Automating human-oriented tasks and complementing humans in their work with rail vehicles, may it be in relation to management, maintenance, or network operation, can be facilitated by the use of artificial intelligence (AI), which carries the promise to make human-oriented tasks more efficient and less error-prone. It is the firm belief of the authors, given the global lack of skilled rail expertise and the increased complexity and efficiency requirements of the operational environment that AI will become a complementary companion to humans in that work environment with the potential for making railways more efficient. The aim of this chapter is therefore to provide a backgrounding regarding some of the beneficial aspects of AI and, based on concrete examples, give recommendations on how to avoid typical pitfalls when using and implementing AI in your organization.

Vehicles are complex systems composed of multiple interdependent subsystems. A design change in one subsystem can lead to either beneficial or detrimental effects on others, making it essential to understand how design choices propagate through the system. This paper focuses on the traction chain in rail vehicles, specifically the interaction between two critical subsystems: the traction motor and the inverter. Since the inverter supplies the current and voltage required by the motor, changes in motor design can affect the inverter’s thermal behaviour. We analyse this interaction to understand how motor design influences the temperature evolution of inverter power electronic components. To achieve this, we apply a framework that integrates network theory, structural equation modelling (SEM), and sensitivity analysis. First, analytical models of a three-phase induction motor and an inverter thermal model are developed. A reverse breadth-first search is used to identify all input parameters that influence the inverter temperature. Global sensitivity analysis (GSA) isolates the most influential inputs, enabling the construction of a reduced network graph. Then, SEM and local sensitivity analysis (LSA) are applied to quantify the relationship between motor design parameters and inverter temperature, yielding a coefficient that captures the strength of the dependency. This approach provides an alternative representation of the multidisciplinary interactions between subsystems. It also helps identify key change propagation paths, making it easier for designers to anticipate and manage the consequences of design changes. By reducing analytical complexity and clarifying subsystem interdependencies, the framework supports more efficient early-stage design, potentially reducing the number of iterations needed to reach a satisfactory solution.

Predicting rolling contact fatigue crack hot spots or regions with increased local driving forces in rails is challenging due to the wide range of factors that influence crack initiation. Rail sections experience fluctuating creepage conditions, contact positions, and loads throughout their lifespan, influencing the development and location of fatigue cracks. A new computational method is proposed that predicts the orientation and regions prone to rolling contact fatigue cracks under realistic service loading. It combines multi-body simulations, finite element analysis, and critical plane approaches. A novel multi-variable sampling technique simplifies loading spectra into representative traction profiles, which are then analyzed using finite element analysis and the Smith–Watson–Topper damage indicator parameter (DIP<inf>SWT</inf>). The maximum DIP<inf>SWT</inf> value identifies the critical plane and potential crack orientation. A case study on the Swedish heavy haul train line (Malmbanan) considers measured traffic and loading conditions, analyzing the wheel load spectrum for a 384 m long section of a R = 450 m curve. Results show that the DIP<inf>SWT</inf> is highest for the locomotive with a loaded payload configuration, with a maximum value of 3.84 × 10⁻⁸ located at 38.59 mm from the lower gauge face corner. The DIP<inf>SWT</inf> critical plane aligns with experimental measurements of RCF cracks orientations near the gauge corner. This computational method, when combined with other predictive tools, can efficiently identify conditions that lead to RCF cracks and determine their possible locations and orientations in railway tracks.

Active wheelset steering can improve the curving performance of railway vehicles and thus reduce wear. Several control strategies have been proposed to achieve a perfect steering condition which may require feedback signals that are difficult to measure. This study proposes a control strategy for an active wheelset steering system via wheelset angular velocity measurements aimed to minimise longitudinal creepages in curves. The desired wheelset angular velocity is derived from the relation of longitudinal creepages and the wheelset movement in curves. Co-simulations with a conventional railway vehicle with two two-axle bogies are carried out for 24 cases. First, the strategy based on actual wheel-rail geometry is used to evaluate the effectiveness of the proposed control strategy; in the second step, a simplified strategy is tested using the approximated equivalent rolling radius approach. Curving performance indicators, including wheelset movements and wheel-rail wear number, are used to evaluate the performance of the control system. The results indicate the effectiveness of the proposed control strategies. Challenges and practical considerations are also discussed.

Wheel-rail contact friction coefficient is often assumed to be constant through the entire contact patch for the calculation of surface traction. In reality, however, the friction value in a certain point decreases when transitioning from adhesion to slip regimes. Including this friction coefficient behaviour in the estimations of surface traction on the contact patch can potentially provide more accurate calculations of wear and rolling contact fatigue (RCF). In the present work, a slip velocity-dependent friction coefficient is implemented in the tangential contact solver using the concept of ‘Friction Memory’. The effect of this implementation on traction estimations and on the prediction of wear and RCF is analysed by comparing the results with a case with constant friction coefficient in the contact patch. Furthermore, the slip velocity-dependent friction coefficient provides a creep curve with a maximum creep forces value, and a decreasing creep force for higher creepages. This is commonly known as one of the possible mechanisms of curve squeal noise generation. The results provide insights into the likelihood of curve squeal generation, and an on-set curve squeal noise detection technique is proposed that also accounts for the influence of profile changes due to wear.

Negotiation of turnouts imposes challenges for an active wheelset steering system due to lack of smooth transition curves and existence of rail discontinuities. These affect the performance of both the vehicle and control system in turnouts, which is investigated in this paper. The simulation is carried out with a conventional railway vehicle with two two-axle bogies passing through a crossover onto a parallel track. The Swedish 60E1-R760-1:5 turnout is used in this study. The results reveal that an active wheelset steering system can decrease the wheel-rail wear index. However, peaks in wear number take place in the switch toe and the crossing nose and they are considerably higher than other regions. A control scheme with preview is proposed by considering wheelset lateral positions on discontinuous rail profiles to avoid flange contact. The proposed control system with preview results in a further reduction of the maximum wear number by 66%.

The advent of silicon carbide (SiC) semiconductors in electric traction enables several benefits, including the shift to passive cooling. However, it requires a conjugate heat transfer analysis to understand the temperature distribution and variation. While steady-state solutions exist, transient conditions in rail vehicles remain challenging. This paper develops two analytical models to predict temperature distribution and variation, validated against numerical simulations. An electric motor model estimates power losses in the converter, defining heat dissipation. The complete model is tested under realistic drive cycles, linking operational conditions to power losses and free flow speed. The results show the model effectively captures temperature variations, with higher losses during acceleration and larger temperature surges of around 70 K at lower speeds. Furthermore, the temperature at the junction was observed to be 20 K higher than at the base position and to exceed 420 K at a more downstream location. Thus, the proposed method captures the temperature variations considering different physical effects with reasonable accuracy and significantly faster computation times than transient numerical simulations.

 Vehicles are complex systems composed of interdependent subsystems, where a change in one component can propagate through others, causing knock-on effects and impacting overall performance. This interconnected nature makes early-stage design challenging due to limited knowledge of the consequences of design choices. Existing tools address different aspects of this gap in isolation and typically do not quantify the cascading changes i.e., knock-on effects or the total impact of an input on the output, highlighting the need for an integrated approach. This paper addresses these limitations by proposing a framework to quantify the magnitude and direction of the knock-on effects and consequently calculate the total impact by integrating curve fitting, sensitivity analysis, and structural equation modeling. This framework derives representative functions, quantifies the influence between every interacting factor pair, and calculates the total impact. To demonstrate the framework, a rail traction system is used as case study. Results indicate that while rated power had comparatively less impact on the output than frequency, rated power had 6 times as many paths and 1.86 times as many vertices as frequency, indicating higher knock-on effects. To validate the structural model, for each input and output, the total impact value was compared with the direct sensitivity coefficient, yielding a maximum error of 3%. Additionally, comparison with results of Global Sensitivity Analysis confirmed consistent input rankings and influence directions. This demonstrates the framework’s capability to quantify knock-on effects and total impact while preserving true input-output relationships.

This work presents an analysis of the sensitivity and impact of various vehicle parameters on the estimation of ballast settlement cost in passenger and freight systems in the Universal Cost Model (UCM). The study compares the vertical force obtained from multibody simulation results to analytical solutions for passenger and freight wagons. These results highlight the significance of considering unsprung mass, vehicle type, and vertical suspension parameters when estimating ballast settlement cost with the analytical representation of the P2 force. For passenger vehicles, the simulated P2 force is sensitive to all investigated parameters, with an average sensitivity of approximately 18%. The choice of method for determining vertical force for simulated track irregularities also influences the impact of these parameters. The simulated P2 force for the track section with track irregularities exhibits a higher percentile difference compared to other track cases, indicating that the parameters have a greater impact on the simulation of P2 force. In contrast, for freight vehicles, the simulated P2 force shows sensitivity only to two out of seven investigated parameters: the vertical stiffness of the outer and inner coil springs. The impact of these parameters is dependent on track irregularities and the chosen method for determining vertical force. The standard deviation measure has the highest impact on the outer coil spring, while the 99-percentile and mean value measure show a medium and minimal impact, respectively. The inner coil spring has an overall minimal impact across all track cases and comparison methods. These results show that the existing UCM methods can be improved by incorporating the identified parameters, thereby enhancing the accuracy of ballast maintenance cost estimation modules.

This compendium is mainly intended for the MSc education in rail vehicle engineering at KTH Royal Institute of Technology. It now appears in its 4th edition. Together with the compendium "Part 2: Rail Vehicles", they comprise 20 chapters and more than 600 pages.