A new model for the prediction of track sound radiation

Abstract

The TWINS model is a widely used and well-established model for rolling noise which has been validated against field measurements in terms of overall noise spectra and levels. However, there are still some areas that can be improved. In particular, the radiation from the rail is based on a model of a rail in free space and there are also limitations in the model for the sound radiation from the sleepers. This paper draws on recent research into the effects of the proximity of the rail and sleeper to an absorbing ground on their sound radiation. Moreover, the ballast is acoustically absorbing to some extent because of the gaps between the ballast particles and this can affect the noise radiation by the rail and sleeper. The ballast absorption is represented here using the Johnson-Allard model with measured values of flow resistivity and porosity. Additionally, the Delany and Bazley model is introduced with a higher value of flow resistivity for comparison. These are used to produce the normal impedance of the ballast layer which is introduced in boundary element calculations of the radiation from the rail and sleepers. Comparisons are made first with the sound radiation from a 1:5 scale track model which has been measured reciprocally in the reverberation chamber. It is shown that the impedance using the measured flow resistivity is inadequate for use in the numerical models; probably an extended reaction model would be more appropriate. However, the Delany and Bazley model with the higher flow resistivity gives better agreement with the measurements and is a practical solution. The new models have also been used together with TWINS to predict the sound radiation from an operational track and the results have been compared with an example field measurement. The new models are found to give an improvement at low frequencies, where the sleeper is the dominant noise source.