Sharp changes in the vertical stiffness levels of a track can increase train and infrastructure deterioration to the point where there is a serious risk of a derailment. Major overloading and unloading forces are created between the different track and vehicle components. This phenomenon has grown in importance as the operational speeds of trains have increased with the expansion of high-speed lines. In order to solve this problem a method has to be found to smooth the changes in vertical stiffness levels along the track. In the present paper, the combination of transition regions and under- sleeper pads (USPs) has been studied. The research has been performed by means of a dynamic vehicle–track interaction model created by synthesizing a series of sub-models of individual effects. The analysed variables allow various track configurations, train travelling speeds and the stiffness of the USPs to be investigated. The obtained results show that combining transition zones with USPs pads allows more homogeneous vertical stiffness levels to be achieved along the tracks which results in improved dynamic behaviour of the vehicle–track system.

Railways are an efficient transport mean with lower energy consumption and emissions in comparison to other transport means for freight and passengers, and yet there is a growing need to increase their efficiency. To achieve this, it is needed to accurately predict their energy consumption, a task which is traditionally carried out using deterministic models which rely on data measured through money- and time-consuming methods. Using four basic (and cheap to measure) features (train speed, acceleration, track slope and radius of curvature) from MetroValencia (Spain), we predicted the traction power using different machine learning models, obtaining that a random forest model outperforms other approaches in such task. The results show the possibility of using basic features to predict the traction power in a metropolitan railway line, and the chance of using this model as a tool to assess different strategies in order to increase the energy efficiency in these lines.

In the National Spanish railway network, two types of track gauge with continuous welded rails are currently in use: the “Iberian” wide gauge (1668 mm) and the standard gauge (1435 mm). In order to improve links and freight traffic between different lines and with the rest of Europe, a dual gauge track with three rails was developed. This solution modifies the classical track configuration, so it is necessary to develop new methodologies and studies to understand its behavior. Among other loads applied on a continuous welded rail track, a considerable rise in temperature induces compressive stresses in the three rails that can lead to lateral track buckling. Moreover, on dual gauge tracks, the addition of the third rail increases the axial compression, which may lead to track instability. For this reason, a three-dimensional continuous welded rail model is developed in this study to be used for dual gauge track buckling analysis on straight tracks subjected to temperature load. The continuous welded rail dual gauge track model consists of beam, solid and spring elements, in which a non-linear behaviour of the ballast is considered. The results obtained may be used to predict the buckling capacity of the continuous welded rail on dual gauge tracks with respect to different parameters such as lateral resistance, lateral imperfections, sleeper spacing or torsional stiffness.

This paper presents the application of machine learning systems based on neural networks to model the energy consumption of electric metro trains, as a first step in a research project that aims to optimise the energy consumed for traction in the Metro Network of Valencia (Spain). An experimental dataset was gathered and used for training. Four input variables (train speed and acceleration, track slope and curvature) and one output variable (traction power) were considered. The fully trained neural network shows good agreement with the target data, with relative mean square error around 21%. Additional tests with independent datasets also give good results (relative mean square error = 16%). The neural network has been applied to five simple case studies to assess its performance – and has proven to correctly model basic consumption trends (e.g. the influence of the slope) – and to properly reproduce acceleration, holding and braking, although it tends to slightly underestimate the energy regenerated during braking. Overall, the neural network provides a consistent estimation of traction power and the global energy consumption of metro trains, and thus may be used as a modelling tool during further stages of research.