A modified turnout noise model and field validation

Abstract

This paper presents a modified empirical model for prediction of noise emanating from turnouts and crossovers. Existing empirical models use a generic correction factor on top of normal rolling noise, assuming turnout is also a dipole line source. Field measurements were conducted at various setbacks to validate the noise model. It is found that turnouts behave as monopole point source rather than dipole line source. The distinct noise source characteristics of turnout necessitate a different set of modelling equations. Calculation of Railway Noise (CRN) tends to underestimate the noise contribution from turnout, particularly at close distance or large angle of view, with a discrepancy of 4–7 dB(A). The modified noise turnout can address such discrepancies by adopting monopole point source assumption. Noise and vibration sensors were also placed at the track to study the relative noise contribution from rail, sleeper and wheel. The empirical relationship between noise level and various dependent parameters are derived. The noise level has higher dependence on train speed which an empirical relation of 35log(speed) rather than 20log(speed) in analytical models. However, it shows little dependence on unsprung mass, probably due to the fact that the wheelset is not a rigid body, the gear box and traction motor does not have significant addition to effective mass.