use the native balise reader instead for trackside interface but have the same cab functionality.

Some LS online references :
http://plkold.evo.pl/fileadmin/pdf/ertms_konf/10_sesja_2_referat_03_-_migracja_do_ETCS_LS.pdf

http://www.irse.nl/resources/ETCS_LS_IRSE_Wildenrath-V2.pdf

Report for RSSB :

https://www.researchgate.net/public...pervision_for_GB_application_high-level_study

ETCS Regional has the advantage over 'conventional' L2 that intermediate blocks for flighted trains become far more practical - as all you need is an intermediate balise in an area with radio coverage suitable to obtain the movement authority.

This could prove useful in improving the effective capacity of single track lines by allowing multiple trains to follow each other through a block without having to wait for it to clear. (For example a freight coudl be put into a section first and a few minutes later a passenger train could enter the section, making several stops but timed so it catches up at the next passing loop, where it could stack behind the freight train).

I believe RETB has some of these intermediate plain track section breaks, including shunt limits for use with special shunting tokens in certain locations. We still have the challenge of train length to solve in a more general purpose ETCS solution. If you have radio coverage at your intermediate block, there could also be a radio linked axle counter sensor at the block marker which could safely confirm 'out of section' for all trains including variable length ones such as freight.