The WLE train ran between Lippstadt and Warstein, in northwestern Germany. Along the line, roughly 30 kilometres long, WLE usually moves construction material as well as beer and also provides single wagonload services. The company has been at the forefront of DAC testing, as it carried out the first commercial tests for traditional DAC last summer.

DAC tests recap

Since the beginning of 2026, the amount of tests announced and launched in the context of DAC escalated. Other than the WLE tests, in cooperation with coupler manufacturer Voith, 2025 also saw DAC tests in Sweden to assess the coupler performance under extreme weather conditions.

This year is the turn of Austria Rail Cargo Group, which is running similar tests to the ones done in Sweden. Large-scale commercial tests are scheduled for 2027, where seven trains will be equipped with the coupler and will run in nine different countries with raw materials or carrying out intermodal services.

AI-supported forecasting introduces the concept of targeted availability, with easily available parts “planned more leanly” and parts more expensive and difficult to get are “reliably secured”. Moreover, the model provides information on mileage and maintenance levels creating the conditions for better forecasting.

The oil pump case

One example of how implementing this AI-supported model in Darmstadt has improved forecasting of spare parts is the case of oil pumps. “While the old method did not identify any demand, the AI model predicted five units – actual consumption was six. With delivery times of around 500 days, this determines whether a vehicle is out of service or remains operational”, DB Cargo explained.

The ongoing conversation about ETCS – its cost, its pace of rollout, and whether alternatives might serve the industry better – is one the rail sector genuinely needs to have. The system embodies three decades of hard-won interoperability work across dozens of national signalling regimes and regulatory bodies. That institutional weight is not trivial. Critics who question its economics and timelines are right to do so; defenders who warn of fragmentation risk if it were abandoned outright are also right. Both arguments are valid.

But both are also talking about the same layer of the problem: signalling. And that may not be where the most urgent capacity gains are hiding.

Consider the numbers. A European Commission study published in February 2026 found that just 19% of EU rolling stock currently carries onboard ETCS equipment. By 2030, that estimate rises to 40%. More than half the current fleet – 51% – has no fitment plans at all. Full network-wide benefits, which require near-universal fleet coverage, are realistically a matter of decades, not years.

Meanwhile, rail freight’s modal share in the EU stands at roughly 17% of inland freight tonne-kilometres, essentially flat for over a decade, against a European Commission target of 30% by 2030. Freight customers are already being turned away today because the network lacks capacity. The gap between ambition and reality is not a signalling gap alone. It is an operational gap – and it is widening.

The lesson from a network that did scale

In the 1960s, telecommunications engineers faced a version of this same challenge. The telephone network operated on what became known as circuit switching: every call reserved a dedicated line, end to end, for its entire duration. The system worked. It was reliable, safe, and technically sound. It was also catastrophically inefficient. The reserved wire sat idle most of the time. Scaling was exponentially expensive. The economics broke down the moment volume grew.

Packet switching did not replace the cables, but it changed the protocol for using them. Data was broken into packets, each finding its own path through the network dynamically. There was no reserved capacity, no end-to-end pre-allocation, and no idle infrastructure. The physical layer remained largely the same; the operating logic was reimagined entirely. The internet followed. Then email, video, cloud computing, and the digital economy as we know it – all built not on new cables, but on a smarter way to use the ones that already existed.

European rail still operates on the circuit-switching model. Every train holds a dedicated slot through the network, allocated end to end. Even if only a portion of a freight train needs sorting, the entire consist uses up a yard slot upon entry, delaying the onward movement of the wagons that did not require sorting. A delayed service creates ripple effects across adjacent paths. Capacity is consumed by buffers and waiting time, not by the movement of goods. The tracks exist. The wagons also exist. What has been missing is a different way of thinking about the operating protocol.

Operational innovation as an overlay, not a replacement

The question worth adding to the signalling debate is this: what is rail’s equivalent of packet switching? What operational concepts allow trains to use existing slots more dynamically, within the signalling infrastructure already in place, without waiting for a network-wide upgrade to unlock them?

One emerging direction is the ability for train sections to couple and decouple at operational speed. Take the shunting yard scenario: a freight train where only part of the consist requires sorting no longer needs to commit the entire train to the yard. The section that needs sorting diverges on the run; the rest continues on the main line, preserving the slot from the outside rather than consuming it from within.

The yard, now freed from holding the full consist, can dispatch an already-sorted section that couples on the run with the moving section. The same slot enables two operations. No new track is required. Neither are new signals, nor a change to the signalling certification. The logic of the network changes without the network itself changing.

This is what an overlay approach looks like in practice: operational concepts that sit on top of existing infrastructure, extracting capacity that the current system structurally leaves unused every day. It is not a replacement for signalling modernisation. Rather, it is a parallel track of progress that does not have to wait for the long migration to ERTMS to complete.

Layers, not sequences

The deeper lesson from the internet is not that infrastructure does not matter – it is that infrastructure and operating protocol evolve best together, in parallel, not in sequence. The physical network mattered enormously; packet switching would have had nothing to route across without it. But framing new protocols as something to pursue only after the infrastructure is fully upgraded would have been a costly mistake. Both layers advanced simultaneously, each making the other more valuable.

The same logic applies to rail. ERTMS migration should and will continue; moving block and digital supervision will eventually deliver the headroom that fixed-block signalling cannot. That work is necessary and worth doing well. But treating it as a prerequisite for operational progress means accepting a decade or more of avoidable stagnation on a network that is already turning away demand.

ERTMS migration should and will continue

The industry needs to think in layers: the long signalling migration on one track, and a parallel investment in overlay operational concepts on another. Modular train formation, smarter slot logic, on-the-run coupling and decoupling – innovations that do not require the signalling layer to change before they can begin delivering value. Both advancing simultaneously. Both contribute to a network that can actually carry the freight volumes Europe needs it to carry if the 30% modal share target is ever to be more than an aspiration.

Rail has the infrastructure. The slots exist and the tracks are there. What the internet showed us is that the operating logic is not fixed: rather, it is a choice. And choosing a smarter one does not require waiting for the hardware to catch up.

Is the industry ready to have that conversation in parallel with the signalling debate – or will operational innovation remain a second-tier topic until the migration is done?

Currently, more preliminary DAC tests are being carried out. The first ever commercial pilot was launched in Germany in the summer of 2025, while in Sweden the ability of DAC to work under extreme weather conditions was tried. Currently Austrian Rail Cargo Group is testing DAC on different types of wagons and will continue to do so for the whole of 2026.

The DAC cost problem

One of the main points raised by DAC critics is the cost and who will have to bear them. Given the lack of clear guidelines from EU institutions, estimations vary and seem to be increasing each time they are reviewed. Currently, many agree that equipping a unit should cost between 22,000 and 25,000 euros, while the total cost of the project is somewhere around 15 billion euros.

Cheaper than new infrastructure

The technology would allow for bogies to automatically increase or decrease their width when the rail gauge changes. Ukraine has the same rail gauge as Russia (1520mm). The idea of implementing a standard gauge (1435mm, like most of Europe) was already in the air, but the war intensified the need for solutions. Given the high costs of building new infrastructure, variable gauge bogies gained in popularity.

Spain also has a different gauge compared to the rest of Europe (1600mm) and has been working on variable-gauge bogies since at least 2014. Thus, their expertise could prove significantly useful for Ukraine and its rail freight sector, which is in desperate need of new connections towards the West. Possible cooperation between the two countries was first explored in the summer of 2023 with the signing of a MoU in March 2024.

Fuel additives generally mean adulterating diesel with additional chemicals. This trial will assess whether Fuelcare’s additive can enhance diesel engine performance. The partners also want to measure any reduced emissions and extend the distances GBRf locomotives can travel between refuelling.

Committing biocide to protect engines

Fuelcare is a UK-based specialist in fuel quality management, industrial additives, and biocides, serving sectors from rail and marine to aviation and offshore energy. The Shrewsbury-headquartered company provides products and services designed to prevent microbial contamination, maintain fuel integrity, and optimise diesel and hydrocarbon fuel performance. Its expertise includes fuel testing, laboratory analysis, additive injection systems, and cloud-based monitoring solutions, all supporting operational reliability and regulatory compliance.

Biocides are additives that prevent microbial growth in diesel fuel, helping to keep engines and fuel systems clean, reliable, and free from corrosion or filter blockages. The trial using the technique aims to evaluate its effectiveness in an operational environment. Running until 22 March, it is among the largest fuel additive initiatives undertaken by any UK rail freight operator. There is a Europe-wide drive towards cleaner fuels, even though rail freight offers a significantly cleaner way of moving goods, when compared with other diesel-powered options – notably road haulage. Success could improve operational efficiency across GBRf’s fleet and reinforce the company’s commitment to a greener, more sustainable railway.

Perfect partners

It’s no surprise that the Peterborough-based operator should be the partner in such a trial. GB Railfreight (GBRf) continues to take a progressive approach to modern traction, combining operational efficiency with environmental performance.

A key part of this strategy is the significant leasing of a fleet of bi-mode Class 99 locomotives, capable of running on electric or diesel power. This flexibility allows GBRf to reduce emissions, operate across both electrified and non-electrified routes, and support a more sustainable rail freight network while maintaining high levels of reliability and service.

The pilot in Sweden also aimed at testing DAC under extreme weather conditions, but with a much heavier train moving steel. After country-specific tests, next year it will be time to take it a step further with large-scale ones. The plan is to equip 250 wagons and 15 locomotives with DAC and have them run throughout the Old Continent.

On the operational side, Groenewoud underlined that reality rarely matches the plan. Planning is less a static timetable and more “continuous re-planning” across multiple constrained resources: paths, locomotives, wagons, driver duties and repositioning. A disruption on one axis quickly cascades across the rest. Back-office systems face different pressures. Here the litmus test is throughput with a lean team. Feature completeness and ease of use matter most, while a brief outage is inconvenient rather than catastrophic.

Single-wagon load: the toughest digital puzzle

Not all freight is equal from a software perspective. Block trains on simple A–B lanes are well served by many vendors. The difficulty rises sharply with single wagonload (SWL). Few tools tackle this well, Groenewoud noted, describing it as a genuinely complex multi-resource puzzle. The business challenge compounds the technical one: SWL flows are often less profitable, yet shippers expect providers to handle both block and SWL traffic. Operators that cherry-pick only the straightforward work risk losing the entire account.

Why non-functionals now outrank features

Compared with 20 years ago, Groenewoud argued that non-functional requirements — reliability, performance, scalability and especially cyber security — are now “more dominant than the functional requirements”. Generative coding assistants can spin up basic business applications quickly, but turning them into mission-critical systems remains hard because the craft lies in meeting those non-functionals at scale. “A generative ‘programmer on your shoulder’ can build features. It cannot, by itself, deliver the non-functional part,” he said.

AI’s role: three stages of impact

Ab Ovo is already experimenting with voice-to-process tooling that converts spoken descriptions of workflows into business process models and starter applications. The company is working to embed safeguards for security, scalability and performance from the outset, and to align outputs with “green software” principles.
Energy, data centres and ‘green software’

As AI scales, energy use in data centres will grow. Policymakers are likely to constrain capacity in some locations, making software efficiency a strategic concern. Groenewoud advocates for “green software principles” and notes that language choice at runtime matters: carefully engineered low-level implementations can reduce consumption in production environments. “You’d better make sure mission-critical applications are extremely efficient for using less space in a data centre,” he said.

People, knowledge and the adoption curve

Despite the hype, AI will not replace locomotives, wagons or cargo. Value will accrue around the surrounding processes, where the business case remains decisive. A near-term priority is knowledge retention as experienced staff retire. Here AI can act as a persistent memory layer for procedures, constraints and best practice, shortening analysis cycles for new routes or customers and reducing reliance on large external consulting teams.

Bottom line

For rail freight operators, the software brief is crystallising. Keep trains moving through robust, cyber-secure platforms engineered for reliability and performance. Use AI to compress cycle times, predict needs and simplify interfaces — but don’t mistake fast feature generation for mission-critical resilience. And build with energy efficiency in mind, because scarcity in the data centre could soon be a competitive factor.

Mainline railways were historically – and many still are – controlled by a wide range of national signalling systems. In the mid 1990s, when cross-border locomotives and multiple units were introduced, the need arose for a unified European signalling and communications system to replace the old ones.

This took the name of European Train Control System (ETCS) and was enhanced by GSM-R communications technology to become the European Rail Traffic Management System (ERTMS), maintaining the same block system philosophy in which trains are separated by fixed blocks with lengths that are at least as long as the braking distance of trains running on the line.

This leads to intervals between trains that are longer than the necessary safe distance between them, which is determined by the braking distance between the moving end of the 1st train and the front of the 2nd plus a safety margin.

This is called “moving block” or “ETCS Level 3” and has been discussed for decades, but there is no concrete implementation in view for the next years or even decades. It is needed above all on densely trafficked lines such as major parts of the Scandinavian – Mediterranean and the North Sea – Rhine – Mediterranean TEN-T corridors and specific bottlenecks around Lyon, Île-de-France and the Baltic capitals. Awaiting infrastructure upgrades, a special focus should therefore be placed on increasing capacity at the bottlenecks.

Much has changed since the eighties

Since ETCS development, IT has rapidly advanced. Processor speeds are ~10,000x faster, and mobile data transmission went from 0.2 to 10,000 Mbit/s. While slow, unreliable data transmission doomed 1980s remote brake monitoring attempts, this is no longer an issue. Despite this, rail infrastructure managers plan to spend decades implementing the expensive and already dated ETCS across Europe, with full rollout anticipated a century after its initial introduction.

Implementing ETCS could be a suboptimal solution. New and more advanced signalling systems are available. Some rely on fixed signalling installations and some do not require them. Examples include the German Aerospace Centre’s (DLR) TrainCAS, a decade-old train-to-train communication system operating successfully on the Harzer Schmalspurbahn with potential for SIL4 upgrade even in dense fog at high speed.

The French Urbanloop, developed by University of Lorraine students, is a small, AI-assisted urban transport facility using pods that follow each other closely, localised by passive lineside beacons, and has been operating in Paris.

Ecotrain, also in France, is developing a solution for driverless trains on secondary lines to exchange data with level crossings to prevent collisions. Given that modern cars have sophisticated control systems for collision avoidance and autonomous driving (see the image above), it is unclear why simple ETCS train equipment costs hundreds of thousands of euros, when a complete Urbanloop pod or a high-tech car costs only a few thousand. See the price comparison in the graph:

To maximise railway capacity (after prioritising safety, switch control, timetable adherence, and energy efficiency), the signalling system is crucial. It must accurately track the position, direction, and speed of every train within its area. This essential information enables the safe organisation and optimisation of all train operations.

📌 Possibilities for Train Localisation and Safe Operation (click to expand)

A number of possibilities for the determination of the location and provision of safe operation of a train have been used or can be used. All have their own drawbacks when it comes to precise and at the same time reliable positioning.

• Track circuits for positioning and train integrity supervision.
• Axle counters for positioning and train integrity supervision.
• Fixed electric contact shoes for triggering braking.
• Electromagnetic solenoids for triggering braking.
• Continuous lineside data transmission antennae for positioning and speed control.
• Discrete lineside balises for positioning and speed control.
• Discrete passive lineside tags for positioning.
• Radio connection for speed control.
• Satellite-based positioning (GPS, Galileo, Starlink).
• Wheel revolution supervision for positioning and speed and acceleration/deceleration measurements.
• Odometry for positioning and speed and acceleration/deceleration measurements.
• Local earth magnetic field variations.

As there is now a sufficient number of different independent localisation methods available – each with varying technologies, safety certifications (up to SIL4), availability, and reliability – moving block operation (where trains dynamically adjust their spacing based on real-time data) becomes feasible. This is where we can start to think about alternatives to ETCS.

An idea for a better alternative to ETCS

A proposed signalling system should rely on three of these three localisation methods. Normal operation of trains at maximum speed and minimum headway shall be allowed if all three systems produce matching information. Trains will still be able to operate when two systems agree but the third differs. Speed and headway will then be set to values that depend on which systems are in line with each other.

If all three systems give diverging information, trains will not be allowed to run except at low-speed emergency level (typically “on sight”, probably with a maximum speed of 30-40 km/h) through secured human authority procedures.

Railways are systems and the interaction between their subsystems must be managed. It is therefore also imperative to talk about the detection of train completeness, i.e. loss of wagons. These must be secured by on-train equipment depending on the type of train.

For trains coupled with standard “screw” couplings, it is a fact that today no solution for SIL4 train integrity detection exists for serial operation even if several demonstrations have already proven some kind of feasibility. Therefore, a certain level of redesign of freight trains (and adaptations to passenger trains) will be needed, likely by equipping freight trains with electric power and digitalising them.

It is possible to quite quickly introduce such a modern highly responsive IT-based system to allow train operation under moving block conditions and thus increase line capacity, mainly for freight trains. This could allow a drastic cost reduction for signalling systems.

It could eliminate the need for many of the expensive lineside elements and their cabling, such as track circuits and axle-counters, beacons (balises) that require frequent inspection and maintenance and that are prone to meteorologic impacts and vandalism.

It will also lead to a reduction in the number of interlockings and control centres. In combination with other improvements in infrastructure, terminals and rolling stock design, it will entail a substantial boost at a fraction of the cost of ETCS. It must be designed in such a way as to allow a gradual introduction in mixed operation with ETCS or another legacy system. As long as other trains operate nearby under a traditional signalling in mixed operation, trains that are equipped with the new system must run according to the rules of that system.

Years, not decades

Modern trainborne IT technologies, which require no extensive fixed infrastructure, can be implemented as an overlay and eventual replacement for legacy signalling systems. This approach facilitates a rapid transition to moving block operation on critical bottlenecks. As a result, significant improvements and cost reductions for freight lines can be realised within years, rather than decades.

To ensure an effective improvement of rail transport, political support is mandatory. Politicians should not only suggest but also finance the ideas and transform them into legal instruments. The railway sector at all hierarchy levels should take benefit from new developments in the aerospace and automobile industries. But the full impact on capacity will only materialise if synergies with parallel improvements in logistics, infrastructure design and maintenance, energy supply and freight train design are integrated.

Do you want to share your view? You can reach out to the RailFreight.com editorial team, or to Reinhard Christeller via the button below. You can also leave a comment.

Belgium clearly implemented ETCS at lightning speed. At the same time, the country did not have much choice in implementing the technology. The 2010 rail accident at Buizingen, which left 19 dead and 171 wounded, highlighted the shortcomings of the nearly one century-old national safety system. It could not prevent trains ignoring red signals.

A patchwork of ETCS Levels

Since Belgium already had in-house ETCS expertise after it had taken the first implementation steps in 2009, it opted to accelerate that implementation. In order to do cover the entire mainline network with ETCS within an acceptable timeframe and at a reasonable cost, Belgium chose to install a mix of three variants of ETCS:

• ETCS Level 1 Full Supervision (ETCS L1 FS)
• ETCS Level 2 Full Supervision (ETCS L2 FS)
• ETCS Level 1 Limited Supervision (ETCS L1 LS)

Infrabel chose to implement ETCS L1 FS in places where that was already planned before the 2011 Masterplan, where ETCS L2 FS was not technically possible (such as in large stations), or on short sections between ETCS L1 FS zones. Otherwise, the infrastructure manager installed ETCS L2 FS, except on the more quiet sections. In the latter case, it installed ETCS L1 LS.

What is the difference between Full and Limited Supervision? Infrabel explains:

With ETCS Level 1, both Full and Limited Supervision, a train’s on-board computer receives information about the maximum permitted speed, whether the next signal is open or closed, the gradient of the tracks, etc. With Limited Supervision, the on-board computer receives this information over a shorter distance than with Full Supervision.

Furthermore, with Limited Supervision, this information is not visualised on the train driver’s screen. ETCS Limited Supervision is a system – a mode of ETCS L1 – that runs in the background. The train driver looks outside and follows the signals, as in situations without ETCS. If he or she does make a mistake, for example by driving too fast, the system intervenes and performs an emergency brake application. With Full Supervision, the train driver sees all the information on the screen in the driver’s cab.

The present day

That brings us to 2026. Fifteen years after the 2011 Belgian “ETCS Masterplan” and 2.8 billion euros later, the mainline network is ready for ETCS-only operations. Some 80% of expenditures went to interlocking, a base system for the control of switches and signals. Belgium is making the switch to the digital interlocking systems SmartLock and SIMIS W.

Milestone completed. Still, that does not mean that there are no more challenges on the horizon. In the coming decade, the 2G communication technology GSM-R should reach end-of-life status and be replaced with the 5G system FRMCS, which offers increased reliability, speed and higher levels of cyber security. This will require more expensive retrofitting and infrastructure work, just as Belgium has completed ETCS implementation.

Infrabel expects that we will end up with a dual system of both GSM-R and FRMCS, and so it does not worry that GSM-R will be defunct from one day to the next. The infrastructure manager is preparing to keep its GSM-R network operational for longer to retain its communications infrastructure.

Simultaneously, Infrabel is preparing GSM-R towers for FRMCS installation. However, not all equipment is commercially available or is even in existence, which means that planning for the switch is difficult. However, Infrabel adds that it maintains its own standalone rail fibre optic network, which offers security and a lack of interference. The “backbone” is there, so the migration to FRMCS has been “prepared” already. To complete the upgrade, Infrabel just needs to change the hardware of the communication system.

Who coordinates problem analysis and resolution?

Operationally, ETCS-only operations also bring about several challenges. Hans-Willem Vroon, director of the Dutch rail freight association RailGood, had earlier explained to RailFreight.com that ETCS operations in the Netherlands are not flawless. A key problem concerns independent investigation of incidents and the allocation of responsibility. The many involved stakeholders, such as part manufacturers, software developers, the infrastructure manager operators and train drivers make that process difficult. “As is so often the case in chains, the temptation is great for chain players to hide behind each other and for important players to disappear.”

The question is who, then, takes responsibility for the independent investigation of incidents or determines where operations derail?

Infrabel explains that it organises quarterly consultations with stakeholders by default. Moreover, the infrastructure manager says that it has traffic reports of all delays that occur on its network. If ETCS is involved in the incident, it consults with stakeholders to identify the problem. This, according to Infrabel, “goes quite well”. In other words, there is no formally appointed entity that takes responsibility, but these processes take place in good faith and in a cooperative spirit.

However, Infrabel does not expect system-breaking issues to arise. The use of technology is as watertight as possible with the help of certificates and homologation procedures. Still, not every supplier implements everything correctly – or identically to one another, despite certification – 100% of the time. There are some hiccups from time to time due to cost reductions, which can cause operational issues.