The coupled vehicle-track system (CVTS) dynamics have been extensively investigated for decades. However, the calculation accuracy of prevailing vehicle-track coupling models needs to be improved in the high frequency range due to the inappropriate model simplification and neglect of material nonlinearity. In this study, we propose a refined numerical model of the CVTS that considers the nonlinear properties of the railpads and primary suspension using the fraction derivative Zener model. Furthermore, we more realistically simulate the wheelset, rail and railpad configuration with the elastic axle, solid finite element and surface-support models, respectively, and improve the computation efficiency by employing the mode superposition method. The results demonstrate that the refined CVTS model is more accurate than the classical model in simulating vehicle-track coupling dynamics above 2 kHz. In particular, there are significant differences in the dynamic response of the elastic wheelset model compared to the rigid model over a broad frequency range, with an 11% difference in the bogie acceleration response at the first dominant frequency. When the railpads are modeled using the surface-support model, the rail acceleration differences exceed 41% near 1 kHz and 44% near 2650 Hz, compared to the point-support model. Additionally, the rail response at various locations across the rail cross section can be calculated using the finite element method in this refined model. Overall. the proposed CVTS model provides high accuracy and efficiency for random vibration analysis, especially in the high frequency domain.

To investigate the train-induced ground vibration, an explicit time-domain, three dimensional (3D) finite element (FE) model is developed. The train, track, embankment, pile-board structure and nearby ground soils are all fully coupled in this model. The complex geometries involving the track components and pile-board structure are all modelled in detail, which makes the simulation of wave propagation more realistic from the train to the ground. The model is validated with in situ tests data collected in the Beijing-Shanghai high speed railway line. Good agreements have been achieved between the numerical results and experimental results both in time domain and frequency domain. The proposed model is thus capable of reproducing the dynamic ground response induced by a typical high speed train. Soil responses induced by different number of vehicles are compared. With more vehicles, the spectral peaks of soil responses are more prominent at the integral multiples of the vehicle passing frequency. Too few vehicles will not bring about such phenomenon, thus sufficient number of vehicles should be included in a train to properly model train-induced ground vibration. With the proposed model, the influence of the pile-board foundation on the ground vibration is investigated. It is found that the pile-board foundation can significantly attenuate the low frequency ground vibration. The attenuation of the ground vibration as a function of distance from the track is simulated and the influential factors to the local vibration amplification are investigated. It is found that soil Young's modulus and soil impedance contrast are the two main factors influential to the local vibration amplification. The softer the natural soil, the larger the amplification. The larger soil impedance contrast makes the amplification more obvious. The soil stratification and geometric discontinuity at ground surface are not the main cause of the local vibration amplification in this work.