Simulating rolling noise on ballasted and slab tracks: vibration, radiation, and pass-by signals

Sammanfattning: Shifting to rail-bound freight and passenger traffic is key in Europe's strategy towards transport decarbonisation. However, increasing railway traffic can increase environmental noise pollution. Rolling noise is often the dominant noise source. It originates from the interaction of the rough running surfaces of wheel and rail. Predicting rolling noise and performing acoustic optimisation of existing and new tracks requires validated, flexible, and physics-based prediction tools. This is especially relevant for the different designs of ballastless tracks, which are increasingly used for high-speed lines. Therefore, this thesis aims to develop and implement a modelling approach for rolling noise in the time and frequency domain to increase understanding of sound radiation, investigate noise mitigation measures, and allow research of the perception of transients in rolling noise. To achieve this, models for vibration in wheels and several types of ballasted and slab tracks have been implemented using the Waveguide Finite Element method. This method allows an efficient prediction of the track vibration up to high frequencies. Next, models for the sound radiation from wheel and track were implemented using adaptions of the Boundary Element method (BEM), such as the Fourier series BEM and the wavenumber domain BEM. The computational efficiency was addressed in multiple ways. Finally, an approach to simulate the sound produced at a stationary track-side receiver has been developed and implemented based on moving Green's functions. The simulations were largely implemented in in-house Python code. The ballasted and slab track dynamic models have been tuned and compared with measurements on full-scale tracks. The developed models have been used to analyse the vibrations in track and wheel and the acoustic radiation from these vibrations. This allowed the investigation of noise mitigation measures. Further, the necessary complexity of the dynamic track model for predicting rolling noise in time domain was investigated. Two parameter studies were carried out with a focus on track design with lower noise emission. Slab tracks with booted sleepers showed a potential for noise reduction without increasing loads on the track structure. A continuous rail support lowered the radiated sound power at high frequencies. The contributions of different wheel modes to the radiated sound were investigated considering the directivity of each mode, and dominant modes were identified. The established models produce time signals usable for auralisation, which, among others, has the potential to research human perception of transients in rolling noise.