Curve squeal is commonly attributed to self-excited vibrations of the railway wheel, which arise due to a large lateral creepage of the wheel tyre on the top of the rail during curving. The phenomenon involves stick/slip oscillations in the wheel/rail contact and is therefore strongly dependent on the prevailing friction conditions. The mechanism causing the instability is, however, still a subject of controversial discussion. Most authors introduce the negative slope of the friction characteristic as source of the instability, while others have found that squeal can also occur in the case of constant friction due to the coupling between normal and tangential dynamics. As a contribution to this discussion, a detailed model for high-frequency wheel/rail interaction during curving is presented in this paper and evaluated in the case of constant friction. The interaction model is formulated in the time domain and includes the coupling between normal and tangential directions. Track and wheel are described as linear systems using pre-calculated impulse response functions that are derived from detailed finite element models. The non-linear, non-steady state contact model is based on an influence function method for the elastic half-space. Real measured wheel and rail profiles are used. Numerical results from the interaction model confirm that stick/slip oscillations occur also in the case of constant friction. The choice of the lateral creepage, the value of the friction coefficient and the lateral contact position on the wheel tread are seen to have a strong influence on the occurrence and amplitude of the stick/slip oscillations. The results from the interaction model are in good qualitative agreement with previously published findings on curve squeal.

Most available wheel/rail interaction models for the prediction of impact forces caused by wheel flats use a Hertzian spring as contact model and do not account for the changes in contact stiffness due to the real three-dimensional wheel flat geometry. In the literature, only little information is available on how this common simplification influences the calculation results. The aim of this paper is to study the influence of contact modelling on simulated impact forces due to wheel flats in order to determine the errors introduced by simplified approaches. For this purpose, the dynamic wheel/rail interaction is investigated with a time-domain model including a three-dimensional (3D) non-Hertzian contact model based on Kalker's variational method. The simulation results are compared with results obtained using a two-dimensional (2D) non-Hertzian contact model consisting of a Winkler bedding of independent springs or alternatively a single non-linear Hertzian contact spring. The relative displacement input to the Hertzian model is either the wheel profile deviation due to the wheel flat or the pre-calculated vertical wheel centre trajectory. Both the 2D model and the Hertzian spring with the wheel centre trajectory as input give rather similar results to the 3D model, the former having the tendency to slightly underestimate the maximum impact force and the latter to slightly overestimate. The Hertzian model with the wheel profile deviation as input can however lead to large errors in the result. Leaving aside this contact model, the correct modelling of the longitudinal geometry of the wheel flat, is actually seen to have a larger influence on the maximum impact force than the choice of contact model.

A time-domain model for the prediction of long-term growth of rail roughness (corrugation) on small radius curves is presented. Both low-frequency vehicle dynamics due to curving and high-frequency vehicle-track dynamics excited by short-wavelength rail irregularities are accounted for. The influence of non-Hertzian and non-steady effects in the wheel-rail contact model on rail wear is studied. The model features a contact detection method that accounts for wheelset yaw angle as well as surface irregularities and structural flexibilities of wheelset and rail. The development of corrugation on a small radius curve is found to be highly influenced by the wheel-rail friction coefficient. For vehicle speed 25 km/h and friction coefficient 0.3, predictions of long-term roughness growth on the low rail show decreasing magnitudes in the entire studied wavelength interval. For friction coefficient 0.6, roughness growth is found at several wavelengths. The corresponding calculation for the high rail contact of the trailing wheelset indicates no roughness growth independent of friction coefficient. The importance of accounting for the phase between the calculated wear and the present rail irregularity is demonstrated.

Rolling noise is excited by the roughness of the wheel/rail running surfaces. The contact patch acts as a filter attenuating the excitation at wavelengths that are short in comparison with its length. Additionally, the excitation depends on the variations in roughness profile height across the width of the contact. While most available wheel/rail interaction models include the contact filter effect by roughness pre-processing, a time-domain model is presented in this paper that includes the contact filter effect dynamically by an appropriate two-dimensional (2D) or three-dimensional (3D) non-Hertzian contact model. The 2D contact model is based on a Winkler bedding, while wheel and rail are locally approximated by elastic half-spaces in the 3D contact model. The wheel/rail interaction model is applied to evaluate the contact filter effect for different sets of roughness data measured in several parallel lines. It is found that the 3D contact model gives, as a general tendency, a contact force level several dB lower than the 2D model. The differences increase with a decrease in correlation between the roughness on parallel lines and vary significantly with the choice of roughness line in the 2D model.

The prediction of impact forces caused by wheel flats requires the application of time-domain models that are generally more computationally demanding than are frequency-domain models. In this paper, a fast time-domain model is presented to simulate the dynamic interaction between wheel and rail, taking into account the non-linear processes in the contact zone. Track and wheel are described as linear systems using impulse-response functions that can be precalculated. The contact zone is modelled by non-linear contact springs, allowing for loss of contact. This general model enables the calculation of the vertical contact forces generated by the small-scale roughness of rail and wheel, by parametric excitation on a discretely supported rail and by discrete irregularities of rail and wheel. Here, the model is applied to study the excitation caused by wheel flats by introducing a flat on a rotating wheel whose profile in the contact zone is updated in every time step. To demonstrate the functioning of the model, simulation results are compared to field measurements of impact forces and a brief parameter study is presented.