Aerodynamic noise simulation of high-speed train bogie

Abstract

Aerodynamic noise from high-speed trains becomes more and more important as the train speed increases. Of the various sources, the train bogies contribute significantly to the overall aerodynamic noise, especially the leading bogie. This research aims to reveal the aerodynamic noise generation mechanisms from bogies of high-speed trains and propose suitable noise reduction measures. Numerical simulation has been a great challenge for the simulation of the aerodynamic noise of bogies. This is due to the complex geometry, which makes the discretization very difficult and, meanwhile, the grid will be very large. To overcome these challenges, a hybrid grid system is explored, which can guarantee a high-quality grid in the boundary layer while maintaining the overall number of cells in the grid at an acceptable level; the model size and flow speed are both scaled down to further reduce the number of the grid cells. The Delayed Detached Eddy Simulation is used to investigate the flow and obtain the noise source information to feed into the Ffowcs Williams-Hawkings acoustic analogy for far-field noise prediction. The hybrid grid system and the numerical methods are validated by simulations for a circular cylinder, square cylinder and an isolated wheelset. After that, the hybrid grid system is applied and further developed for the simulations of a bogie in a simplified cavity. The results show that the rear part of the bogie and cavity have strong pressure fluctuations and the noise generated by the cavity is much greater than that by the bogie. A more complex model of a bogie under a leading car is then investigated. It is found that the bottom of the cowcatcher and the bogie, the cavity rear surface, and the side dampers, which are directly flapped by the highly turbulent wake and detached shear layer, form strong pressure fluctuations. The far-field noise levels and the sound power levels emitted by the cavity are greater than that of the bogie. The effect on the aerodynamics and aeroacoustics of the lateral position of the bogie’s side components relative to the car body is investigated. The flow field results show that the protruded side components shield the detached shear layer from upstream, preventing it from impinging on the rear part of the cavity. The pressure fluctuation on the side components increases, while it reduces at the rear surface of the cavity, as a result of which only a small difference in the sound power is found between the various cases. Based on the analysis of the flow field and pressure fluctuation distribution of the simulated cases, a noise control technique based on a dual staggered jet is developed to reduce the noise level of the leading car. The wake at the bottom and the detached shear layer at the two sides of the cavity is pushed away by the jets, which reduces the pressure fluctuations on the bogie and the cavity. A reduction of 2 dB for the sound pressure levels and 3.5 dB for the sound power levels is obtained. Finally, to reduce further the computational cost of the models, a novel decomposition method for CFD simulation is developed. A model of tandem square cylinders is adopted to validate the method showing good agreement between the decomposed model and the complete model. The computational time of the decomposed model is 29.6% less than that of the complete model. The collected inflow data is compressed by a convolutional Variational AutoEncoder neural network and the compression ratio achieves 63.5. The decomposition method then is applied to the simulation of a half-width leading car model.

More information

Identifiers

ORCID for Yuan He: orcid.org/0000-0003-3768-519X

ORCID for David Thompson: orcid.org/0000-0002-7964-5906

Catalogue record

Export record