An investigation into inconsistencies between theoretical predictions and microphone array measurements of railway rolling noise

Theoretical models, such as TWINS, and microphone array measurements have been widely used to gain better understanding of rolling noise. However, the array measurements are often inconsistent with the TWINS predictions and give less prominence to the rail than the theoretical models. The objectives of this thesis are to make validation work of the TWINS model for Japanese railway lines, and to explore the reason why the microphone array gives a correct estimate of sound power radiated by the rail. A comparison in terms of noise and rail vibration has been carried out for six wheel/rail conditions of Japanese railways. The TWINS predictions show good agreement with the measurements. After confirming the applicability of the TWINS model, the effects of wheel load on noise and rail vibration are investigated, and the predictions show similar trends to the measurements. The acoustic properties of a rail as measured with a microphone array have been investigated through simulations and field tests. In the simulation the rail is modelled as an array of multiple sources. Results are given for two situations: (a) the multiple sources are incoherent, which is assumed in determining sound power from a microphone array, (b) the sources are coherent, which is more representative of the rail radiation. It is found that the microphone array cannot detect a large part of the noise generated by the rail in the frequency range where free wave propagation occurs. Sound measurements were carried out to validate the radiation model of the rail by using a shaker excitation of a track. It is found that the noise is radiated from the rail at an angle to the normal when free wave propagation occurs in the rail, and that the predictions based on coherent sources show good agreement with the measurements. Sound measurements for a moving train were also performed with a microphone array. It is shown that the microphone array misses a large part of noise generated by the rail, when directed normal to the rail. This leads to an underestimation of the rail component of the noise in the array measurements