In response to the growing demand for rail transport, next-generation signalling systems are increasingly investigated by the railway community. In particular, the concept of Virtual Coupling (VC) is progressively gaining ground thanks to its potential ability to reduce safe train separation to less than an absolute braking distance. That enables trains to move synchronously in a vehicle-to-vehicle radio-connected convoy. One of the major concerns associated with this concept is the safe and effective control of trains in a convoy when considering varying train resistances and risk factors due to, e.g., sudden degradation in the train and communication performance. This paper develops a novel Predictive Artificial Potential Field (PAPF) approach for safe and effective real-time train control under realistic VC operations. The proposed approach uses a realistic homogeneous strip model of train motion. Moreover, it incorporates a dynamically changing safety margin to take into account risk factor occurrences, such as delays in train control and communication, or sudden emergency braking applications. A simulation-based assessment of the developed method is performed for a high-speed rail corridor in China. Results show that the proposed PAPF control algorithm effectively supervises the safe train separation, preventing activation of emergency brakes even when risk events occur. The method contributes to advancing the state of the art on VC train control.

Along with much of Europe and the global trend towards in-cab signalling, Great Britain (GB) rail is transitioning to Radio-based European Train Control System (ETCS). In collaboration with several universities and research partners, Network Rail have undertaken a programme of R&D to build on the opportunities that Radio-based ETCS offers, including the move towards automatic train operation (ATO), and integrated, centralised traffic management systems that maximise the potential in capacity, performance and energy efficiency for passenger and freight customers. Bringing the train planning and timetabling capabilities into the modernised, data driven, information rich world is a significant puzzle piece of turning opportunities offered by the signalling and control technologies into the day-to-day operations of the railway. This paper covers the research carried out into the characteristics of a radio-based ETCS railway that can be analysed for a goal-based state of the art train planning capability. It considers the advancement of tools, techniques, processes and skills that are required to plan, operate and regulate the railway through automatic train operations and future traffic management systems, ultimately harmonising planning, real-time operations and post-operations performance analysis.

To further improve the capacity on the European railway network, next-generation distance-to-go signalling systems are being developed in the context of the European Train Control System (ETCS). This paper investigates the impact of track discretisation granularity on conflict detection and resolution for ETCS with onboard train integrity monitoring. The study enhances a previously developed model for fixed-block distance-to-go signalling by introducing a track discretisation procedure and reformulating safe train separation constraints at switches. The assessment is performed on a junction and a corridor case study, using track discretisations with maximum section lengths from 50 to 800 m. Though finer discretisations potentially improve the model objective, computation times quickly increase. While the results show minimum effects of the track discretisation on the conflict detection and resolution, they suggest that maximum section lengths of 200 or 400 m may offer a good balance between solution quality and computational complexity, depending on the track layout and traffic density. Generally, reliable rescheduling decisions can already be obtained with a 800-m discretisation.

Widespread congestion in metro systems often hinders passengers from boarding the first arriving train, making them compelled to adopt an alternative route, some of which involve travelling backwards. While this travel strategy has direct consequences for forecasting passenger flow distribution in congested networks, little is known about the travelling backwards phenomenon and why people adopt this travelling behaviour. The aim of this study is to understand passengers’ perception of time in various segments considering travelling backwards. To achieve this, we develop a route choice model using revealed preference data from smart card records. We find that passengers exhibit a greater aversion to waiting time and onboard time while travelling backwards. Specifically, passengers perceive each minute spent waiting on the turn-back stations’ platform as equivalent to 1.97 min on the origin platform. Similarly, each minute spent onboard the backwards train is perceived as equivalent to 1.24 min on the forwards train. Ignoring this difference in perception would result in the underestimation of the expected social benefits of demand management policies. Finally, we assess the potential benefits of travelling backwards under various passenger flow conditions, offering valuable policy insights regarding whether and how this behaviour should be regulated or promoted.

The increasing train traffic over railway networks stretches the demand for capacity of railway yards and rolling stock maintenance locations, which increasingly limits performance and further growth. Therefore, the scheduling of rolling stock maintenance and the choice regarding optimal locations to perform maintenance is increasingly complicated. This research introduces a Maintenance Scheduling and Location Choice Problem (MSLCP). It simultaneously determines maintenance locations and maintenance schedules of rolling stock, while considering the available capacity of maintenance locations. Solving the MSLCP using one large Mixed Integer Programming appears not to perform well enough. Therefore, to solve the MSLCP, an optimization framework based on Logic-Based Benders’ Decomposition (LBBD) is proposed by combining two models, the Maintenance Location Choice Problem (MLCP) and the Activity Planning Problem (APP), to assess the capacity of an MLCP solution. Within the LBBD, four variants of cut generation procedures are introduced to improve the computational performance: a naive procedure, two heuristic procedures and the so-called min-cut procedure that aims to exploit the specific characteristics of the problem at hand. The framework is demonstrated on realistic scenarios from the Dutch railways. It is shown that the best choice for the cut generation procedure depends on the objective: when aiming to find a good but not necessarily optimal solution, the min-cut procedure performs best, whereas when aiming for the optimal solution, one of the heuristic procedures is the preferred option. The techniques used in the current research are new to the current field and offer interesting next research opportunities.

As regenerative braking systems become more widespread in railways, rising attention is paid to collaborative train operations under optimized timetables to enhance regenerative braking efficiency. The effective usage of regenerative braking energy (RBE) is determined by the dynamic nature of the traction power supply network, driven by constant changes in train power and positions. Solving the power flow with multiple trains significantly, however, increases the computing time required to solve the optimization model. Most existing methods have to solve optimization problems neglecting the dynamic power flow analysis, which sacrifices the accuracy of regeneration efficiency. In order to address this challenge, we propose a data-driven model that emulates the power flow analysis and reduces the computational demands. Initially, data from both single and multitrain simulators are collected and stored in a database, from which critical information regarding train position, power, and substation power is extracted. A neural network is then used to develop a data-driven model that predicts the power of a substation in a power supply network based on train positions and powers. Case studies with Beijing Yizhuang Metro line data show that the calculation time of the data-driven model is 0.33% of the power flow simulation while keeping the accuracy above 99%. Based on this data-driven model, by optimizing train speed profile and dwell time, the energy supplied by substations can be reduced by up to 13% compared to traction optimization.

Passenger railway demand fluctuates daily, peaking at the start and end of the workday due to commuting to school and work. During the off-peak the volumes drop and most people travel for other purposes, like leisure and social visits, which results in different travel destinations. Despite this, many European Railways use fixed line plans and cyclic timetables that remain constant throughout the day. While this approach makes schedules easy to remember and provides ample off-peak travel options, it is primarily designed for peak-hour demand, making it less efficient for the off-peak. Furthermore, due to the different mix of travel purposes, a schedule based on peak-hour demand is not necessarily optimal for off-peak demand. This paper aims to combine the benefits of a cyclic timetable with the flexibility of an acyclic timetable in order to follow the time-dependent demand more closely. We propose a mixed-integer linear programming model that finds a timetable for a day consisting of several periods which each have its own line plan. The resulting timetable is required to be cyclic within each period and provide a good transition between the periods. The model is successfully tested on a case study with changing stopping patterns using data from the Dutch railway network, for which an optimal timetable is found. In this timetable, the transition between cyclic schedules can be done without cancelling trains or shifting trains from the new cyclic times.

In response to the escalating demand for rail transport, the concept of Virtual Coupling (VC) train operations is progressively gaining ground within the railway sector. The concept of VC aims at reducing safe train separation to less than the absolute braking distance by letting trains move synchronously in radio-connected convoys. One of the major concerns associated with VC is ensuring safe train separation considering realistic risk factors, such as heterogeneous train braking performances and varying track conditions. To address such a safe train separation problem under VC, this paper proposes a novel train control model based on the Artificial Potential Field (APF) method to safely supervise the complete braking process of trains moving in a VC convoy. The proposed model uses a homogeneous strip representation of train length and a Dynamic Safety Margin (DSM) to take into account accurate train dynamics as well as potential risk factors, due to different train acceleration/braking rates, communication delays, unexpected emergency train braking applications, and position measurement errors. The method has been applied to the case of a high-speed line in China. Results show that the APF-based control method can effectively adapt to real-time variations in train dynamics and the operational environment to safely supervise the complete train braking process and avoid collisions even in the case of unplanned emergency braking applications. The proposed APF-based approach shows promising real-time performance which can further contribute to advancing the state of the art on safe train control under VC signalling.

This paper provides an overview of the main results achieved within the Horizon 2020 Shift2Rail project named RAILS (Roadmaps for Artificial Intelligence Integration in the Rail Sector). The RAILS roadmapping process provided state-of-the-art, taxonomy, future research directions, and recommendations in three macro areas: Railway Safety and Automation, Predictive Maintenance and Defect Detection, and Traffic Planning and Management. RAILS findings shed light on the potential of intelligent technologies and provided essential guidelines for integrating machine learning into next-generation smart railways.

Automatic Train Operation (ATO) aims to enhance punctuality, energy efficiency, and reliability by automating driving tasks. Specifically, for mainline railways, an ATO onboard component generates and tracks optimised train trajectories based on time targets or windows at critical network locations, known as timing points, across train routes. These timing points and their associated constraints are specified in the Train Path Envelope (TPE), computed to ensure conflict-free operations. The generation of TPEs relies on dynamic updates of the real-time traffic plan from the Traffic Management System and real-time train statuses (e.g., position and speed). Understanding how TPEs are affected by these updates is essential for effective ATO deployment. To address this, this paper proposes a sensitivity analysis using elementary effects of a TPE generation algorithm, evaluating its response to variations in real-time traffic plans and train status updates. A real-life case study on a Dutch rail corridor with heterogeneous traffic reveals that control timing points can be introduced into the TPE as headways decrease, to homogenise traffic by aligning speed profiles and thus resolving conflicts. Timing point locations remain mostly unchanged, while their associated time windows become more sensitive when placed further along the route. Operational tolerance, which defines the latest conflict-free passing time, becomes more sensitive to headway changes and the distance from the previous stop.