Automatic Train Operation (ATO) is nowadays a given technology for metro operation and, although it is still in its infancy when applied to main lines, it is often a chosen option for self-contained industrial or freight railways. But is ATO being used to maximum advantage and what are the criteria for optimising the benefits?
A recent seminar staged in London by the IMechE and the IRSE was aimed at creating a better understanding of ATO, not only its technology but how it needs to be integrated into other systems. Some of the output proved revealing.
The London Underground perspective
George Clark, the director of TfL Engineering and the recently elected President of the IRSE, reminded everyone that the ATO story began on LU when the newly constructed Victoria line was commissioned between 1968/71. This was ground-breaking technology in those days, especially as the system used was invented in house. Based around different frequency-coded track circuits, the system was operational for over 40 years, which is a testament to its good design and maintainability. The system was, however, just that – an ATO application that optimised the driving of trains.
Things have moved on since then and LU now uses modern CBTC (computer-based train control) systems on most of its lines which incorporate ATO functionality. Such systems demand a co-ordinated approach between signal and rolling stock engineers, with the civils fraternity needing to be heavily involved so as to get the right track layouts in support of automated operation.
Over and above the engineering, the train-service planners and operations managers have to decide what kind of train service they want – it is from their decisions that the functional and engineering specifications will be derived.
Whilst ATO is delivering enhanced capacity and safety with maximum acceleration and optimised braking rates, other factors have to be taken into account, particularly if the railway is in an outside environment. Knowing every train’s position is essential for safe ATO operation and this can be a challenge where climate conditions and vegetation affect adhesion. The need for rail lubrication on curves and altered braking rates when rails are wet or contaminated are part of a modern ATO specification.
Closer adherence to the timetable means linkage to either a traffic management system or ARS (Automatic Regulation System) to achieve a design that extends to the margins of operation. The driver (in reality, a train operator) must be assured the system is functioning correctly.
Modern ATO systems require huge quantities of data transfer as part of monitoring performance, predicting maintenance, linking to information systems and updating asset databases, so Big Data and IoT must be part of the communication package. Lastly, as LU has successfully achieved with the Victoria line upgrade (now operating at 36 trains per hour in each direction), sooner or later an ATO system will require an upgrade or replacement and a means of achieving this without service disruption should be part of the initial system design.
ATO on the main line
Introducing ATO on to a main line railway is several degrees more difficult. However, two linked successes can recently be chalked up. The biggest of these is the completed development of layering an ATO package onto an ETCS-equipped railway, the second is the deployment of such a system on the London Thameslink central core section. Press releases claim this latter as a main line first but, in truth, it is more akin to a metro operation with the same type of trains all stopping at the same stations. Nonetheless, it is a milestone and Imtithal Aziz from Network Rail described the need for an ATO package in order to achieve 24tph. It took five years of work to develop and test the system including live running on the Network Rail Hertford Loop test track at Watton-at-Stone.
The UK can be rightly proud of this March 2018 achievement that, no doubt, will be the basis for many more ATO systems in Europe and worldwide. The need to define what has to be added to ATO for it to meet the objectives emerged along the way, starting with an understanding of the Grade of Automation (GoA) levels now set down as standards. These are:
- GoA 1 – Manual Train Operation but including C-DAS (Connected Driver Advisory System);
- GoA 2 – Driver Attended ATO (as LU operates its ATO lines);
- GoA 3 – Driverless Train Operation (akin to Docklands Light Railway where a ‘Passenger Service Agent’ – sometimes called a train captain – is retained);
- GoA 4 – Unattended Train Operation (UTO) with no member of staff on board (Lausanne Metro for example).
Examples of all these exist across the world, so the technology is already established. More difficult is the public acceptance of, especially, GoA 4 and what happens if a train fails or an accident occurs. Network Rail, for the present, has decided to stop at GoA 2, since it is likely that any ATO operation only covers part of a route, with normal driving taking over for some of the journey.
To achieve an optimum journey experience with timetable adherence, optimum performance, maximum capacity and good safety will require other technical systems to be added in. Traffic Management is vital since accurate train regulation is required, particularly if an ATO railway has multiple junctions and routes. On Thameslink, where the TMS has yet to be completed, a 30-second update of timetable to each train is proving insufficient and real time updating is now regarded as necessary. This, in turn, demands a very high data exchange rate to every train, something that is beyond the capacity of the present GSM-R system. Thus, a replacement 4G or 5G radio link may take on more urgency with Wi-Fi enhancements likely to be included as well.
C-DAS is equally important for GoA 2, ideally integrating this with ETCS and ATO on to the driver’s DMI screen. Even for GoA 3 and 4, the C-DAS data may be used as the input commands for the ATO operation.
Other challenges include the definition and location of timing points, automatic or manual door opening releases, train despatch procedure, station stopping positions for different length trains, transition arrangements between ATO and non-ATO operation, interfaces to the train propulsion and braking systems, and keeping the track data base information updated on the train system.
The European Shift2Rail and TEN-T programmes are incorporating ATO requirements and the ERA (European Railway Agency, now renamed European Union Agency for Railways) will be building ATO into the ERTMS/ETCS specifications. Network Rail will be considering how ATO can be accommodated into its Hybrid ETCS Level 3 concept.
ATO as a means of predictable and efficient operation
Having a supplier’s viewpoint is always useful, because they tend to have a wider perspective as to what is going on. Thomas Godfrey from Bombardier confirmed that ATO is very mature in the metro sector, many of which run as UTO. However, most systems are proprietary, with only high-level standards giving general guidance on technical requirements. The IEEE Standard 1474.1, dating from 2004 but with several updates since then leading to CBTC operation, has enabled considerable innovation.
Heavy haul freight traffic in the industrial sector has benefitted from ATO, but extending this to mixed traffic lines is much more complex owing to the many types of services and speeds. UNISIG, the industrial consortium which was created to develop the ERTMS/ETCS technical specifications, is producing an ‘ATO over ETCS’ TSI (Technical Specification for Interoperability) and Network Rail is known to be working on national standards for ATO with the risk that deviating from any international standards will invariably increase costs.
An ATO model will consist of both trackside and train-borne elements and must include TMS for constant updating of regulation schedules, plus balise positioning, GSM-R and speed probes on axles. Speed measurement and accuracy is essential for successful ATO.
The initial Chinese experience of ATO led to a jerky ride owing to delays in system response to commands, but this has now been evened out.
Any GoA 3 or 4 application will need creep facilitation, automatic sleep and wake-up commands for effective berthing, automatic joining or splitting of trains and automatic detrainment in the event of an incident. Some of these present real challenges.
The logistics of ETCS rollout will limit ATO usage as route fitment will lag behind equipping the trains. In the UK, some routes, such as C2C, Merseyrail and parts of the Glasgow suburban network, will be easier than others. A cut down ATO application for automated door opening and assisted braking may be a more practical proposition elsewhere.
Making an ATO railway work
Even when ATO is introduced, there are many operational issues to be sorted out. Pradeep Vasudev, technical director at WSP, outlined the key interfaces that are demanded. A structured approach based on experience needs to consider that, whilst the signalling effectively drives the train, what role does the driver have, if indeed there is one? Door operation, start up commands, degraded mode operation, all need to be decided. Semi-automated disruption planning is more difficult – recovering a failed train might mean signalling a second train into the occupied section to couple up and push forward, not an easy concept for signal engineers.
Operations has effectively entered the Systems domain and must consider:
- Operational change including disruption management;
- Technical integration, including connectivity and software management;
- Programme integration including migration methodology and operator readiness.
Migration from an old ATO to a new ATO system is challenging and the experience of the Victoria line in London could become a template for others. It involved lots of sub-projects and key milestones for both track-based and train-borne equipment.
GoA 4 and the Glasgow Subway
The Glasgow underground railway, which basically has a circular route, is being modernised. This includes new trains, improved stations, depot upgrade and a commitment to GoA 4 UTO operation. Stadler is supplying the trains, working with Hitachi STS to deliver the complete system. It is a complex migration and Stefan Rosendahl explained the challenges.
The 17 new trains will be an articulated four-car design, with the end coaches having two bogies and the shorter middle cars having just one. Walk-through access and information screens are to be provided, along with help points. Maximum speed will be higher (58km/h), capacity will be 110 seats and the intended frequency will be 16tph in rush hours. The Hitachi CBTC product will form the basis of the ATO, including aerials and speed sensors. Platform screen doors (PSD) will be part of the GoA 4 package.
Migration from old to new is the biggest challenge, as the Glasgow authorities wish to maintain a service during the implementation period. This means equipping the new trains with a temporary driver’s area, so that old and new trains can run simultaneously to the existing GoA 2 operation. Once all the new trains are in service, a ‘lift up’ driving desk will replace the temporary driver’s console and UTO can be initiated.
CCTV recording of all passenger activity in every car is a requirement, as is on-demand monitoring of images at the central control room. This, in itself, requires a large bandwidth from the track-to-train communications link.
A 2’ out of 3’ redundancy design for the vital equipment is required to meet safety and reliability requirements. Ten possible operating modes are foreseen, ranging from manual driving, in both forward and reverse directions, through to UTO operation. One requirement is for a second train to rescue a failed train.
Many eyes will be watching this exciting development as implementation approaches.
ATO for HS2
When HS2 is complete, it is expected to take much of the traffic from the West Coast, Midland and East Coast main lines, in part due to the requirement to free up capacity at the London end of these lines. Hence 18 tph is required on the London to Birmingham section, which demands an ATO solution. Tight timetable compliance, with boarding and alighting aimed at two minutes and only a 30 second delay allowed on all services, makes HS2 akin to a high-speed metro – so says Ben Rule, the director of operations.
ATO will enable a high consistency of driving but will include the ability to manipulate the train service for gaps if things go wrong and allow for processes such as an emergency stop. Trains will be ‘flighted’ in three or four-minute groups and, whilst capable of 360kph, will normally not go this fast unless a catch-up situation arises.
Many trains on HS2 will also run on the conventional rail network, which is where the biggest source of delay will originate. Current long-distance services from the north of England and Scotland will divert to HS2 to free up capacity on existing routes, especially the West Coast main line. Once on HS2, the line controller effectively becomes the driver of the train, so as to ensure optimum regulation.
Carine Marin, the acting head of control and communications engineering at HS2, advised that no ATO high-speed product is yet available but anticipates that the GSM-R replacement (FRMCS) will be developed by then. This, when coupled to TMS and GPRS, will allow a suitable ATO package to be produced. A CCS laboratory will be provided to model and demonstrate the system, but a long and robust testing and commissioning plan is envisaged.
Different weather and climate conditions can play havoc to an intensely used railway, so they have to be considered. Phil Dubery, from CPC Systems, informed that positional uncertainty can lead to a see-saw action of propulsion and braking in an ATO railway if conditions are bad. Dropping the brake rate from 0.7 to 0.4 will lead to a 12-second delay for every station stop.
Having a management plan to deal with this is important. The production of a real-time adhesion map is one solution, so information relating to locations where trains will struggle and the associated data on environment, gradient and weather can then link into TMS and ATO intelligence. Provision and observance of temporary speed restrictions could become part of this map.
Evolution of ATO in Hong Kong and Singapore
TC Chew, now with Arup’s global rail business but well-known for his work to develop metro systems in the Far East, suggested some fundamentals in the development of an ATO-operated railway. Hong Kong, with its 11 lines, 223km of route (including the now-integrated KCR railway) and five million passengers per day, modelled its first ATO application on London’s Victoria line. The replacement with more modern systems is well under way, with some lines now operating a GoA 4 system.
Singapore Metro has six lines, building to 330km of route by 2030 and three million passengers per day. It also is updating its ATO systems, but originally opted to retain a ‘front’ driver. This decision has been reversed and four lines now operate to GoA 4.
It is important to understand that customers don’t care about the system, they just need to get from A to B reliably and safely. Both HK and Singapore aim for 99.9 per cent running on time. Funding is crucial and must embrace staff, equipment, failures, emergencies and all associated subsystems.
Both networks have recently experienced embarrassing accidents that caused collisions. Some operators demand that a backup system must be in place in case of ATO failure. In HK, a backup to the backup was also specified, and that caused data irregularities at the interfaces, resulting in the collision. The need for a backup is a contentious issue, but opinion is moving towards investing in reliability for the prime system.
Some misconceptions of a GoA 4 railway need to be exposed. Loss of jobs need not happen, as staff should be redeployed to passenger-assistance tasks. Fear amongst passengers is a myth, as they are not left to their own devices if an incident occurs. The cost of a fully automatic system should not be significantly higher than one where staff are retained.
The two big challenges for GoA 4 are how to bring a failed train to the next station and the need for batteries for emergency propulsion if power is lost. Solve these two and there is no stopping a fully automatic technology.
It is clear that there is much more to successful automatic operation than just buying an ATO package. A final discussion raised some further thoughts that will need consideration over time: industrial relations, artificial intelligence, moving block, timetable initiation, driver training, managing the infrastructure and rolling stock divide. Maybe another seminar in two or so years’ time will have answers to these.