A hybrid computational method for computing urban canyon sound fields exhibiting roughness

Abstract

Predicting the three-dimensional sound field within an urban canyon is essential for urban noise assessment. Several analytical and computational methods exist to predict the canyon sound field. Generally building facades can be characterized as both geometrically reflecting and dispersive. The image source method is well suited to scenarios where surfaces are geometrically reflecting. Given dispersive surfaces the radiosity method predicts the sound field well. On the other hand a canyon exhibiting geometric and dispersive facades requires a computationally intensive technique such as the finite-element method (FEM) or the boundary-element method (BEM). Due to computational limitations neither the FEM nor BEM are well suited for computing large, unbounded, three-dimensional, urban canyon domains. A prediction method which synthesizes adaptive beam tracing and a secondary model for edge diffraction serves as an alternative technique for analyzing urban canyon acoustics, which contain façades exhibiting both geometric and dispersive surfaces. Advantages of the hybrid method include the ability to model unbounded domains simply, no requirement for discretizing geometric boundaries, and the ability to model sound propagation in the time domain. The magnitude of acoustic dispersion from a surface is related to surface roughness. The effect of surface roughness on the canyon sound field is investigated.