Despite having one of the largest subway networks in the world, New Yorkers now experience frustratingly erratic and unreliable service. Underfunding has meant that engineers have been pushing the often-century-old subway signalling hardware decades past its design life, so breakdowns are frequent. This is all too familiar to Londoners who lived through the 1950s to the 1980s, when its transport system atrophied.
This article looks at the role of state for the art train signalling in resolving many of the reliability issues of legacy metro lines, whilst simultaneously adding considerable capacity to them. Another Reconnections article will soon look at these parallels between New York and London’s decline and renaissance in more detail.
In November 2019 the Institution of Railway Signal Engineers (IRSE) held a conference in Toronto, Canada on the topic automatic operation of trains using CBTC signalling. Whilst much of the focus was on North America, understandably given the venue, the subject is just as relevant to London. In particular, the signalling upgrade situation on the New York subway is detailed, as are recent signalling upgrades in London. During the writing of this article came the shock resignation of the head of the New York Metropolitan Transit Authority (MTA), Andy Byford, ostensibly for signalling upgrade related issues.
To address New York’s subway reliability issues, Andy Byford had developed his $40bn MTA Fast Forward Plan to upgrade signalling on all lines, along with other system and bus network improvements. At the core of this plan is state of the art communications based train control (CBTC) signalling, which allows agencies to wring more frequent trains out of existing tracks. CBTC is also being implemented in a number of other metro lines around the world, including a number in London.
George Clark, IRSE President
George Clark, a London Underground signalling engineer, is currently president of the UK based IRSE in a ‘voluntary’ role. He opened the conference by explaining how CBTC has grown in relevance:
- initially developed for and installed on greenfield urban rail lines
- evolved to handle much more complex environments and systems
- now being implemented in many different types of railway systems, such as light rail and mainline railways.
It is still evolving.
He related that initially CBTC was cautiously designed to provide 32 trains per hour (tph) reliably on the Northern, Jubilee and Victoria lines in London. But engineers soon realised that CBTC could achieve more time savings safely. With a bit more work, the Victoria line now achieves 36 tph at peak hours.
London’s Four Line Modernisation (4LM)
We recently covered the 4LM signalling modernisation, as well as the Underground’s Digital Railway efforts in detail, both of which cover the capabilities and implementation of CBTC (called automated train operation – ATO – in those articles) on the Underground, as well as some of the issues.
Note that ATO is the generic term, as is automatic train control (ATC). The Thales version, installed on the Subsurface Railway, Northern and Jubilee lines, is branded TBTC, shorthand for Transmission Based Train Control. The term ATO covers CBTC and TBTC (as well as European Railway Traffic Management System (ERTMS) mainline signalling).
As
stated by a number of presenters at this conference, CBTC evolution is now
principally driven by customers coming up with new signalling scenarios (called
‘use cases’).
Pete Tomlin, another London Underground ex-pat
Pete Tomlin had started his signalling career at London Underground, where he worked on the Jubilee Line CBTC implementation. After that, Andy Byford offered him a job overseas in 2009 driving the installation of CBTC on Toronto’s Line 1 Vaughan extension. Tomlin then followed Byford to New York City in January 2019 to head the CBTC implementation of Andy Byford’s Fast Forward Plan. This will replace the antiquated fixed-block signalling system with CBTC for more than 90% of subway riders in 10 years. This is in addition to the current capital programme which is resignalling the Queens Boulevard lines (with Siemens CBTC), which carries E, F, M and R Line trains.
At the IRSE conference, Tomlin presented his work implementing CBTC in New York. The city’s Metropolitan Transit Authority (MTA) operates 473 stations on 24 subway lines, carrying over 1.7 billion riders a year. There are 818 miles of signalled track, 12,000 signals, 15,000 track circuits, and 200 interlockings, many dating from the 1930’s.
Currently two lines (43 miles) operate under CBTC:
- Canarsie L line (Siemens) since 2007,
- Flushing 7 line (Thales) since 2019.
As well, a third supplier is being qualified (Mitsubishi).
It is no surprise that the Canarsie and Flushing lines are by far the best performing lines on the NYC subway.
Tomlin’s mantra, earned through years of experience, is “Track and truck mounted equipment means trouble”, as undercarriage equipment is exposed to the harsh track environment, which adds to the maintenance burden. Furthermore, maintenance (and training) are key costs in keeping the system operational.
In what is turning out to be a once in a lifetime signalling upgrade, Byford and the MTA plan to eventually have CBTC operational on all of its 24 lines in 20 years. This will improve 90% of passengers journeys, and increase reliability, on:
- 220 miles of CBTC on 11 subway corridors, many with multiple lines,
- including 47 of the 50 busiest stations.
To build some quick wins and momentum, the first five years of the Fast Forward Plan would include five more lines converted to CBTC, upgrading over 50 stations to be accessible, state of good repair at another 150 stations, and redesigned bus routes.
On 13 January 2020 New York announced that the first of next CBTC line upgrades, the 8th Avenue A, C and E lines, had been awarded to Siemens. Having already worked on the Canarsie Line CBTC, this win for Siemens may lead to follow-on contracts in New York.
The next phase of the Fast Forward Plan is to build in 10 years what was previously scheduled to take 40 years.
What’s politics got to do with it?
CBTC’s potential doesn’t come easily or quickly – it requires years of planning, intricate scheduling to keep existing services running, and hard work. Most of all, an on time and on budget CBTC implementation requires experienced hands to guide the way, managers to advocate for and obtain the agency and political buy in, and to keep other projects from impinging on budget and labour. For this reason CBTC experts are a highly sought commodity.
Some politicians like to direct their transport chiefs as if they are staff carrying out their every whim. But public transport, especially rail, is subject to the laws of physics, geography, geology, and technology. Plus the principles of human decency and managing professionals.
The experts fight back
Andy Byford’s resignation from the MTA on 23 January 2020 wasn’t a total shock – he’d actually resigned the previous year in October 2019, only to be coaxed back the same day.
Nonetheless there is only so much a person can take, even a patient and practical man like Byford. It appears that part of the friction between him and his superior, New York State Governor Andrew Cuomo, was due to the latter’s frustration at the nominally slow progress of the CBTC implementations, even though these would be 30 years’ quicker than previous plans.
According to City and State NY, Byford decided to resign when the Governor’s reorganisation plan would have left Byford at only an operational level, without the authority to carry out his Fast Forward Plan.
What Byford’s resignation has done however may have started an exodus of CBTC and other management and experts. The day after Byford resigned, his hand picked CBTC expert Pete Tomlin also resigned – which was completely unexpected. Their last day will be 21 February 2020.
There is no doubt there are already many suitors for Byford’s and Tomlin’s talents, as CBTC is being implemented in metros and LRTs worldwide. Successful cities are based on getting people to their jobs, appointments, classes, entertainment etc quickly and efficiently.
Reconnections doesn’t like to speculate, but with Mike Brown leaving TfL in May 2020, one does wonder if Andy Byford could be a potential candidate for TfL’s top job.
Tight control means lots of debugging
Back to the technical world of CBTC, where tight control is needed is in the integration of the lines, cars, and car components together so that CBTC can optimise performance. All safety-critical moving parts on each car are integrated into the CBTC, such as passenger door operation, to shave seconds off dwell time reliably, consistently and safety. Signalling is sometimes seen to politicians, the media and outsiders as a black art, but it’s a highly sophisticated and highly coupled technology. Installation of automatic train operation (ATO), of which CBTC is one technology, on existing lines can take a decade or more, as lines need to keep operational throughout (so it’s worth reflecting on Governor Cuomo’s intemperacy about timescales).
As a result, integrating and fully testing ATO systems has often been problematic and has gone over budget and past schedule: London, NYC, and other cities. Each ATO implementation must tailored to the unique infrastructure and train characteristics of each line.
ATO also greatly assists railway systems to become more self aware, in terms of self-diagnostics, constant communications with other train systems, and with land based systems.
Extending CBTC beyond
Other IRSE presenters described improvements to CBTC functionality, as suppliers are enhancing their CBTC systems for increasingly diverse and complex rail environments, such as mainline railways:
- controlling different train types concurrently on the same network,
- integrating with different signalling equipment on the same network,
- reducing lifecycle costs by minimizing the number of outdoor components,
- and maximizing energy efficiency.
As cities are getting busier, roads more congested, cities are looking to maximise their rail assets. In cities with older metros, the signalling systems are decades old and are increasingly costly to maintain and keep reliable. ATO offers a step change improvement in train frequency and better reliability.
Public transport agencies want guaranteed quality of services, even in case of problems, especially when ridership grows. This has driven the evolution of CBTC, which now has greater functionality than four or five years ago. Recent enhancements include:
- the ability to access a supervisory control and data acquisition system to determine if track is energized or de-energized
- track junction and work-zone management
- automatic coupling in depot, siding and rescue modes
- more efficient recovery modes in case of software crashes, such as the remote reset of automatic train control computers
- better energy management by optimizing train synchronization at peak hours and more energy efficient speed profiles at off-peak hours
- more efficient crisis management tools in integrated control centres
- a graphical user interface
- redesigned hardware and software platforms for higher availability and better performance.
In the beginning
CBTC began with a loop based system developed by Alcatel SEL (now Thales), called SELTRAC, for the North American SkyTrain systems in the early 1980s. This used inductive loops for track to train communication, introducing an alternative to track circuit based communication.
Despite being started by one company, the benefits of train communications based signalling had widespread appeal. So CBTC has long been an open standard, to the benefit of metros and suppliers, as the competition advances the technology.
CBTC is now the gold standard of rail signalling systems, allowing cities to get up to 40% more capacity out of existing lines, without having to build a new line or dig new tunnels.
CBTC Grades of Automation
CBTC operates under the following grades of automation (GoA):
- GoA 1 – Manual protected operation, no automation (this is the basic fallback operation mode),
- GoA 2 – Semi-Automated Operation (STO), which is automatic train operation (ATO) but a driver present at all times at the front of the train,
- GoA 3 – Driverless Train Operation (DTO), typically a member of staff is on board (who may carry out door opening/closing), but not normally at the front of the train, and in regular operation does not normally have any involvement with train driving,
- GoA 4 – Unattended Train Operation (UTO).
Each Grade falls back to the next lowest level for a managed service degradation.
The following grades were achieved on the following lines:
- GoA2 Toronto’s Scarborough RT 1984
- GoA4 Vancouver SkyTrain 1985
- GoA3 Docklands Light Rail 1987.
For reference, here is another London lines’ automation date:
- GoA2 Victoria Line 1968
Refining the CBTC architecture
CBTC technology is evolving, making use of the latest techniques and components to offer more compact systems and simpler architectures. For instance, with the advent of modern electronics it has been possible to build-in redundancy, so that single failures do not adversely impact operations.
The latest CBTC architecture has less wayside equipment – the diagnostic and monitoring tools have been improved, which makes the CBTC equipment easier to install and replace, and, more importantly, easier to maintain.
There is now off-track signalling equipment as well, in the form of signalling centre(s), to coordinate the trains and optimise operations.
CBTC architecture has evolved to move as much lineside signalling equipment and functionality onto the trains as possible, in order to:
- minimise the number of components to maintain
- reduce wear and tear on sensors and components
- provide much more flexible software control and improved diagnostics.
Finally, CBTC systems have been proven to operate trains more energy efficiently than manual and lesser automated forms of operation.
See some CBTC advances in action
A month before this conference, Thales held an Open House at their signalling division headquarters in Toronto. A Reconnections colleague made this short video of the displays, simulations, and interactivity there.
As an example, San Francisco’s Bay Area’s rapid transit regional rail system (BART) has just awarded the installation of CBTC to Hitachi that will increase capacity 40%, improving reliability, and provide for future ridership increases. CBTC will replace the current fixed-block system from the 1960s.
Not the only advanced signalling game in the world
In Europe, railways had developed over 20 different train control systems, based on their national requirements and operating rules. In particular, the safety critical automatic train protection (ATP) systems can be non-compatible. Any through train crossing European borders has had to be equipped with different ATP systems (such as the first generation Eurostars). Sometimes, it even required changing locomotive or driver at frontiers, as each country typically has its own signalling system for which the drivers have to be trained.
The additional ATP systems take up a much space and weight on-board, and add travel time plus operational and maintenance costs. Unifying the multiple signalling systems would streamline freight and passenger rail services across borders, minimise technical problems, reduce costs, and increase competitiveness.
So in 1989, European Transport ministers decided to develop a single train control system standard to apply across Europe, which became the European Train Control System (ETCS) specification. ETCS is the signalling sub system of ERTMS. Other sub systems include traffic management (ie the control system). ETCS currently provides ATP through the real-time monitoring of movement data, precise train location, and braking curves:
- ETCS Level 0 – line not equipped with any train control (ERTMS/ETCS or national) system.
- ETCS Level 1 – train operating on a line equipped with Eurobalises and optionally Euroloop or Radio infill.
- ETCS Level 2 – train controlled by a Radio Block Centre, with train position and train integrity proving performed trackside with Eurobalises and Euroradio.
- ETCS Level 3 – similar to Level 2 but with train position and train integrity supervision based on information received from the train.
ETCS levels 0, 1 and 2 are NOT moving block – this is reserved for Level 3. Note that the overwhelming majority of installations to date are NOT Level 3.
In 1996, the EU decided that European Rail Traffic Management System (ERTMS) would become the signalling standard for all high-speed and conventional railways.
What is ERTMS?
ERTMS is based on the ETCS, combined with GSM-R (Global System for Mobile Communications – Railways), the radio standard for voice and data communication over a dedicated frequency:
ERTMS = ETCS + GSM-R
Despite their names there is nothing intrinsically European about ETCS or ERTMS. In fact the country with the most miles in operation is China, and there it probably exceeds the entire European track mileage.
ERTMS is similar to CBTC in that both are moving block signalling systems. But ETCS and ERTMS are designed to be truly inter-operative, such that the individual system can be interchanged, and is not proprietary to the supplier.
Similarly, the EU mandates that all new signalling on inter-operative main lines be ETCS, with the option of ERTMS overlay.
Why is this important?
The critical part of ETCS and ERTMS is interoperability. London Underground can be exempted because there is a low chance of a main line railway operating on them. Whereas the East London Line (ELL) is an example of joint NR/LU specification of trains and infrastructure, where some form of ERTMS in its central section might be seen this decade if plans to increase ELL frequencies are taken forward.
Also, despite leaving the EU, the UK use of this standard is unlikely to drastically change. The UK’s Rail Safety and Standards Board (RSSB) also had a significant role writing the ETCS standards. London’s Heathrow Airport tunnels’ Belgian ATP system had to be replaced with an ETCS system to allow Crossrail’s Class 345 trains to go to Heathrow, scheduled for full service in the May 2020 timetable. And to upgrade the Northern City line to Moorgate to ATO, ERTMS will be the only option.
Even Crossrail’s central section should have been ERTMS. It only received an exemption because ERTMS hasn’t yet evolved sufficiently to provide the required frequency, passenger door interfacing, and stopping distance accuracy.
Whilst CBTC has been in operation since the mid 1980s, ETCS and ERTMS are relatively new technologies, starting a decade and a half later. For instance, the stopping accuracy is almost where it needs to be for Thameslink, once ETCS becomes quite cheap solution for many low/medium frequency metro systems.
Not signalling rivals
In Europe, CBTC is only used for metro and light rail lines separate from the main railway network. In the UK, CBTC will likely be used only for urban rail lines, while ETCS and ERTMS will mostly be used for UK mainline railways. In the rest of the world, CBTC is pretty much restricted to metro and light rail systems, plus some North America mainline railways.
CBTC Signalling for LRTs
Perhaps surprisingly, CBTC is also relevant for street-running light rail. Modifying CBTC for LRT lines in mixed traffic segments is a recent innovation, to avoid collisions with vehicles and pedestrians, and to synchronise with traffic lights. This task is an order of magnitude more complex than a fully closed system.
Despite the complexity and considerable software development costs this will entail, this investment is deemed to be sufficiently beneficial, as higher capacity, safer operation, and driverless depot operation will lead to long term savings.
CBTC for mixed traffic LRT requires implementation and integration of front and side sensors for collision avoidance at level crossings and around traffic. In addition, automotive accelerometers and speed sensors are used instead of wheel turn counters for more accuracy, so that wheel slip no longer affects train position accuracy.
A good example of the flexibility and breadth of modern CBTC functionality is Bombardier’s CBTC installation system on the Eglinton Crosstown light rail line in Toronto. The system will incorporate three levels of automation: manual and automatic train protection (ATP) along street areas, automatic train operation (ATO) in the tunnels, and unattended train operations (UTO) in the depot.
Neither snow, nor sleet, nor level crossings can keep us from our appointed runs
Many mainline railways still have level crossings. So obstacle detection needs to be included in CBTC functionality for such lines. Using laser detection and rangefinding (LiDAR) is expensive (€120,000 per crossing), and it cannot always detect small or low objects. They can also affected by adverse weather such as heavy rain, snow etc.
Hence multiple detection systems are required, such as infra-red cameras, along with sophisticated software to reconcile the different views and to filter out nuisance signals.
Further CBTC enhancements planned
Virtual coupling together of trains is being studied to maximise track utilisation, such that a flight of trains is controlled together, not individually. This works ideally with the same or very similar train types, but even braking characteristics can vary between trains of the same class.Obviously, joining and splitting of trains adds another level of operational complexity.
Overall, Britain’s railways are transitioning – slowly – to become a ‘More Intelligent Railway’. Moving block signalling is one basis upon which it will be built.
In conclusion
London Underground and other cities worldwide have proven ATO’s effectiveness in adapting to a breadth of environments and operating regimes. New York will get there, despite personality clashes. The cautious main line environments for automated passenger operations may take longer to resolve, though Thameslink’s ETCS Level 2 is now starting to prove its worth on the core section of cross-London services, at a 20 tph frequency.
“Virtual coupling together of trains is being studied to maximise track utilisation, such that a flight of trains is controlled together, not individually. This works ideally with the same or very similar train types, but even braking characteristics can vary between trains of the same class.Obviously, joining and splitting of trains adds another level of operational complexity.”
Would this include joining/splitting of trains while in motion?
That could be quite an interesting development
The IRSE is the Institution of Railway Signal Engineers
[Corrected. Thanks. PoP]
Whilst in North America proprietary CBTC seems to be the future salvation, I am personally convinced that in Europe (and possibly elsewhere) proprietary CBTC is seen as a necessary long-term stop-gap until the day that ERTMS can accommodate platform edge doors and not be a limiting factor in possible maximum frequency.
America is the land of private and free-market enterprise. In Europe and China governments have more desire to take control and avoid the passenger and taxpayer being seen to be ripped off by suppliers that are effectively a monopoly once they have their feet under the table.
Personally, I think China being a big supplier is almost a non-starter due to Huawei type fears. So that means Europe and North America as dominant suppliers and each, by preference, would initially look to its own suppliers for a solution.
Andy Byford states he may now stay in New York City, but not with the MTA.
LBM and other commentators. Some thoughts:
1) In metro applications CBTC is usually taken to mean “moving block” – i.e. some form of communication of train position back to the control system so that track based detection (axle counters or track circuits) aren’t required.
2) ETCS is the signalling sub system of ERTMS. Other sub systems include traffic management (i.e. the control system). [Updated, cheers. LBM]
3) ETCS levels 0, 1 and 2 are NOT moving block. This is reserved for level 3. The overwhelming majority of installations to date are NOT level 3! [Updated, cheers. LBM]
4) I don’t believe that the specification for ATO for ETCS has yet been published, far less any installations complying with the European specification
5) ATO ETCS installations to date (including Thameslink) have used a “catch all” communications channel called packet 44. (I know the name but be gentle – I’m a mechanical engineer and I don’t really understand what it means except that it’s for individual applications’ special features). Another example is automated pantograph up and down commands at bi-mode trains’ transition locations. The published ATO specification isn’t planning to use packet 44.
6) Virtual coupling is an interesting challenge. On plain line track I can envisage closed loop control systems that take account of different types of trains’ traction and braking capabilities. The big challenge is to form and “unform” the virtual coupling. Let’s envisage an example. To make a point, my example is a the 33 coach train (3 x 11-car Pendolino trains) aimed at taking only one path on the West Coast main line in the UK. The front train is going to Glasgow, the second to Manchester and the third to Liverpool. It is inconceivable to have one platform for all these trains; it would need to be nearly 800m long, so there’d be three trains on three platforms. The Glasgow train would set of first and, after allowing for points to be switched and trains safely dispatched, the Manchester and Liverpool trains would follow. The first train would go a bit slowly to allow the others to catch up and they all proceed towards the point that they diverge. Let’s assume they don’t stop en-route (boggling at dealing with three trains in an 800m platform again and the opportunities for people to be in the wrong portion (well known even on trains with through gangways)). What happens when they diverge? The first train has to decouple, and the second and third trains decelerate to create space for the first train to get away and give the points time to throw to allow the remainder of the train to go the other way. This is repeated ad the next divergence.
The point of this lengthy description is to make the point that virtual coupling might help plain line capacity but plain line capacity generally isn’t an issue; it’s at nodes where it is…..junctions and stations, and there, virtual coupling doesn’t really help.
Good points about the virtual uncoupling operation: for the very little it is worth, I had visualised some sort of revival of the slip coach (trains virtually uncoupled from the back to stop at a station)?
Re: “Virtual coupling”
My interpretation of this was that trains would not actually be physically connected, but under common control (virtually coupled) thus allowing them to run closer together and/or faster by reducing the need to maintain a safe stopping distance. Obviously there are issues there with gradients and different braking profiles that’d have to accounted for.
Does it help capacity? Maybe, for fast long distance services on a spine with differing stopping patterns/end points, especially if control of key junctions is integrated and they’re already grade separated. It sounds to me like a thing that would increase capacity/allow higher speeds on modern high speed lines rather than anywhere else.
One further thought about virtual coupling – I see it frequently on motorways, only there it’s called tailgating. Very dangerous.
“Would this include joining/splitting of trains while in motion?
That could be quite an interesting development”
Isn’t splitting a train in motion called a slip coach (very old capability)?
Also loading and unloading in motion (mailbags)
MilesT…I did once judge a research competition where one of the entries involved trains on adjacent tracks running at the same speed perfectly synchronised so that passengers could change from one to the other whilst they continued down the track at speed. This entry didn’t win a prize
And, of course “Banking” – where special signalling arrangements usually applied, allowing the banker to drop off at, or very close to the top of the climb, as used to be done regularly on the Lickey & Shap inclines.
@Malcolm: It wouldn’t be like tailgating, as the following train would know when the front train is changing speed and what to….
This kind of running would require some kind of radar to allow the following train to maintain the distance between it the leading unit as well as some chatter between the units…
Not impossible, but not simple…
Re: Virtual Coupling
Surely in most scenarios it’s a non-starter. Doing any of the things suggested above would break the fundamental safety principle of the railways – that a train always has a safe stopping distance between it and the train in front.
It’s all very well controlling them together, but as soon as the front train collides with anything (e.g. at a level crossing, or with a trespasser on the track), or derails for any reason you’re looking at any trains close behind colliding with it at speed, likely with deadly consequences. The only scenario I see it being useful is for low speed shunting manoeuvres that are today carried out unprotected or with limited signalling protection (e.g. A loco following a train out of a platform into a siding in order to wait to attach to the front of the next train).
Another point I haven’t seen mentioned yet, is the increasingly apparent issues of ETCS and dead end platforms. It appears that the braking curves programmed into ETCS are extremely conservative (compared to that of legacy signalling, by virtue of it not taking into account local conditions due to its broad standardisation), and in unautomated cases the communication of the braking curve to the driver is far from ideal, meaning drivers end up creeping into stations in order to avoid being tripped (which results in an emergency brake application to a standstill and enormous disruption, rather than just braking automatically until the braking curve is reestablished). The end result is a significantly reduced terminal capacity. For most light rail or metro networks this is less of an issue, because terminal platforms are rare (with the significant exception of London underground), but in the mainline it is a significant problem, that seems to he being largely ignored.
PS: Because of the new design of the website, it took me until now to realise that a new article had been posted, despite visiting several times over the last few days. The enormous header (that takes up the entirety of the screen) completely hides the chronological list – and this article is not yet in it.
Good article LBM. Might be worth noting the US ‘Positive Train Control’ term (PTC) is equivalent to European main line ATP.
Virtual coupling must rely on keeping the units forming a virtual consist following tightly with a very small gap little or no different to that provided by mechanical coupling, so any braking of the forward unit will result in the rear one almost immediately buffering up to make physical contact and the whole mass safely decelerating together.
I don’t think splitting and joining at speed is very likely in the future, but virtual coupling could still have some use for carrying out these operations at stations without the big bump typical of the mechanical equivalent. It could be quicker, more reliable, and for joining might allow boarding to continue in the forward portion while the rear section is buffering up. I can’t see through corridor connections across a virtual coupling being safe though! There might be more value in developing better mechanical coupling systems and precision final positioning in ATO that could give some of the same benefits.
Older legacy national protection systems like UK AWS/TPWS are known as ‘Class B’ systems within ERTMS. Where the old trackside transponders remain, they can be read using special plug in ‘Specific Transmission Modules’ (STM) on board, with additional antennae, etc as required, and the Class B data is input to the train’s ETCS computer for partial supervision. Some countries have implemented alternative schemes where an older train protection scheme is adapted to use Eurobalises instead. This avoids any extra physical hardware on the train. A Eurobalise transmits multiple data packets, and packet 44, known as the ‘reserved national datagram’ can encode the simple signal values from the old system in parallel with other ETCS data packets. Older trains without a modern ETCS computer and its interfaces on board can be equipped with a simple specialised Eurobalise reader that converts the datagram signals. This strategy allows a flexible transition where old ATC transponders co-exist with Eurobalises on the track until all trains have a Eurobalise reader. Using special software, the newer ETCS-compliant trains can emulate the legacy class B system entirely using the standard onboard train computer and cab displays.
Switzerland replaced legacy Signum and ZUB inductors with Eurobalises in a project completed in 2017, when the whole standard gauge network became fully interoperable by exploiting the new ‘limited supervision’ mode now included in the international specifications. Belgium’s TBL system, on which the GW ATP pilot scheme was based, was modified to use Eurobalises in a similar way and France’s KVB was conceived from new as a balise-based limited supervision system.
Where ‘Class B’ systems use standard ERTMS communications (eg Eurobalises) they use ‘packet 44’ for their non-standard data. In addition to the Thameslink ATO scheme, another UK example is the WCML TASS (Tilt Authorisation and Speed Supervision, used by Pendolinos and Super Voyagers). This demonstrates that Class Bs are not always legacy systems eventually to be superseded. Strategically, functions of packet 44 systems are meant to be integrated into the general scheme if they have a long term value, so packet 44 can also be thought of as a prototyping area. That is how the ‘limited supervision’ mode made it into the standards from work done in Switzerland, and Thamelink’s ATO experience will inevitably find its way into a standard implementation in due course.
Another example of a Class B with a long life expectancy is the train stop system used by Berlin’s S-Bahn. The mechanical arms and trip cocks are being replaced by Eurobalises in the new ZBS control system. This is not transitional pending ‘full ETCS’ in the future, but a fully featured long term solution. The trackside and train use standard ETCS components but with bespoke software, although any ETCS train could, in theory, be switched to work in ZBS mode given appropriate emulation software.
@DM1, 21 February 2020 at 20:35
In UK there’s already an issue with very slow approach to terminal buffer stops, first appearing with the so-called defensive driving policies, then locked in by TPWS fitment at these locations which is very restrictive. In ERTMS, especially when supplemented with ATO, I would have thought this could be optimised. Each train calculates its own safe braking curve based on known track characteristics and the extent of forward movement authority. Theoretically, a train with better brakes could approach more confidently.
I suppose another way of thinking about ‘virtual coupling’ is as a level 3 moving block overlay on top of an otherwise fixed long block railway. Platoons of trains could be authorised to follow each other, crucially still maintaining minimum braking distance between them, but allowing multiple trains in a single fixed block. The trains would still need good low latency vehicle to vehicle communications to manage this cooperatively between themselves rather than all processing going via the trackside. Only the lead train would get the conventional movement authority. Others in the convoy would get an order to follow a specific train. Conceptually this gets round many of the problems associated with moving block as the processing and communications would largely be distributed on the vehicles.
VM1….I’m sure it’s not coincidental that many of the early adopters of ETCS relied on their legacy systems in and around terminal stations. In addition to the points you make, each train in these areas requires a continuous GSM-R channel and until the advent of GPRS or some other technique this was challenging to achieve.
One thing that confuses me each time I’m reading about CBTC is that it’s never really clear to me whether the term “CBTC” denotes one particular signaling system (which multiple vendors can supply hardware for), or a loose family of conceptually similar but technically distinct and mutually incompatible systems.
What I read in e.g. the Wikipedia article sounds most like it’s a loose family of systems, though that is never really said in so many words. On the other hand, in material about one particular line or transit system, the term CBTC is invariably used in context that sound like it’s the technical designation for the particular signaling system the organization has chosen to install.
This LR article seems to place itself somewhere between the two chairs. On one hand it says: “So CBTC has long been an open standard, to the benefit of metros and suppliers …” which sounds like CBTC is a single standard, where you go to the market and contract to have a signaling centre built and then go to the market and buy some trains, and the only thing you need to explain to each of those vendors about what the other vendor provides is is the four letters “CBTC” and each can then look up in that open standard to find frequencies, modulation, data formats, semantics?
But later the article goes on to explain that ETCS/ERTMS is interoperable between different vendors — implying (though again without saying it in so many word) that CBTC is not. And then I’m confused again …
Could someone who actually knows please make a declarative statement saying that CBTC is one or the other?
HENNING MAKHOLM…Personally I thought LBM explained it quite well! But here goes on trying to respond:
There are lots of different signalling systems around that claim to be CBTC. In general, CBTC stands for “Communications-Based Train Control” and the essential feature – as I mentioned earlier in the comments – is that trains report their position to the system to deliver a much more accurate position than is the case with track detection.
Metro CBTC systems are generally proprietary systems from suppliers such as Alstom, Bombardier, Hitachi STS, Siemens and Thales. In general, these systems use different architectures and different communication systems which mean that trainborne equipment from one supplier won’t work with the trackside eqipment of the other. Indeed, there is no guarantee that trainborne equipment from a supplier in a given city will work with the same supplier’s trackside kit in another city. On London Underground for example, the Metropolitan line Thales system is somewhat different to that on the adjacent Jubilee line, so one line’s trains cannot be signalled on the other line’s tracks.
ETCS was an industry initiative to break out of all this to meet an aspiration of the European Union that has since become law. The intention is that trainborne equipment from one supplier will work with the trackside equipment from any other ETCS supplier. To achieve this, the means of communication (beacons/readers, and GSM-R radio*) are defined and mandated, as are all the messages and protocols that will be used to transmit messages to and from trains. Even the packet 44 (referred to by Mark Townend) which is used for a multitude of purposes – some not envisaged when the system was developed – is managed and controlled.
ETCS levels 1 and 2 still use track detection of trains and ETCS level 3 is true CBTC. ETCS level 2 uses the radio to provide in-cab signalling in the form of a continuously updated movement authority.
* I recall seeing that one operator (KAZAHSTAN) at least has implemented ETCS using Tetra radio.
100ANDTHIRTY
24 February 2020 at 16:20
Finland is also implementing ERTMS using Tetra radio. Many fewer base station ‘cells’ are required to cover a given distance compared to other solutions, particularly important in a large nation with widely spaced centres of population and difficult access to intermediate equipment, especially in the depths of winter.
UIC, the International Union of Railways, not just in Europe but around the World, was first formed in 1922 and recognised many decades ago that individual nations’ railway administrations had become prisoners of their legacy signalling and train protection systems. They realised that at each cycle of renewal, railways were prone to proprietary lock-in to their latest suppliers who could often force widespread renewals on their own commercial terms by their planned obsolescence tactics. With different, incompatible systems on each national network, organising cross-border working was a difficult enterprise, requiring time-consuming and expensive loco and crew changes at each border encountered en route, or undesirably complex multi-system equipped traction.
The UIC concluded that interoperability, particularly at the key train-track control interface, would become a key tactic of their organisation in promoting its core mission “to promote rail transport at world level and meet the challenges of mobility and sustainable development.” Railway administration clients would be able to take advantage of more competition in supply of more standardised products and systems on track and vehicles that would work seamlessly together within nations and across borders.
In Europe the UICs interoperability objectives coincided happily with the EC/EUs interests in facilitating and increasing cross-border trade and rail operator competition so they enthusiastically cooperated in developing the ERTMS concept with European funding. The EUs force of law enabled national railways to insist that suppliers offered compatible systems within Europe, and the core principles and more standardised products have since also found markets outside the EU.
ERTMS is now becoming a mature technology after two decades in development, and all major suppliers are now offering serious capabilities in the field. There is no realistic way back to the proprietary incompatibilities of the past, nor any enthusiasm for that. Would any modern customer deliberately want to limit their supply base and risk lock in and early obsolescence tactics? ERTMS is by no means perfect nor even ‘complete’ today but realistically it is the only show in town. It is, in its favour, modular, flexible and extensible and most importantly an open standard system that no individual supplier can monopolise in any particular market.
Some interesting points here….and a couple of corrections…
The 20tph currently provided on Thameslink is actually under conventional signalling. There are only a handful of trains operating each day in ETCS Level 2 (L2) (and possibly ATO should the driver choose) and these are outside the peak hours. We have actually squeezed 24tph through the core during disruption under conventional signalling too though not often and not consistently.
The Northern City Line (Moorgate – Finsbury Park) is being converted to signals away L2 ETCS but will not have ATO (more akin to the Cambrian than Thameslink core).
Note the comment “Thameslink’s stopping accuracy is nearly where it needs to be”…it IS where’s it needs to be. It is incredibly accurate, time after time when running in ATO.
There was some interesting research by the RSSB a few years ago noting the high cost of implementing L2 systems on rural lines and wondering whether it is actually worth it (T1043 for those with access to Sparkrail via RSSB). They discussed the possibility of using a ETCS L1 Limited’s supervision system in these cases. Basically a limited number of signals would be fitted with infill loops and fixed signals would be used but with ETCS providing ATP. This would be more likely in rail corridors between L2 systems.
It will be interesting to see how difficult it is to get two RBCs interacting and also how 700s would cope with different ETCS systems in the Thameslink core (overlay) and East Coast Main Line when that eventually comes online (signals away). Both the core and Cambrian line have one RBC each – although mainland European countries have systems with RBCs interacting I do believe this can cause performance issues. And then there’s the problem of different ETCS baselines…. interesting times to work in rail operations without doubt.
Ah here it is…
https://www.researchgate.net/profile/Tom_Endersby2/publication/299762180_Viability_of_ETCS_limited_supervision_for_GB_application_high-level_study/links/5705087c08ae44d70ee304b6/Viability-of-ETCS-limited-supervision-for-GB-application-high-level-study.pdf?origin=publication_detail
Thanks, 130. That clears up the confusion.
I suppose that means the statement in the article that “CBTC has long been an open standard” is mistaken? At least can’t be what I’m used to “open standard” meaning.
HENNING MAKHOLM, 29 February 2020 at 15:16
Perhaps the CONCEPT of train control with ubiquitous radio, cab signalling, moving block type concepts, ATO, integration with other safety systems like tunnel ventilation, platform screen doors etc can no longer be considered proprietary as a whole as some manufacturers may have tried to claim in the distant past. The different systems aimed at metros in particular and their various detailed methods of achieving the functionalities required are still very much protected intellectual property, however.
On the other hand, with computerisation it is becoming increasingly easy theoretically for a train from one manufacturer to emulate the functionality required for another manufacturer’s control system without physically needing some of that manufacturer’s hardware integrated into the rolling stock design, and many suppliers are also converging on very similar trackside hardware too, with the proprietary Siemens CBTC being used on the tunnel section of the Elizabeth line, for instance, utilising digital beacons identical to standard Eurobalises. Clearly, as a non-ERTMS system, the messages being passed need not comply with the European standard but the overall message size limits and transmission protocols are identical. This means the trains’ standard Eurobalise reader hardware and antenna can be used for that part of track-train communications.
Britain’s first digital railway takes major step forward as funding and partners announced in Rail Business Daily.
“Network Rail has confirmed Siemens and Atkins as its partners in a major programme to introduce in-cab signalling on the southern section of the East Coast Main Line – a scheme that will reduce passenger delays by thousands of hours.
“The partners will play a critical role in delivering the East Coast Digital Programme (ECDP). The first £350 million investment in the ECDP by the government is already being used to begin the introduction of real-time digital signalling on the route, and lay the foundations for wider national roll-out.
“The ECDP will be the first intercity digital railway in the UK, fitting trains with the latest in-cab signalling technology and removing the old lineside signals. It will mean that signallers will be able to talk to trains continuously rather than only at fixed points, instructing and responding in real time and reducing delays and significantly improving performance.
“Network Rail launched a procurement process to find private sector partners to help deliver the programme back in September 2018. It was an entirely new way of working, to team up with suppliers from the start to design, develop and deploy the European Train Control System (ETCS) technology. The procurement has concluded with Siemens confirmed as the programme’s train control partner (TCP) and traffic management partner (TMP), and Atkins as rail systems integration partner (RSIP)…”
An update on New York MTA’s CBTC technological progress with UWB (Ultra-Wide Band) wireless technology that offers faster and less-expensive installation of modern CBTC.