A time-domain model for railway rolling noise

Conference poster, 2023

The poster presents a modelling approach for railway rolling noise prediction developed at Chalmers during a recent PhD project. Rolling noise, which is caused by the roughness-excited vibration of the wheel and the track, is the dominant noise source in a wide range of vehicle speeds. The presented modelling approach is based on the time-domain, non-linear contact model WERAN. The model has been extended with a numerically efficient description of the structural vibration of the wheel and the track based on moving Green's functions. Further, efficient models for the sound radiation from the wheel and track were developed and implemented, again using a Green's functions approach. The Green's functions are computed using combinations of the Waveguide Finite Element method (2.5D FE), the Wavenumber domain Boundary Element Method (WBEM / 2.5D BE), the Fourier domain BEM (FBEM), and spherical harmonics equivalent sources. This model provides a physics-based, time-domain description of the radiated sound based on the combined roughness between the wheel and the rail. There are several possible applications for a time-domain rolling noise model, for example in component design, condition monitoring, and, by auralising the noise, as an effective tool for communication with a broader public.