Railway rolling noise prediction and validation

Abstract

Railways play an increasingly important role throughout the world as they are environmentally friendly, helpful to economic development and an efficient mode of transportation. Meanwhile, they have disadvantages, one of which is railway noise. This thesis is focussed on railway rolling noise prediction. The most effective method of reducing railway noise is to control it at its source. The dominant noise sources must be identified and the parameters that influence them should be understood and predicted before control implementations are made. Therefore, there is a requirement for development of theoretical models for these purposes. Over the past few decades, much work has been performed to develop theoretical models for railway rolling noise prediction, to validate these models against full-scale running tests and to use the models to aid the design of quieter trains. One of the most effective models, the Track-Wheel Interaction Noise Software (TWINS) model, has been validated under European conditions and Japanese conditions. Based on the TWINS, a prediction model called Railway Rolling Noise Prediction Software (in short ‘RRNPS’) model was developed in this thesis. It has been shown to provide reliable predictions compared with those by the TWINS for typical European conditions. To extend the range of conditions for which the RRNPS model is validated, new measurement data of noise from suburban passenger trains and vehicle/track parameters were gathered in a typical railway rolling noise situation in Australia to simulate and validate the RRNPS model. Comparisons between simulations and measurements have shown that this software model gives reliable predictions in terms of both overall A-weighted sound pressure level and noise spectrum. The model provides the basis for predictive model-based strategies to control and mitigate railway rolling noise. It also forms the foundation from which the effect of environmental factors and friction modifiers on normal rolling noise can be modelled and quantified. The RRNPS model was developed further, integrating a 2D variable speed corrugation model, and used to predict tangent rail roughness growth and corresponding railway rolling noise growth over time. A series of field experiments were performed in a typical railway rolling noise situation to gather the relevant parameters for simulation and validation purposes. Through comparisons between simulations and measurements, it is shown that the developed RRNPS model gives reliable predictions on tangent rail roughness growth and corresponding noise growth. The simulation model can be utilized as a predictive tool to investigate new strategies to control and mitigate railway rolling noise growth due to corrugation and rail roughness growth. For the purposes of quantifying the effect of humidity, temperature and friction modifiers on normal rolling noise, a series of corresponding experiments were performed with a two disk test rig. The RRNPS model was modified for the test rig conditions and used to simulate the effects of the three factors. Comparisons between simulations and measurements indicate that the modified RRNPS model can reasonably predict the effects of these environmental factors on normal rolling noise. Although the effect of humidity, temperature and FMs is small under the conditions tested, its quantification provides useful insight into the control of railway noise under other conditions. The most significant contributions of this thesis are the development of simulation models for predicting tangent rail roughness growth and corresponding railway rolling noise growth over time and quantification of the effects of humidity, temperature and friction modifiers on normal rolling noise. Secondary to this, field validation of modelling predictions of railway rolling noise under a typical situation in Australia is provided.