National operator DSB is working on a pioneering programme to equip København’s S-bane for fully automated operation without using platform edge barriers. Contracts are due to be let later this year for an obstacle-detection system that could pave the way for cost-effective GoA4 applications worldwide, as Executive Vice President for Strategy & Rolling Stock Jürgen Müller explains to Benjámin Zelki.
‘We are operating a very stable, very robust, but brownfield system; the 170 km, seven-line S-bane network serves 350 000 passengers a day. It would just be very costly and too disruptive to retrofit our legacy stations with platform screen doors, some are 90 years old’, says DSB’s Jürgen Müller as he explains why the operator decided to rule out the use of platform screen doors at the very start of its S-bane automation project.
He explains that the operator looked at international examples of how to protect automated railways and metro lines, and ‘we quickly realised that platform screen doors are the default option to make stations safe. So we started a careful investigation, after which we decided to use an obstacle detection system instead, to ensure safety in an open environment.’DSB describes the obstacle-detection system it is working on as ‘one of the most pioneering elements’ in the automation programme. The ODS-P technology is not readily available on the market, and DSB expects that the planned development could pave the way for cost-effective GoA4 applications on other networks worldwide.
Deer and newspapers
‘The scale of the København automation project makes it globally significant including the network length, the amount of trains involved’, Müller adds. ‘A major challenge is that our network is open-air, crossing urban areas, woods and nature reserves, with occasional deer crossing. It is also prone to the weather, sometimes heavy snow. Therefore, we are undertaking several shadow operating phases before we start revenue services’, Müller says.
‘It was very important to have the Danish Civil Aviation and Railway Authority onboard from the beginning. They have been supporting us throughout, and we agreed that the safety level of the new system had to be at least as good as the legacy train control equipment. And we’re now very confident that the system is very safe.’
Separating false alarms
Müller explains that today, DSB believes the risks of false alerts leading to service disruptions is a much higher risk than bone fide safety incidents. ‘We looked at statistics for Lyon’s automated metro Line D, where a similar obstacle detection tool has been deployed, and we found that more than 90% of the stops were false alarms.’
‘Can we get the tool to identify if it is a bird flying through the track area, a newspaper falling from the station, or a person in the tracks, to avoid stopping the trains unnecessarily?’ Müller asks. ‘And if the stop happens, we need to get a person to identify the reason quickly, and to restart the system in the case of a false alarm. These are the issues we are working on now.’
There will be a contractual target for malfunctions in the planned contract package for the obstacle detection system, along with targets for an as yet unknown number of true and false alarms.
DSB anticipates that not every alarm will impact the service, as the operators in the traffic control centre will be able to check an an alarm via the surveillance system and cancel it before a train reaches the affected area. The operator explains that a long development phase will see shadow operation at selected stations, including proof-of-concept installations, adding that this will provide a clearer picture of the number of true and false alarms that could be expected.
Three large contracts
The København automation programme for GoA4 unattended operation is divided into three large contract packages.
Siemens Mobility is equipping the S-bane network for GoA4 operation under €270m contract awarded in April 2024. In January this year, DSB selected a consortium of Siemens Mobility and Stadler to supply and maintain a fleet of at least 226 electric multiple-units suitable for driverless operation.
The third package, covering the platform-based obstacle detection system branded as ODS-P, and including the provision of CCTV, passenger information systems and emergency stop buttons, is expected to be awarded later in 2026. Siemens Mobility and Hitachi Rail (previously Thales GTS) have been prequalified.
Müller explains that the tender package for ODS-P includes functional requirements based on dialogues with other operators using various versions of live obstacle-detection systems worldwide, as well as on market dialogues with suppliers. However, the winning bidder will have to design the tool and have a choice on the technology. During the tender process, a competitive dialogue has been undertaken by DSB, with a focus on safety targets.
‘When the Supporting Systems contract is awarded, we will have all the puzzle pieces in our hands’, Müller says. The entire programme is valued at DKr30bn, and is funded by DSB through ticket sales and the service level contract with the government.
From GoA2 to GoA4
Under an August 2011 contract, Siemens equipped the S-bane network with CBTC as a replacement for the legacy HKT signalling system. CBTC went live across the entire network in September 2022, and operation at Grade of Automation 2 was achieved in May 2023.
DSB reports that the introduction of CBTC has significantly improved punctuality, which reached what the operator says is a record high of 95·6%. This was measured to within 3 min of time at each station, as an average over the course of 2025.
CBTC has also slightly reduced travel times by optimising braking curves, saving a few seconds at each stop. DSB expects the transition to GoA4 to further enhance efficiency and reliability by eliminating human variability, allowing for more precise and consistent operations.
Asked about the cost implications of upgrading the CBTC for the enhanced functionality rather than replacing it outright, DSB told Railway Gazette International that ‘to enable GoA4 unattended operations, the existing CBTC system will be upgraded to include new functionalities, including software updates for existing onboard units to ensure compatibility with the new GoA4 trains and the upgraded trackside system. Regardless of whether the new fleet is GoA2 or GoA4, DSB would need to procure new onboard units for those trains and upgrade the trackside CBTC.’
Managing the interfaces
Working with three contract packages and co-ordinating multiple suppliers is another major challenge for DSB, especially when working on a historic network.
‘It was clear for us after the market consultation that no supplier would be willing to undertake such a large turnkey contract, so we split the project into three large packages’, Müller explains.
To manage the complexities, DSB chose to keep system integration in-house. This approach allowed for early and active involvement with suppliers, as well as the establishment of tools like an integrated test lab to facilitate successful delivery. These methodologies could serve as a model for other large-scale automation projects, the operator believes.
Platforms not rebuilt
DSB opted against rebuilding platforms across the entire network to avoid years of disruption, but it is looking to implement other measures to improve accessibility. These include measures to bridge gaps and height differences between the trains and the platforms.
The new S-tog fleet will retain the wide-body design of the current rolling stock to help minimise the platform gaps. However, the new trains will be longer and slightly narrower — 3 550 mm instead of 3 600 mm. Most doors will have fixed steps to minimise the stepping distances. The doors serving those vehicles with an accessibility zone will be equipped with automated sliding ramps and lower floors to manage the height differences and ensure unassisted accessible boarding.
Platform gaps vary between stations and are bigger on curved platforms. DSB is therefore looking at undertaking a small package of infrastructure enhancements, including the deployment of rubber gap fillers where necessary — these components have been used in several other metros and suburban operations.
Among the modifications being evaluated is the addition of small humps or ramps on some platforms to give an extra 20 mm of height. This would enable passengers with prams, luggage or limited mobility to board and alight more easily.
Line F as a pilot
Current plans envisage that the orbital Line F will be the first route to be fully automated. This north-south semi-circular line is the only one that does not pass through the central section including København H main station, making it ideal for testing unattended train operation without risking disruption to the entire network. During this phase, DSB expects to gather operational data and refine processes in a less complex environment, which will allow it improve efficiency and reliability when rolling out automation on the other lines which share tracks through the cross-city core.
Deployment on Line F is expected to begin in 2031 and last a year, with automated services starting by the end of 2031. After that, the line-by-line rollout is expected to take eight years, meaning the automation programme will likely last through most of the 2030s. DSB expects some service interruptions, but Müller confirms it is working to minimise the impact.
Retaining staff
DSB expects its frontline staff to be retained across the business even as driverless operation comes in. Existing drivers will be offered opportunities to transition to roles on regional and long-distance trains, or to be redeployed as onboard supervisory and customer service staff on the automated S-bane.
Roving onboard agents will be allocated to defined zones across the network, assisting passengers and handling issues such as stranded trains or alarms from the obstacle detection system. This will ensure a strong focus on passenger service and operational safety in the automated network, the company says.
International partnerships
‘We have been looking at other cities and have been in contact with colleagues at other operators around the world where similar automated lines operate with obstacle detection systems’, Müller says. These include the SkyTrain automated light metro in Vancouver, the Kelana Jaya monorail in Kuala Lumpur, Nürnberg U-Bahn, Lyon metro Line D and sections of the Barcelona metro. ‘We are also in contact with other relevant operations, including DB for the S-Bahn networks in München, Hamburg, Stuttgart, as well as Paris’s RATP.’
He emphasises the importance of academic research to underpin the programme. ‘We work with universities which specialise in automation, including Technical University of Denmark and TU Dresden. We are trying to get knowledge to avoid mistakes and go into difficult areas with open eyes.’