Every two years, the signalling industry showcases its achievements, innovations and challenges under the umbrella of the Institution of Railway Signal Engineers (IRSE) ASPECT conference, which this year was held in October at the Technical University in Delft, Netherlands.
Founded in 1842, the university has a Department of Transport Operations and Management, where Prof Dr Rob Goverde heads up railway operations and engineering. Resilience engineering and safety management can also be studied – very pertinent with the increase in passenger demand across the world’s railways.
George Clark, the IRSE president, emphasised that engineering, operations, cyber security and skill sets are all part of obtaining resilience. The recent UK power outages, which lasted only 15 minutes, caused train service disruption for several hours, an example of resilience not being effective.
Resilience in many forms
Wim Coenraad, a past president now working for Movares (a Dutch rail consultancy), spoke of the need for ‘business continuity’ as a means of keeping the railway running when large scale disruptions occur.
A power failure in Switzerland, during which back-up supplies and a diesel generator did not function as planned, resulted in delay to 1,500 trains and 200,000 passengers, with compensation of three million swiss francs (£2.3 million) being paid out. It can be an expensive business when things go wrong.
Learning lessons is all important, so companies should try and stabilise the situation, aim for a skeleton service, load share if possible, not overwhelm staff with information and focus on the essentials.
Managers need to try and analyse the risk of unimaginable consequences by understanding the operational, information and signalling technologies. They should try to put these together into a disaster recovery plan based on security and contingency planning, creating a scenario for the predictable that includes a crisis and emergency centre and having staff trained with empowerment to act should be part of the safety culture.
On the same theme, Andrew Love from SNC-Lavalin talked on ‘The approach to planning for things that might not happen’. An example would be a national emergency such as a pandemic flu outbreak where more than 1.13 per cent of the population is likely to die.
Staff absence in any such emergency would be a major problem, so having alternative suppliers on tap for catering and cleaning would be an asset, as would staff being equipped to work from home. Companies must be aware of the power of social media for the communication of information and test the resilience of their systems if possible.
Disney Schembri from Siemens spoke of climate change and the growing risk of extreme wet weather and flooding, of extreme heat and potential rail buckling, of the associated crowd build-up at stations and the general impact on people. With half of all rail travel being commuting to work or education, there is predictable frustration if trains are cancelled or late.
Intelligent monitoring systems need to include drones, cameras and soil sensors, while plans for entry and egress at stations, particularly at peak times, has to take advantage of personalised smart phone technology. Much of this has commercial implications which may be more difficult to solve than the technical issues.
Filling the skills gap is another resilience factor, with 30,000 apprentices required in the UK by 2022. Such a programme will need funding of £206 million per year plus an element of government support, a significant sum for the industry to find.
Resilience in design
Building ever-larger control centres (ROCs – Rail Operating Centres – in UK speak) is seen as designing-in resilience, according to Victor Abbott from Jacobs Australia. The ability to work in Normal Mode, Degraded Mode and Emergency Mode is necessary to take into account technical, customer, operator and external factors.
A ROC should be able to respond to all situations through a developed framework using a hierarchy of control. These can be different depending on the type of railway – a metro needs a local centre, an intercity railway – something for the entire route, and smaller countries can have just one centre for its entire network. Whichever solution is selected, the controls should be capable of managing track, rolling stock, signalling, telecoms, power, buildings and external factors. Having different control interests in different buildings creates a silo effect and does not stack up, as data and visual integration is essential.
Alexandra McGrath from VicTrack Australia spoke of experiences in Melbourne, where the 1970s city loop now cannot cope with passenger numbers, resulting in bottlenecks and the signalling struggling to perform. A major control centre failure in 2017 resulted in a backlog of trains and took several hours to recover since the crowd build-up prevented staff from getting to the emergency control centre. Lessons learned from this include a twice-daily information update, much improved liaison with other forms of transport, the removal of level crossings and a detailed exercise to compare the conventional and resilience approach to engineering elements.
Keith Upton of Atkins questioned the value of the Network Rail GRIP process (Governance of Railway Investment Projects) as being too project-management biased. The eight stages are output definition, feasibility, option selection, single option development, detailed design, construction test and commissioning, scheme handover, project close.
He suggested that many of the early stages should be a tick box exercise, but was concerned that the present culture leads to prolonged dissent resulting in delays, overruns, budget overspend, contract disputes and cancellations.
While he accepted that capturing requirements is important, he noted that stakeholders are not always obvious. An assurance process made up of ‘technical stage-gates’ would be more applicable and would explain to everyone what is actually involved without it becoming too bureaucratic. Recent project “fiascos” could have been averted.
Resilience in engineering
Can the IoT (Internet of Things) assist the resilience of older technology, questioned Bob Janssen from Siemens Netherland?
A solid-state interlocking imports and extracts lots of data to and from the outside world, and this could be used to monitor performance or potential problems. Taking a shunt signal as an example, measuring voltage and current is easy but how about supplier, wiring, identification, point status and protection?
Such enhancements would allow the data to be enriched beyond the interlocking, enabling a minute-by-minute state of the railway to be exported to a dedicated server, accessible to approved third parties. With clever algorithms, a particular failure could be calculated for the impact on the whole rail route.
Providing resilience whilst upgrading a railway is a challenge. Ian Jones from Siemens UK quoted the London Underground’s Victoria line, the Thameslink transition to ETCS and ATO, the S-Bane refurbishment in Copenhagen and the Riyadh metro. All of these projects introduced complex software that failed to perform as expected.
A soft migration strategy requires regular testing with the ability to go back before finally adopting. Participation by all levels of management helps to give confidence to both in-house engineers and the suppliers. In Riyadh, a dedicated, circular test track facilitated testing of all operational aspects, including environmental controls.
Railways worldwide are seeking to increase capacity without having to build expensive infrastructure. The greatest challenge is with metro and suburban lines but additional capacity is also needed elsewhere.
Aaron Sawyer from SNC-Lavalin described the situation at London Heathrow Airport terminal 5. This has a rail link to connect the main terminal building with its satellites at T5A, B and C. Built as a two-track transit system, the initial operation used two trains, one on each track, shuttling up and down.
However, this proved insufficient to handle the number of passengers boarding or alighting from ever-larger aircraft. A four-train operation was therefore devised, with up and down lines and crossovers at each end. It sounds simple, but could be fraught with problems if things go wrong, and how to test it?
The solution was an N-gauge scale model using COTS (commercial off-the-shelf) components, including a Raspberry Pi processor, all connected on an Ethernet backbone. Whilst clearly not failsafe, the operators acted out real situations in an off-site location. It has worked well and built confidence in controlling the full-sized system.
The upgrade of London’s Victoria line using a fixed block ‘distance to go’ radio system to achieve 36 tph (trains per hour) has been a remarkable achievement, but staging the introduction to build up to that level and not trying to achieve it in one go was a sensible precaution.
On the Sydney suburban network in Australia, a 375 million annual ridership is continuing to grow. Increasing capacity includes using ETCS Level 2 with axle counters and a traffic management system, which means a completely new way of working for 4,000 of the network’s 10,000 employees.
To plan for this, studies were made of similar situations around the world. An integrated and collaborative approach between engineers, operators and suppliers was thought to be essential, with configuration of standard products ensuring there would be no bespoke supplier lock-in.
The need to achieve early wins for the customers was important, but getting into the details too early could have led to an unmanageable situation, so services were introduced incrementally on a progressive deployment basis. Maybe that could be something for Crossrail to consider?
Japan has one of the world’s most reliable railways, but it, too, is struggling with capacity challenges. Yoichi Sugiyama from the Railway Technical Research Institute explained that, in the logical sequence of timetable » operations control » train control » signalling systems, managing capacity relies on knowing exactly the time and location of every train (not just at stations) and leads to a speed-control system that matches the operational control. This allows routing and revised timetabling to be calculated constantly, so as to optimise headway.
Every signalling conference must have an update on ERTMS/ETCS. At ASPECT, much was made of the ‘Hybrid Level 3’ concept, whereby existing track circuits or axle counters are retained for trains equipped only with Level 2 equipment. However, full Level 3 trains – those whose train integrity can be guaranteed – would take advantage of Level 3 operation, including moving block and closer movement authorities.
Freight trains, which on occasions can become uncoupled, must be proved complete and must operate to Level 2 rules, with more limited movement authorities.
Many railways are considering this solution, including the Netherlands and the UK, but as yet no such system is in operation. Delft University has made some calculations based on a particular route.
Increasing capacity by just increasing the number of trains under the existing legacy signalling system would give 104 per cent utilisation, meaning the timetable could not operate. Switching to Level 2 ETCS would give 90 per cent utilisation, while Hybrid level 3 would bring it down to 84.7 per cent, allowing an extra two trains to run.
The often-mentioned problem with freight trains is proving their integrity. With continuously braked freight trains, it might be thought that this should not be a problem, as a split train would result in an emergency brake application in both parts of the train. However, if the locomotive compressor was able to overcome the resultant brake pipe leakage, the front part of the train would be able to travel some distance from the separated rear part of the train. Although unlikely, such eventualities have to be considered.
One might ask why signal engineers should be responsible for proving train integrity, when fundamentally it is a rolling stock problem, but anyone who can invent a train proving system for freight trains that is cheap, safe and practical would become an entrepreneur of note.
The progression to ERTMS seems inevitable, but it will take many years to get nationwide implementation in the larger countries. Karel van Gils, innovations director at ProRail (the Dutch equivalent to Network Rail), says it will not be achieved until 2050, assuming the necessary funding continues – €2.5 billion was allocated this summer. It will need a change on how the railway is managed, with systems needing to be procured on a partnership basis.
Full ETCS Level 3 may not yield the benefits that are predicted according to Maryam Akbari, an MSc student with Mott MacDonald. Eight challenges were listed:
- operation of level crossings, stations and different types of trains;
- train integrity, as previously mentioned;
- optimising the migration path from legacy to Level 2 to Level 3;
- recovery from degraded mode working and the different ways of determining train position;
- executing maintenance arrangements – possessions, work zones and hand-backs with assurance of clear tracks;
- level crossing protection if communication is lost;
- life cycle costs and managing train borne kit and software updates;
- deployment of ETCS in busy station areas where communication paths may be limited.
All of these are real constraints, plus they don’t answer the question of what to do about GSM-R. This is a problem that is perpetually overlooked by signal engineers and operators, but it is a major challenge and will involve significant expense. The future radio options must be considered right now.
Level crossing design and operation remains a controversial subject. Maarten Bartholomeus from ProRail believes that ETCS can improve crossing performance simply because, as the position and speed of trains will be better known, which will improve calculation times for when barriers have to close. Enhanced pedestrian information could also be added, by providing a display screen showing an illuminated approaching train image with a yellow light indicating a risk to cross.
When things go wrong, failures can close the barriers for long periods and cause significant disruption. With ETCS, it should be possible to keep the barriers up but restrict any train movement authority to that point and show a ‘Not Protected Crossing’ indication on the driver’s display. Once at the crossing, extending the movement authority under a caution approach would be possible.
Most major cities have eliminated level crossings over time. However, while London still has 13 and Berlin 46, 620 still remain in Tokyo’s inner city, causing both road and rail challenges.
Measures to decrease accidents are succeeding through improved crossing protection and activation systems. Conventional obstacle detection using laser and LiDAR is expensive (€120,000 per crossing) and LiDAR cannot always detect low level objects.
As an alternative, high level infra-red cameras with a dual vision processing unit are giving encouraging results, according to Ryuta Nakasone from Japan Railways. The system retains an image of the crossing in ambient conditions that is compared with the real time image when a train is due. The technology is proven to work in rain, snow and in the hours of darkness without artificial lighting.
The proliferation of level crossings on Japan’s non-Shinkansen lines has led to modularising the analysis, design and operation. Known as the Functional Resonance Analysis Method (FRAM), it involves devising software logic for different crossing scenarios, all aimed at improving workability and maintenance.
Three basic elements exist – train detection layer, train tracking layer, warning layer – all of which are adapted for individual crossings.
Human factors were a major influence on the introduction of ERTMS on the S Bane network in Copenhagen, according to Amanda Elliott from Innovace Design in the UK. The project was complex and needed new operational rules, a changed safety management system and proof of confidence to deliver operational performance.
Human factors testing was necessary to check whether staff could handle both normal and unusual operating conditions. 240 test cases were devised, based on real life scenarios in class room conditions. Observers watched how staff reacted, with results marked as P = Pass, D = Difficult, F = Fail (also TF = technical fail, if the equipment did not perform as expected).
The tests were followed by an improvement cycle with debrief and feedback, the collation of findings, improvement actions, re-testing and closure. While initially suspicious, staff came to appreciate the usefulness of the exercise, learning that, in any degraded mode operation, they should not focus too much on ‘the rules’ for a pragmatic solution.
Undertaking even simple projects where all parties need to reach agreement can be fraught. The provision of a multi-duct cable route in Melbourne was an example described by Alexandra McGrath. Establishing an organisational structure is vital to identify conflicting interests and who specifies what. Many contradictions appear, as actions that are forbidden by one person are required by another.
Empowering a key person from each stakeholder is a key requirement. The management team needs to map the problem, ask the right questions, listen and watch reactions, don’t contradict or override and be aware of national and workplace regulations appertaining to safety and legal issues.
Is innovation the key to unlocking performance and capacity challenges?
Distinguishing between ideas that have a realistic outcome and those that are never likely to see the light of day is important. ASPECT yielded some novel thinking, but readers must decide for themselves whether these ideas and concepts are worth pursuing.
Jeong-Ki Hong from Rail Research Korea described the KRLYNX concept, which links interlockings with outstation subsystems entirely by IP communications using a closed network and an internal ring around the interlocking.
An IP control unit, that replaces the conventional relay interfaces, would have the capability of communicating to points and signals up to 40km away while the system has to be capable of verifying safety and reliability requirements.
The IP interfaces would use a maximum telegram length of 1023 bytes, including the header and payload, and an IP address for each device would need to be allocated.
Interoperability testing with three manufacturers’ products on a dedicated test track has been undertaken in 2019. Standardisation, improved life cycle cost and protection against a supplier going bankrupt are the benefits claimed.
Joāo Martins from EFACEC of Portugal saw IP and the maintenance of safety standards as a challenge where formal verification of SIL 3/4 software systems to EN50128 has to be maintained. Initial testing, using standard verification techniques on a Frauscher axle counter within a high-level signalling system, looked promising.
A controversial concept by Matthew Slade from CPC Systems is to create virtual control centres using cloud computing. The control centre hardware would be provided by a third party (Amazon?) to which applications such as TMS, SCADA and signalling controls would be connected, either by a dedicated fibre, if sufficiently close, or by the internet. The railway would benefit by having less hardware to maintain.
However, such a significant cultural shift would mean thinking through lots of issues. Would it be more reliable, would the comms diversity be robust enough, could the cost of maintaining the interfaces outweigh the cost benefit, cyber security and service level agreements could be troublesome? Above all, would the business case stack up, even if renewing out-of-date control centre hardware?
Virtual Coupling of trains keeps cropping up as the ultimate in track utilisation, whereby movement authorities to one train are sent to subsequent trains at the same time, thus all trains move simultaneously.
Egidio Quaglietta from Delft Technical University has researched this and recognises there are significant factors involved. Different train characteristics, diverging junctions, no operational principles, proving the technology and safety would all be major challenges. With ETCS Level 3 and moving block, the separation of trains on a high-speed line would be about four kilometres, and virtual coupling would improve on this. However, for lower speed lines, the advantage of coupling compared to Level 3 becomes minimal, so perhaps an idea best forgotten!
More hopeful is the idea of Vehicle-Based Train Control, whereby train movements are initiated and controlled by the trains rather than a manned control centre, so says Ying Lin from HollySys in China.
Aimed primarily at the metro market, each train would have provision for Automatic Train Supervision (ATS), Objective Controller Server (OCS) and a VOBC (Vehicle On-Board Controller) comprising Automatic Train Protection (ATP), Automatic Train Operation (ATO) and its own communication management.
The system relies on a digital map, showing where every train is at any point in time, enabling communication with every other VOBC every 250msec. Trains generate their own movement authority for the intended route and direction.
Other innovation papers included:
- Valise – a virtual balise system based on comparing real forward-facing images with retained digital images of the track ahead – a separate article in Rail Engineer on this concept will be written during 2020;
- Electronic Track Relay – being designed to improve track circuit performance, where leaves, bad insulated joints and limited rail running area can cause failures – Harm van Dijk from Movares in Holland believes that an electronic relay can improve monitoring and diagnostics and thus make failures less disruptive;
- EULNYX – this project has existed for some time, aimed at standardising the interfaces in signalling architecture to allow subsystems to be ‘uncoupled’, in other words to prevent ‘lock in’ to any particular supplier. The vision, according to Bob Janssen from EULYNX Netherlands, is a standard set of data preparations to achieve automation of tedious design tasks, easier simulation of a signalling system including ATO and formalised testing.
In any review of a three-day conference, it is impossible to report on every paper and discussion. In addition to the topics reported, many improvements to how signalling projects are designed and implemented emerged, including automated verification of signalling data (Systra UK), point condition monitoring (Balfour Beatty), how to manage transition staging during a project implementation (Shard Group Australia) and off site testing when disparate systems are employed (Siemens UK in connection with Crossrail).
Cyber security, and the factors that can influence it, featured in a number of sessions, but this subject deserves a separate conference.
The one disappointment was the absence of any telecom or radio presentations especially when the T in ASPECT refers to telecommunications (ASPECT = Automation, Signalling, Performance, Equipment, Control and Telecommunications).
Hopefully, this review will provide a flavour of the topics while sufficient descriptions are given to enable a drill down into the organisations and suppliers mentioned if further information is needed.
Well done the IRSE for staging this truly international event!