It is a salutary thought that the European Train Control System (ETCS) was first introduced into the UK in 2010, now 16 years ago. This was on the Cambrian line from Shrewsbury to Aberystwyth and Pwllheli, which acted as a proving ground.
This was to herald the start of a rolling programme where the technology would be adopted for the majority of main lines by the end of the 2030s. Sadly the rate of roll out has fallen far short of expectations and the only other routes now equipped are the central core of Thameslink that has the important add on of Automatic Train Operation (ATO), the GWR route into Heathrow Airport which is now extending back into Paddington, and the recently introduced Northern City Line (NCL) from Finsbury Park to Moorgate in London.
So why the slow progress, and has it been a blessing in disguise?
Over time, the technology has developed with associated upgrades in software releases but the difficulties of fitting a complete route with a mixed fleet of trains has proved to be something of a challenge. The creation of an industry-wide partnership has been necessary to move the problem forward. Two of the lines equipped so far have a single type of rolling stock that makes the logistics much simpler. The ultimate aim of dispensing with lineside signals has been achieved on the Cambrian which was easy as, being previously equipped with Radio Electronic Token Block (RETB), it did not have signals. The NCL meanwhile is a self-contained line where traditional signals were removed in May 2025. At the two other locations, the ETCS exists as an overlay so that unfitted trains can operate.
The first big test for ‘signals away’ will be the East Coast Main Line (ECML) where the East Coast Digital Programme (ECDP) has been actively planned for some years. The initial contract which included the NCL was signed in Q1 2022. Implementation of the first main line section from Welwyn Garden City to Hitchin is almost ready for cutover but must be rigorously assured for both safety and operational reliability before this can happen. This section was chosen because it has the biggest capture of drivers and fleets from different train companies. The main supplier of the infrastructure equipment is Chippenham-based Siemens Mobility, a company that has a long pedigree in the supply of signalling systems.
The scale of the task
To explain what is involved, Siemens invited selected members of the railway technical press to witness what is required and the progress being made. As well as the ground equipment, the section of route requires a number of train types to be fitted; these include classes 800 (the LNER, Hull Trains, and Lumo fleets), 700 (Thameslink), 717 (inner suburban that already operate over the NCL), 387 (outer suburban), 180 (Grand Central), plus some freight locomotives and on track plant and track measurement trains.
Testing all of this on the real railway would be very disruptive, so as much testing as possible is carried out off site. A vital element of this is the building of a digital twin that replicates in the factory what is being installed on the ground. A System Integration Laboratory has been created which is a sizeable room containing all the elements of an ETCS-controlled railway. These include:
- A replica of the signaller’s work stations that exist in York ROC that control the Welwyn Hitchin section.
- The Trackguard Westlock interlocking for that area.
- The interface between the interlocking and the work station (Controlguard Westcad).
- The Radio Block Centre (RBC).
- A driver’s ETCS on board control console including a video screen simulating the real railway from Hatfield to Arlesey.
- A Siemens train mobile kit.
- A GSM-R cell site covering the lab, connected by a leased comms link back to the Network Rail GSM-R test network in the Midlands.
As such, the link between the radio network and the interlocking is via the air gap.
To ensure interoperability and in the interest of mutual co-operation, other suppliers’ train borne equipment has been loaned to Siemens so that these can be tested to the ground equipment via the radio link. The precedent for multi-supplier testing took place at Hitchin in recent years where different suppliers installed ETCS infrastructure that connected to a test train on the Hertford loop.
Results so far
Creating the digital twin is proving its worth. Having the capability to not only test the technology but also how the system will be used operationally is of immense value. Being given the chance to ‘drive’ a train through the section proved popular with those present. The Movement Authority (MA) together with the permitted speed is shown by a grey band around the outside of the speedometer. The lineside signals will remain for a period after the cutover and are aligned with the MA. It will not be possible for an MA to go beyond a red signal as drivers in the real world would find passing a red signal very unnerving.
Keeping the train speed within the limit could be challenging, not only when slowing down as needed but to maintain the speed necessary to meet the timetable. The grey band changes to yellow if the allowed speed is exceeded firstly as a warning to slow down and then to red with the brakes being automatically applied if the required action is not taken. The console also includes details of the line characteristics such as gradients and speed restrictions.
The changeover from conventional signalling to ETCS once operational will be automatic, providing the driver is trained and certified for ETCS operation. This will require the driver’s capability to be part of the sign on process for the journey, another thing that the twin can test.
The simulation of the York work station shows the traditional white lights for route availability, red lights for section occupied, but also green lights showing the extent of the MA given. The twin has also shown up some technical deficiencies needing attention, some associated with interpretation of interoperability between the different suppliers. Another challenge is backward compatibility as it is certain that the software level currently standardised at an international level will be updated over the life of the system. The current thinking is that train equipment with an earlier version of software will work to infrastructure with the latest software. However, earlier software infrastructure will not support later versions of mobile software, which could enforce updates into the future.
While testing on the twin is invaluable, it is not sufficient to implement the system without additional onsite testing taking place. This has proved very disruptive with multiple closures of the ECML between Peterborough and Kings Cross at weekends being deemed necessary and consequential loss of revenue and passenger dissatisfaction. The disruption is acknowledged but work is being done to keep access to a minimum. This already includes action to deploy exclusion zones where quadruple lines exist such that two tracks are taken out for testing under a possession leaving the other two lines available for normal traffic.
The future
The plan is to commission the Welwyn Hitchin section in July. Not all rolling stock will be fitted by then and thus lineside signals will need to be retained for the immediate future, which will also facilitate conversion training for drivers in ETCS operation. No date is yet set for when signals can be removed.
After this, the roll out on the ECML will continue to eventually cover the whole section from Kings Cross to Stoke tunnel just short of Grantham and is expected to be completed by 2029, still three years away.
The bigger question is which lines will be converted to ETCS after that? Active planning is happening for further elements of the ECML, the northern section of the West Coast Main Line, the Midland Main Line, the Trans Pennine route and even the Brighton main line. It is likely to be renewal driven as replacing signalling with more ‘lights on sticks’ will fail to deliver the advantages and cost reduction that ETCS can yield.
None of these are certain and it will be up to GBR to make the decisions. Funding will be crucial but equally important will be keeping the teams with the vital expertise in place. Future projects are unlikely to need the degree of assurance testing once the principles for the ECML are established. This essentially means a rolling programme that governments of whatever party can sign up to. A little like electrification where the start/stop funding situation has caused so much criticism.
Thanks are expressed to Siemens for allowing Rail Media an insight into the factors associated with ETCS roll out.
Network Rail Pragmatism
Coinciding with the article on the Siemens ETCS Integration Lab, a talk on ETCS progress in the UK was given recently to the IRSE London & SE section by James Dzimba, chief control & communication engineer at Network Rail. This was a brave and honest assessment but left a lot of questions to be answered.
Signalling assets are ageing faster than the renewals that are taking place. The current actual spend in CP7 is less than £600 million per year, increased to £660 million per year when level crossing renewals and minor works are included. The expected spend was predicted to be £830 million per year, so a significant shortfall is occurring.
Where does ETCS fit in this? The use of ETCS is driven by deliverables, not the technology but concerns exist as to why it is costing so much. In Europe, the success of ETCS is driven by country mandate and a system authority which helps to create a stable ongoing work plan.
There is also the question of strategic safety versus functional safety, namely fundamental versus site specific. Over-specified ETCS undermines the strategic safety and functional over scope and so drives unaffordability since the accreditation inflates the cost.
Cost must be driven out of ETCS if it is to be the preferred system for ongoing renewals. Despite interoperability requirements, there is risk in always using the same supplier. There is no common way of delivering ETCS. Having it as an overlay system will always be expensive and not beneficial to the business case. Lessons have been learned from the application of ETCS to the Cambrian route, the Thameslink central core, and the Northern City line from Finsbury Park to Moorgate but all of these have only one type of rolling stock in day-to-day operation. Equipping a mixed traffic railway as in the ECML makes it much more difficult.
While it might be perceived that ETCS is highly desirable, it is not inevitable. Alternatives exist and others are possible:
- Modular and low-cost traditional signalling.
- Development of a low cost ETCS alternative – Regional / Lite / Sniffer.
- Only go for targeted Automatic Train Operation (ATO), not blanket deployment.
- Deliver only the core functionality, not enhancements.
ETCS pundits claim the system will be safer but are we not safe enough? From an engineering perspective, there must be a degree of realism prevailing. A one-size-fits-all situation is not deliverable or sustainable. ETCS should be regarded as an evolution, not a step change. There is a big risk of going backwards.
CP7 funding constraints are evident and unlikely to improve. While further routes have been identified, such schemes are being deferred but any extended migration erodes the business case. CP8 funding must be made to work, and for this a central system authority, a mandated fleet fitment that fits the timescale, and long term stable funding, are all essential.
James complimented the creation of a system integration lab by Siemens as very helpful to the engineering development and associated testing. He acknowledged the damage being done to rail passenger confidence by having to close the ECML for several consecutive weekends from Peterborough to London for testing purposes. New methods of carrying out testing are being developed.
This talk was remarkable in that it brought out many of the concerns voiced by professional railway engineers and operators over several years. One big question that remained unanswered is that there seems to be no ‘directing mind’ for the future of ETCS. That is a question that GBR might have to address very soon after its creation.
