Digital Railway

[moggie]

It has taken nearly twenty years to get to this point. Railway Safety (remember them?) and then the SRA (remember them?) started to look at ERTMS and develop an ERTMS rollout programme in about 2001, under the management of Mr Waboso. The SRA did recognise the problems that the MoU tries to address and created a budget with a separate funding line for train fitment, but when the SRA was abolished in 2005 and the programme handed over to Network Rail the money was all thrown into the general signalling pot. The SRA did manage to launch the Cambrian early deployment, but that's all that has happened - except Thameslink (see below). One problem is that no manager of a major signalling scheme wants to be first.

Precisely. Snails pace. Cambrian will already be approaching technical obsolescence with the various age / marks of kit employed. I doubt any 'manager' would be embracing the opportunity to carry out any significant alteration to the signalling arrangement there as obtaining compatible extras might involve a bill far greater than the estimated costs. Fortunately (and probably one reason the line was selected as the test bed) there's unlikely to be any demand necessitating such a change.
As for 'being the first'. Someone has to be but as you observe they won't be if we're talking purely ETCS as this now exists. The system employed to deliver ETCS functionalities might be a first of type depending on whoever is awarded contracts but I suspect most (not all) Project Directors wouldn't have a clue whether they were dealing with a first of type anyway being as they will be appointed prior to award of equipment supply contracts. So to them it will be just another challenge to overcome, that is until the problems start compounding.

Crossrail has created its own problems by implementing about the most complex system you could imagine, with CBTC in the core section and transitions to ERTMS for Heathrow Express (and maybe GWML eventually) and to AWS/TPWS for everything else. Thameslink, which picked ERTMS for its core, works already, so citing Crossrail as an ERTMS problem is misleading.

The key problem which is still not solved is that in order for ERTMS to have a business case you have to eliminate the conventional signalling, but that makes the switchover from one to the other very risky. The obvious way to solve that is to put ERTMS in as an overlay, with conventional signals as well - but then it costs more and doesn't deliver capacity benefit unless you do complex things with intermediate blocks. I don't see that the MoU does anything to help that.

Indeed it has but WHO (metaphorically) specified or advised on what Crossrail engineers are now trying to do? I suggest the very same types pushing the ETCS roll out across large parts of the network. One thing is certain. Those that advise and specify are generally nowhere near the job once design, construction and testing of the chosen system starts. It's not their specialism - which is why major projects employing 'novel' technology offerings often end up with dogs breakfast solutions such as Crossrail signalling seen as a whole system.
As for the cost savings. It's all very well stating these are obtained by eliminating conventional signalling but THAT's the reality. You simply cannot ignore the reality of system migration from the conventional arrangement to the new ETCS arrangement on an existing railway, so it is illusionary to pretend otherwise. Just as one cannot avoid stagework and enabling works required to execute large scale conventional signalling projects. You cannot simply wish it away or park it off the balance sheet somewhere convenient. Migrating from one system to another while maintaining train operations has to be done and done effectively with minimal disruption to the service. Crossrail's Heathrow branch is a good example of ineffective system migration planning - late and costly delivery caused by poor execution of system migration from ATP to ETCS. Yet, it is precisely those pushing the 'digital solution' including those in NR who are responsible for effective (on both cost control and maintenance of train service) migration planning enabling ETCS roll out. Another example - ECML. Currently we have Kings Cross resignalling. Much needed system renewal which paves the way for ETCS introduction with the introduction of ETCS compatible equipment BUT, inclusive of lineside colour light signals and the physical structures. So where are the 'savings' here I ask? The answer is there are none. But a much needed signalling renewal has resulted in a system adapted for current train service needs and compatible with future L2 ETCS. But NO savings. Indeed it can be argued significant extra cost of the ETCS requirement compared with conventional signalling renewals without an ETCS component. Of course in reality the costs are spread across many budgets such that the 'signalling costs' exclude significant ETCS components as costs for these are tucked away in other money pots. Yes it's a complex process rolling out whole system ETCS to a working railway but that's why those responsible are paid well to effectively scope, plan specify and manage their project requirements.
Thameslink was a first and was to all intents successfully implemented but is it yet to be employed in daily use at it's full operating capability?

Then take a look at system architecture. About the ONLY component of a conventional signalling schemes avoided by ETCS (migration strategy ignored) is lights on sticks and structures which hold them in the air. Train detection, switch and crossing, level crossing, power, lineside data comms - all common to conventional / ETCS. Interlocking - same. Front end control systems - similar. Communication systems - ditto. BUT, ETCS requires a cubicle full of processing equipment on board each and every train - not cheap. Then there's clearly an augmented system integration and testing regime to undertake. Far more involved than that required for conventional signalling systems which by comparison is negligible. As we've learned system integration is an ongoing sore for Crossrail accountants. Then of course there TMS at the control centre which is usually specified as part of an ETCS solution and augmented train detection to facilitate closing up of trains enabled by ETCS. All at significant cost across the whole ETCS system.

So when one adds up what is saved from conventional, the cost of lights on sticks / gantries offset by added train detection and more routes / MA's for ETCS and then add the rolling stock ETCS kit and all the processes required to validate it's safe operation. I am far from convinced the headline 'savings' at outturn are real. They might be at the estimating stage before a single line of signalling code is written and it might be that costs are 'adjusted' to place them out of the ETCS budget. But let's not kid ourselves that ETCS L2 is the holy grail. L3 might be - it it can be obtained for the cost of train detection architecture.

 

[Belperpete]

Crossrail has created its own problems by implementing about the most complex system you could imagine, with CBTC in the core section and transitions to ERTMS for Heathrow Express (and maybe GWML eventually) and to AWS/TPWS for everything else. Thameslink, which picked ERTMS for its core, works already, so citing Crossrail as an ERTMS problem is misleading.

The interface problems are not what are currently holding up CrossRail. It will initially be commissioned as a self-contained system - the interfaces come later. And the interfaces to AWS/TPWS are actually fairly simple.

 

[Belperpete]

Rather than start a new thread I thought this one could be revived. Network Rail and DfT have recently signed a Memorandum of Understanding, committing themselves to support the Digital Railway's "Long Term Deployment Plan" which was published in June 2019:

https://www.networkrail.co.uk/wp-co...Digital-Railway-Long-Term-Deployment-Plan.pdf

The MoU is too long to quote in full, but essentially it sets out how NR and DfT wil work together to ensure that digital signalling (i.e. ERTMS) is deployed when current signalling assets are life expired. Here are some extracts which give the flavour:



I would be interested to know what others think of the document.

There was an article in IRSE News (I think, may have been Rail Engineer) awhile back explaining the need for a long-term strategy. If I remember correctly, train fitments needed to be already well underway to meet the timescales set. And resignallings also needed to be started early, to avoid serious overloading of the suppliers at later stages. I remember thinking at the time that is never going to happen.

Its all very well having a strategy, but the proof of the pudding is going to come when DfT puts its hand into its pocket and starts actually paying for all this work. Particularly for early works that are going to cause serious disruption to travellers without offering any immediate benefits.

 

[MarkyT]

There was an article in IRSE News (I think, may have been Rail Engineer) awhile back explaining the need for a long-term strategy. If I remember correctly, train fitments needed to be already well underway to meet the timescales set. And resignallings also needed to be started early, to avoid serious overloading of the suppliers at later stages. I remember thinking at the time that is never going to happen.

Its all very well having a strategy, but the proof of the pudding is going to come when DfT puts its hand into its pocket and starts actually paying for all this work. Particularly for early works that are going to cause serious disruption to travellers without offering any immediate benefits.

As fleet renewals continue, the trains are all going to end up capable soon anyway, especially now the DB freight fleet is underway. I believe we need an intermediate stage of limited supervision to replace and augment AWS/TPWS like Belgium and Switzerland have done to replace their legacy protection systems. Meanwhile level 2 can continue to be implemented on major corridors and major city approaches with or without lights on sticks. The thing is the cost of 'conventional' has come down enormously with led signals and axle counters, especially on rural railways where only simple straight posts are required, and that includes the maintenance overhead, which has decreased markedly. There's little obvious business benefit in migrating. Obviously on wide multitrack approaches to termini in big cities like London, the calculation is totally different with the huge gantries etc avoided. Also, L3 is somehow thought of by some as being able to offer unlimited capacity but this is absurd as the number of trains being run simultaneously on a rural railway for instance is not really determined by the signal system but by the single line occupancy and the positions of passing loop, and even if you could run 20 trains flighted a mile apart in one direction along the Cambrian fro instance, the computational and communications requirements to reliably manage all those radio movement authorities is bound to influence cost, complexity and reliability, quite apart from the problem of where to put the trains and the unattractiveness of the service in the other direction while the notional massive convoy passes!

 

[Roast Veg]

It seems logical to me that RETB and other token based lines ought to be converted when the time comes, as per the Cambrian, rather than providing them with lights on sticks.

 

[LOM]

It seems logical to me that RETB and other token based lines ought to be converted when the time comes, as per the Cambrian, rather than providing them with lights on sticks.

The two remaining RETB areas in Scotland were renewed in 2013/14 giving them a 40 year lifespan.

 

[MarkyT]

The two remaining RETB areas in Scotland were renewed in 2013/14 giving them a 40 year lifespan.

An ETCS based system that could be a logical replacement for RETB was installed on a very rural line in Sweden. This uses the ETCS level 3 based technique of virtual fixed blocks, avoiding much fixed train detection, and the RETB-like technique of intermittent rather than continuous movement authority transmission. Radio block centre and interlocking functionality were combined into the same installations. Conventional fixed train detection was retained around points for extra safety however, so small islands of trackside signalling equipment were necessarily retained at station passing loops using the track - train radio system for communication with the interlocking along as well as the movement authority (again rather like the later modifications to RETB).

 

[Belperpete]

As fleet renewals continue, the trains are all going to end up capable soon anyway, especially now the DB freight fleet is underway.

What the article I was referring to showed, was that if you waited until all stock running through an area was ETCS-compatible, before resignalling that area with ETCS, then you inevitably pushed all your resignalling schemes too far back to achieve the ETCS target date, and in addition created a mountain of resignalling work close to the target date that the supply industry could not hope to meet. In order to flatten-out the resignalling curve, signalling work would need to be carried out earlier, which meant stock would need to be ETCS-fitted earlier than currently-planned fleet renewals allow for. Fitting of ETCS to trains with only a few years life does not make sense, so inevitably it meant renewing stock earlier than currently planned.

As I said, the proof of the pudding will be when DfT actually start paying for this. Currently they seem to be concentrating on schemes where the stock has already been renewed with provision for ETCS fitment, or is due to be renewed. There is no way that they will hit the ETCS target date that way.

L3 is somehow thought of by some as being able to offer unlimited capacity but this is absurd as the number of trains being run simultaneously on a rural railway for instance is not really determined by the signal system but by the single line occupancy and the positions of passing loop, and even if you could run 20 trains flighted a mile apart in one direction along the Cambrian fro instance, the computational and communications requirements to reliably manage all those radio movement authorities is bound to influence cost, complexity and reliability, quite apart from the problem of where to put the trains and the unattractiveness of the service in the other direction while the notional massive convoy passes!

The same applies on any railway, not just rural single lines. The signalling system affects headways between following trains. The track layout, and in particular stations and conflicting junctions, also affects headway. And a mixture of service types, eg stopping and non-stopping and freight, is another factor. The signalling system does little to help with these other factors. As you say, it is all very well having a convoy of trains following one another closely down a length of straight line, but what happens to them all when they get to the end of the line?

To take just two examples, both Thameslink and CrossRail have required substantial remodellings and resignallings of their adjacent conventional railway to cope with the frequency of trains being thrown at them.

 

[Re 4/4]

Not sure how comparable this is, but Switzerland was going to convert the whole country to ETCS a while ago and that didn't work out, we have some form of it (L2 I think) in the new 'base tunnels' though. SBB's 'Smartrail 4.0' new grand plan seems to be actually taking a step back from ETCS everywhere for a moment, because

• It turned out that fitting it (particularly loco conversions I think) is massively more expensive than planned.
• It actually reduced capacity on some lines. Reason is, to get it approved with today's safety standards, they had to model slightly less steep braking curves than the ones they currently use to make sure the worst possible train under poor adhesion conditions can still always stop short of the required point.

 

[Belperpete]

SBB's 'Smartrail 4.0' new grand plan seems to be actually taking a step back from ETCS everywhere for a moment, because

• It turned out that fitting it (particularly loco conversions I think) is massively more expensive than planned.
• It actually reduced capacity on some lines. Reason is, to get it approved with today's safety standards, they had to model slightly less steep braking curves than the ones they currently use to make sure the worst possible train under poor adhesion conditions can still always stop short of the required point.

Same problems experienced here with fitting GW ATP onto the old HSTs. Took much longer than expected, as virtually every one was different in some way, and so cost much more than expected.

And the Cambrian's timetable had to be padded to suit ETCS, which won't allow up and down trains to enter loops simultaneously (unlike the old RETB) due to the lack of overlaps. They were supposed to be amending the ETCS to fix this, but it got quietly swept under the carpet. However, the problem of new signalling giving less capacity than the old due to modern standards is not unique to ETCS.

 

[MarkyT]

Not sure how comparable this is, but Switzerland was going to convert the whole country to ETCS a while ago and that didn't work out, we have some form of it (L2 I think) in the new 'base tunnels' though. SBB's 'Smartrail 4.0' new grand plan seems to be actually taking a step back from ETCS everywhere for a moment, because

• It turned out that fitting it (particularly loco conversions I think) is massively more expensive than planned.
• It actually reduced capacity on some lines. Reason is, to get it approved with today's safety standards, they had to model slightly less steep braking curves than the ones they currently use to make sure the worst possible train under poor adhesion conditions can still always stop short of the required point.

So with characteristic Swiss inventiveness, SBB invented a new system called Level 1 Limited Supervision (LS) and got it incorporated in the international standards. They then proceeded to replace all their legacy protection system transponders with active eurobalises and fitted all their older traction with a balise reader and a small box of tricks that uses these to replicate the old systems' functionality on-board. Meanwhile, all the newer traction with built-in full ETCS functionality received software upgrades to read the new eurosignum and eurozub transponders and emulate the system under the new limited supervision mode. While the older traction is restricted to the LS classic routes and can't venture onto their level 2 equipped trunk main lines, modern full ETCS equipped traction, from any EU nation, can at least theoretically from a signalling perspective go anywhere on the standard gauge network, so Switzerland has been 100% interoperable since project completion 2017. They got modern nationwide digital train protection, network-wide, by decoupling it from signalling renewals. Belgium has achieved something similar with TBL1+. France already had it, arguably, with their KVB which has very similar balises that can be read by a standard eurobalise reader. Other nations, notably Germany, are also investigating the approach and even the Berlin S-Bahn have implemented Eurbalises as a modular replacement for their mechanical trainstop arms. Level 2, with or without signals, is undoubtedly the best option for fast and high capacity lines, but L1 LS could offer a cost-effective solution for much of the UK, perhaps up to three-quarters of total route mileage I suspect, replacing TPWS and AWS with similar and possibly enhanced functionality that could, for instance enforce a continued slowdown after a cancellation of a distant warning and prevent a speed up after an overs-peed trap, both for a given distance or time after passing the transponder. I suggest a trial on the Wherry Line, where there is shiny new signalling, and new ETCS-ready swiss designed rolling stock. I suspect Stadler may already be very familiar with the approach.

 

[Belperpete]

replacing TPWS and AWS with similar and possibly enhanced functionality that could, for instance enforce a continued slowdown after a cancellation of a distant warning and prevent a speed up after an overs-peed trap, both for a given distance or time after passing the transponder.

So the system would still enforce braking for a set time even if the signal clears - sounds very restrictive to me. Don't see how it could be for a given distance, unless the unit has some way of measuring distance, which would increase the complexity.

What you are proposing is effectively equivalent to another national TPWS fitment programme, whereby every signal that currently has AWS/TPWS would need to be fitted with a balise, and every train would need to be fitted with a balise reader. It would cost in the same order as the TPWS fitment programme, and probably take as long. I am not convinced that it would be cost-effective on its own, but could be as part of a long-term migration strategy.

However, it would have its own migration issues. You couldn't remove the existing AWS & TPWS trackside equipment when you laid the balises, until all stock running over that line were equipped with balise readers, and likewise every train would need to keep its existing AWS & TPWS receivers in addition to its balise reader until every line it may run over is equipped with balises. How did the Swiss stage their changeover?

 

[MarkyT]

So the system would still enforce braking for a set time even if the signal clears - sounds very restrictive to me. Don't see how it could be for a given distance, unless the unit has some way of measuring distance, which would increase the complexity.

What you are proposing is effectively equivalent to another national TPWS fitment programme, whereby every signal that currently has AWS/TPWS would need to be fitted with a balise, and every train would need to be fitted with a balise reader. It would cost in the same order as the TPWS fitment programme, and probably take as long. I am not convinced that it would be cost-effective on its own, but could be as part of a long-term migration strategy.

However, it would have its own migration issues. You couldn't remove the existing AWS & TPWS trackside equipment when you laid the balises, until all stock running over that line were equipped with balise readers, and likewise every train would need to keep its existing AWS & TPWS receivers in addition to its balise reader until every line it may run over is equipped with balises. How did the Swiss stage their changeover?

The functionality would be similar to how German PZB ('Indusi') works today. The enforced deceleration at a distant warning would have a distance limit after which it would release and at an overspeed the no acceleration rule would also apply for a given distance before release. With such a digital system these values and speeds could be modified on a site by site basis like the Swiss ZUB system which provided a site based deceleration profile. A different cab indication could be transmitted and different responses engineered for each aspect in the euro-AWS for finer control. All these speed supervision and distance measurement functions should be practical to achieve in a ETCS fitted unit with its odometry and tight speed measurement and control. I don't know how the Swiss changed over their legacy rolling stock, but in UK the latest stock emulates AWS/TPWS within its ETCS systems anyway using plug in STM sensors for the legacy track transponders so is notionally compatible with both systems already and could retain the legacy transponder readers as long as is expedient. There is already a programme to equip all stock with a long term future with ETCS, and I wouldn't change that at all. Changeover of the infrastructure should be easy once all stock working the areas is suitable and the actual timescales would thus be very flexible. Benefits would be additional safety from enhanced functionality and a fully failsafe system (TPWS isn't intrinsically failsafe), wider fitment at all signals with active balises, and all speed boards with passive balises, greater reliability and damage resitance of balises, easier maintenance of standard lighter and easier to handle equipment available from different suppliers, and an end to heavy AWS magnets.

 

[Ianno87]

Same problems experienced here with fitting GW ATP onto the old HSTs. Took much longer than expected, as virtually every one was different in some way, and so cost much more than expected.

And the Cambrian's timetable had to be padded to suit ETCS, which won't allow up and down trains to enter loops simultaneously (unlike the old RETB) due to the lack of overlaps. They were supposed to be amending the ETCS to fix this, but it got quietly swept under the carpet. However, the problem of new signalling giving less capacity than the old due to modern standards is not unique to ETCS.

Although the Cambrian trial has flushed this out as being an issue, this means that (in theory) any unintended restrictive consequences can be specified and designed out in future installations.

i.e. it's not an automatic consequence that ETCS makes everything more restrictive; it won't be more restrictive if you factor that into the requirements and design in the first place.

 

[Bald Rick]

Although the Cambrian trial has flushed this out as being an issue, this means that (in theory) any unintended restrictive consequences can be specified and designed out in future installations.

i.e. it's not an automatic consequence that ETCS makes everything more restrictive; it won't be more restrictive if you factor that into the requirements and design in the first place.

And it is, of course, much less restrictive in many other ways. Not least that there don’t need to be any speed restrictions specifically because of signalling issues.