Railway wheel-flat and rail surface defect modelling and analysis by time–frequency techniques

Damage to the surface of railway wheels and rails commonly occurs in most railways. If not detected, it can result in the rapid deterioration and possible failure of rolling stock and infrastructure components causing higher maintenance costs. This paper presents an investigation into the modelling and simulation of wheel-flat and rail surface defects. A simplified mathematical model was developed and a series of experiments were carried out on a roller rig. The time–frequency analysis is a useful tool for identifying the content of a signal in the frequency domain without losing information about its time domain characteristics. Because of this, it is widely used for dynamic system analysis and condition monitoring and has been used in this paper for the detection of wheel flats and rail surface defects. Three commonly used time–frequency analysis techniques: Short-Time Fourier Transform, Wigner–Ville transform and wavelet transform were investigated in this work.     

Further study on the wheel–rail impact response induced by a single wheel flat: the coupling effect of strain rate and thermal stress

A comprehensive dynamic finite-element simulation method was proposed to study the wheel–rail impact response induced by a single wheel flat based on a 3-D rolling contact model, where the influences of the structural inertia, strain rate effect of wheel–rail materials and thermal stress due to the wheel–rail sliding friction were considered. Four different initial conditions (i.e. pure mechanical loading plus rate-independent, pure mechanical loading plus rate-dependent, thermo-mechanical loading plus rate-independent, and thermo-mechanical loading plus rate-dependent) were involved into explore the corresponding impact responses in term of the vertical impact force, von-Mises equivalent stress, equivalent plastic strain and shear stress. Influences of train speed, flat length and axle load on the flat-induced wheel–rail impact response were discussed, respectively. The results indicate that the maximum thermal stresses are occurred on the tread of the wheel and on the top surface of the middle rail; the strain rate hardening effect contributes to elevate the von-Mises equivalent stress and restrain the plastic deformation; and the initial thermal stress due to the sliding friction will aggravate the plastic deformation of wheel and rail. Besides, the wheel–rail impact responses (i.e. impact force, von-Mises equivalent stress, equivalent plastic strain, and XY shear stress) induced by a flat are sensitive to the train speed, flat length and axle load.     

Modelling of wheels and rail discontinuities in dynamic wheel–rail contact analysis

A classification of wheel–rail contact is given. Difference is made between modelling of a running wheel with continuous single-point-contact, as is common practice in wheel–rail contact analysis, and a wheel with transient double- or multi-point-contact, which may occur for rail irregularities with curvatures larger than that of the wheel circumference. It is shown that application of the first model for these irregularities will strongly underestimate the contact forces as it does not describe occurring mechanisms correctly. Further, it is shown that in principle it is not possible to describe the second type of contact fully correct with a lumped wheel model. Both wheel models are formulated mathematically for some basic contact cases. Afterwards, results are applied to a linear track model. Analytical closed-form solutions are found in the frequency domain for arbitrary type of contact and numerically transformed to the time domain. Finally, the necessity is shown to avoid situations where transient multiple-point-contact may occur (like rail joints) in practice.     

Indirect method for wheel–rail force measurement and derailment evaluation

Wheel set flange derailment criteria for railway vehicles are derived and the influence of wheel–rail contact parameters is studied. An indirect method for wheel–rail force measurement based on these derailment evaluation criteria is proposed. Laboratory tests for the calibration of strain–force devices on the bearing box are carried out to determine the relationship between the applied force and the measured strain. The simulation package, SIMPACK, is used to develop a passenger car model to generate wheel–rail forces and vibration signals. Different cases are considered in this model to provide an accurate validation of the identified wheel–rail forces. A feasibility test is conducted in the Beijing Loop test line using a passenger car equipped with a set of strain gauges on the wheel set. The comparison of the force time history applied to the instrumented wheel set and that obtained using the indirect method is presented.     

A dynamic wheel–rail impact analysis of railway track under wheel flat by finite element analysis

Wheel–rail interaction is one of the most important research topics in railway engineering. It involves track impact response, track vibration and track safety. Track structure failures caused by wheel–rail impact forces can lead to significant economic loss for track owners through damage to rails and to the sleepers beneath. Wheel–rail impact forces occur because of imperfections in the wheels or rails such as wheel flats, irregular wheel profiles, rail corrugations and differences in the heights of rails connected at a welded joint. A wheel flat can cause a large dynamic impact force as well as a forced vibration with a high frequency, which can cause damage to the track structure. In the present work, a three-dimensional finite element (FE) model for the impact analysis induced by the wheel flat is developed by the use of the FE analysis (FEA) software package ANSYS and validated by another validated simulation. The effect of wheel flats on impact forces is thoroughly investigated. It is found that the presence of a wheel flat will significantly increase the dynamic impact force on both rail and sleeper. The impact force will monotonically increase with the size of wheel flats. The relationships between the impact force and the wheel flat size are explored from this FEA and they are important for track engineers to improve their understanding of the design and maintenance of the track system.     

Design and optimisation of wheel–rail profiles for adhesion improvement

This paper describes a study for the optimisation of the wheel profile in the wheel–rail system to increase the overall level of adhesion available at the contact interface, in particular to investigate how the wheel and rail profile combination may be designed to ensure the improved delivery of tractive/braking forces even in poor contact conditions. The research focuses on the geometric combination of both wheel and rail profiles to establish how the contact interface may be optimised to increase the adhesion level, but also to investigate how the change in the property of the contact mechanics at the wheel–rail interface may also lead to changes in the vehicle dynamic behaviour.     

An analytical mathematical method for calculation of the dynamic wheel–rail impact force caused by wheel flat

The simplified method to determine a vertical impact force of wheel with flat and rail interaction is presented in this article. The presented simplified method can be used to identify maximum contact force and its distribution in the contact length between the damaged wheel and the rail. The vertical impact force depends on geometrical parameters of the rail and wheel with flat, speed of vehicle and the angle of deviation of rail. This article demonstrates the influence of wheel with flat geometrical parameters, speed of vehicle to maximum contact force and its distribution in the contact zone. The obtained values of the simplified method for determination of a vertical contact force are compared with the results obtained from field measurements.     

Modelling of wear and fatigue defect formation in wheel–rail contact

The model for analysing wear and fatigue defect formation is developed based on the approaches of contact and fracture mechanics. The model includes the solution of the contact problem for the wheel and rail to find the shape, size and position of the contact zones and the contact stresses and calculation of the surface wear and the function of damage accumulation in the rail and wheel. The wear rate and the worn-profile evolution of the wheel surface are calculated using both statistic and deterministic approaches to modelling of vehicle dynamics (tribo-dynamic modelling). The influence of the evolution of the wheel–rail profiles due to wear on the damage accumulation process is analysed. It is shown that for some values of the wear rate coefficient, the wear process can prevent the crack initiation under the wheel surface.     

Out-of-round railway wheels—a study of wheel polygonalization through simulation of three-dimensional wheel–rail interaction and wear

The objective of this study is to develop a tool for investigation of wheel tread polygonalization with radial irregularities including 1 to 20 wavelengths around the circumference of the wheel. Therefore, an existing multibody system model for simulation of general three-dimensional train–track interaction (accounting for frequencies up to several kHz) is extended with rolling contact mechanics according to FASTSIM. Furthermore, the model is also modified to allow for general wheel–rail profiles. The numerical model uses the concept of an iteration scheme including simulation of dynamic train–track interaction in the time domain coupled with a long-term wear model. A demonstration example including a bogie of a subway train travelling on a straight track is presented. In the example, an initial wheel out-of-roundness (OOR) is applied to the wheels. This irregularity is based on an amplitude spectrum derived from measurements on new wheels. Simulation results show that the most important wavelength-fixing mechanisms of the wheel OOR are (i) the vertical resonance of the coupled train–track system at approximately 40 Hz (the P2 resonance) and (ii) the frequency region including the lowest vertical track antiresonance at 165 Hz, where the dynamic track stiffness is high. Only a straight track is studied, but the model allows for asymmetric train motion on such a track.     

Estimation of wheel–rail contact conditions and adhesion using the multiple model approach

This paper presents the development of a multiple model estimation approach for the identification of the adhesion limit to overcome the problem of the wheel slip/slide at the rail wheel-rail contact. The contact characteristics at the rail wheel-rail interface are both highly nonlinear and subject to changes due to exposure to external contaminations. The detection of adhesion and its changes is therefore scientifically challenging, but would provide a critical information in the control of trains to avoid undesirable wear of the wheels/track but also the safety compromise of rail operations. This study exploits the variations in the dynamic behaviour of the railway wheelset caused by the contact condition changes and applies a bank of Kalman filters designed at selected operation points for the adhesion estimation. A fuzzy logic system is then developed to identify the contact conditions by examining the residuals from the Kalman filters.     

Methodology and verification of calculations for contact stresses in a wheel–rail system

The distribution of contact stresses in the wheel–rail system is a decisive factor for the wear of elements and the safety of rail transport. Analytical calculations of stresses based on the Hertz theory can only be applied to elastic deformation of materials. High dynamic loads leading to plastic deformation (not considered in the Hertz theory) are a predominant cause of problems in the contact vicinity. These problems can be successfully resolved by applying the finite-elements method. Two- and three-dimensional test models were generated to estimate an error in numerical calculations in the MSC.MARC program. We compared the results of numerical calculations with analytical calculations. Based on the obtained results we defined the effect of parameters describing the finite-elements mesh on the calculation error for contact stresses, and adjusted mesh parameters appropriately to achieve as low as possible error in numerical calculations. We also defined the effect of material characteristics on the value of contact stresses on the wheel–rail interface.     

Physical processes in wheel–rail contact and its implications on vehicle–track interaction

Friction within the wheel–rail contact highly influences all aspects of vehicle–track interaction. Models describing this frictional behaviour are of high relevance, for example, for reliable predictions on drive train dynamics. It has been shown by experiments, that the friction at a certain position on rail is not describable by only one number for the coefficient of friction. Beside the contact conditions (existence of liquids, solid third bodies, etc.) the vehicle speed, normal loading and contact geometry are further influencing factors. State-of-the-art models are not able to account for this sufficiently. Thus, an Extended-Creep-Force-Model was developed taking into account effects from third body layers. This model is able to describe all considered effects. In this way, a significant improvement of the prediction quality with respect to all aspects of vehicle–track interaction is expected.     

Comments on ‘the Kalker book of tables for non-Hertzian contact of wheel and rail’

The ‘simple double-elliptical contact’ (SDEC) approach by Piotrowski et al. [The Kalker book of tables for non-Hertzian contact of wheel and rail. Vehicle Syst Dyn. 2017;55:875–901] generates a-symmetrical contact patches in an elegant way. This allows to extend the table-based approach for the wheel–rail creep force calculation towards non-Hertzian contact geometry. This is an important line of research, because FASTSIM is intricate for non-Hertzian contacts, whereas CONTACT requires long calculation times.

Here, we comment on the further motivation that's provided for the approach. According to the authors, ‘the spin creepage generates longitudinal creep force in non-symmetric, non-elliptical contacts’, which is ‘completely lost’ when using elliptical regularisation. We demonstrate that this mainly depends on the choice of contact origin, and that the interaction is much reduced if different choices are made. This suggests that elliptical regularisation may be viable still, if the details are properly worked out. Furthermore, we introduce the spin center and the free-rolling position as means to extend the table-based approach towards more general non-Hertzian circumstances.     

Sensitivity of the wheel–rail contact interactions and Dang Van Fatigue Index in the rail with respect to irregularities of the track geometry

This paper investigates the effects of the track geometry irregularities on the wheel–rail dynamic interactions and the rail fatigue initiation through the application of the Dang Van criterion, that supposes an elastic shakedown of the structure. The irregularities are modelled, using experimental data, as a stochastic field which is representative of the considered railway network. The tracks thus generated are introduced as the input of a railway dynamics software to characterise the stochastic contact patch and the parameters on which it depends: contact forces and wheelset–rail relative position. A variance-based global sensitivity analysis is performed on quantities of interest representative of the dynamic behaviour of the system, with respect to the stochastic geometry irregularities and for different curve radius classes and operating conditions. The estimation of the internal stresses and the fatigue index being more time-consuming than the dynamical simulations, the sensitivity analysis is performed through a metamodel, whose input parameters are the wheel–rail relative position and velocity. The coefficient of variation of the number of fatigue cycles, when the simulations are performed with random geometry irregularities, varies between 0.13 and 0.28. In a large radius curve, the most influent irregularity is the horizontal curvature, while, in a tight curve, the gauge becomes more important.     

Effect of temperature- and frequency-dependent dynamic properties of rail pads on high-speed vehicle–track coupled vibrations

The soft under baseplate pad of WJ-8 rail fastener frequently used in China’s high-speed railways was taken as the study subject, and a laboratory test was performed to measure its temperature and frequency-dependent dynamic performance at 0.3?Hz and at ?60°C to 20°C with intervals of 2.5°C. Its higher frequency-dependent results at different temperatures were then further predicted based on the time–temperature superposition (TTS) and Williams–Landel–Ferry (WLF) formula. The fractional derivative Kelvin–Voigt (FDKV) model was used to represent the temperature- and frequency-dependent dynamic properties of the tested rail pad. By means of the FDKV model for rail pads and vehicle–track coupled dynamic theory, high-speed vehicle–track coupled vibrations due to temperature- and frequency-dependent dynamic properties of rail pads was investigated. Finally, further combining with the measured frequency-dependent dynamic performance of vehicle’s rubber primary suspension, the high-speed vehicle–track coupled vibration responses were discussed. It is found that the storage stiffness and loss factor of the tested rail pad are sensitive to low temperatures or high frequencies. The proposed FDKV model for the frequency-dependent storage stiffness and loss factors of the tested rail pad can basically meet the fitting precision, especially at ordinary temperatures. The numerical simulation results indicate that the vertical vibration levels of high-speed vehicle–track coupled systems calculated with the FDKV model for rail pads in time domain are higher than those calculated with the ordinary Kelvin–Voigt (KV) model for rail pads. Additionally, the temperature- and frequency-dependent dynamic properties of the tested rail pads would alter the vertical vibration acceleration levels (VALs) of the car body and bogie in 1/3 octave frequencies above 31.5?Hz, especially enlarge the vertical VALs of the wheel set and rail in 1/3 octave frequencies of 31.5–100?Hz and above 315?Hz, which are the dominant frequencies of ground vibration acceleration and rolling noise (or bridge noise) caused by high-speed railways respectively. Since the fractional derivative value of the adopted rubber primary suspension, unlike the tested rail pad, is very close to 1, its frequency-dependent dynamic performance has little effect on high-speed vehicle–track coupled vibration responses.     

A method to determine the two-point contact zone and transfer of wheel–rail forces in a turnout

A practical method to determine the zone of two contact points and the transfer of wheel–rail forces between two rails in a turnout is presented in this paper. The method is based on a wheel–rail elastic penetration assumption and used to study a turnout system for a 200 km/h high-speed railway in China. Rail profiles in a number of key sections in the turnout are identified first, and profiles in other sections are then obtained by interpolation between key sections. The track is modelled as flexible with rails and sleepers represented by beams and the interaction between the vehicle and turnout is simulated for cases of the vehicle passing the turnout. Results are mainly presented for two-point contact positions and the characteristics of the wheel–rail forces transference. It is found that the heights of the switch and crossing rail top have significant effects on the wheel–rail contact forces. Finally, the optimised top height for the crossing rails is proposed to reduce the system dynamic force in the turnout system.     

Influence of the track quality and of the properties of the wheel–rail rolling contact on vehicle dynamics

This work describes an analytical approach to determine what degree of accuracy is required in the definition of the rail vehicle models used for dynamic simulations. This way it would be possible to know in advance how the results of simulations may be altered due to the existence of errors in the creation of rolling stock models, whilst also identifying their critical parameters. This would make it possible to maximise the time available to enhance dynamic analysis and focus efforts on factors that are strictly necessary. In particular, the parameters related both to the track quality and to the rolling contact were considered in this study. With this aim, a sensitivity analysis was performed to assess their influence on the vehicle dynamic behaviour. To do this, 72 dynamic simulations were performed modifying, one at a time, the track quality, the wheel–rail friction coefficient and the equivalent conicity of both new and worn wheels. Three values were assigned to each parameter, and two wear states were considered for each type of wheel, one for new wheels and another one for reprofiled wheels. After processing the results of these simulations, it was concluded that all the parameters considered show very high influence, though the friction coefficient shows the highest influence. Therefore, it is recommended to undertake any future simulation job with measured track geometry and track irregularities, measured wheel profiles and normative values of the wheel–rail friction coefficient.     

Detection and Classification of Vehicles by Measurement of Road-Pavement Vibration and by Means of Supervised Machine Learning

Road vehicle detection and, to a lesser extent, classification have received considerable attention, in particular for the purpose of traffic monitoring by transportation authorities. A multitude of sensors and systems have been developed to assist people in traffic monitoring. Camera-based systems have enjoyed wide adoption over the last decade, partially substituting for more traditional techniques. Methods based on road-pavement vibration are not as common as camera-based systems. However, vibration sensors may be of interest when sensors must be out of sight and insensitive to environmental conditions, such as fog. We present and discuss our work on detection and classification of vehicles by measurement of road-pavement vibration and by means of supervised machine learning. We describe the entire processing chain from sensor data acquisition to vehicle classification and discuss our results for the task of vehicle detection and the task of vehicle classification separately. Using data for a single vibration sensor, our results show a performance ranging between 94% and near 100% for the detection task (1340 samples) and between 43% and 86% for the classification task (experiment specific, between 454 and 1243 samples).