This paper addresses an important problem of predicting sound propagation in narrow street canyons with width less than 10 m, which are commonly found in a built-up urban district. Major noise sources are, for example, air conditioners installed on building facades and powered mechanical equipment for repair and construction work. Interference effects due to multiple reflections from building facades and ground surfaces are important contributions in these complex environments. Although the studies of sound transmission in urban areas can be traced back to as early as the 1960s, the resulting mathematical and numerical models are still unable to predict sound fields accurately in city streets. This is understandable because sound propagation in city streets involves many intriguing phenomena such as reflections and scattering at the building facades, diffusion effects due to recessions and protrusions of building surfaces, geometric spreading, and atmospheric absorption. This paper describes the development of a numerical model for the prediction of sound fields in city streets. To simplify the problem, a typical city street is represented by two parallel reflecting walls and a flat impedance ground. The numerical model is based on a simple ray theory that takes account of multiple reflections from the building facades. The sound fields due to the point source and its images are summed coherently such that mutual interference effects between contributing rays can be included in the analysis. Indoor experiments are conducted in an anechoic chamber. Experimental data are compared with theoretical predictions to establish the validity and usefulness of this simple model. Outdoor experimental measurements have also been conducted to further validate the model.

The propagation of sound in long enclosures is addressed theoretically and experimentally. In many previous studies, the image source method is frequently used. However, these early theoretical models are somewhat inadequate because the effect of multiple reflections in long enclosures is often modeled by the incoherent summation of contributions from all image sources. Ignoring the phase effect, these numerical models are unlikely to be satisfactory for use in predicting intricate patterns of interference due to contributions from each image source. In the present paper, the effect of interference is incorporated by coherently summing the contributions from the image sources. To develop a simple numerical model, the walls of long rectangular enclosures are represented by either geometrically reflecting or impedance boundaries. Measurements in a one-tenth-scale model are conducted to validate the numerical model. In some of the scale-model experiments, the enclosure walls are lined with a carpet to simulate the impedance boundary condition. It has been shown that the proposed numerical model agrees reasonably well with experimental data.

In many previous studies, energy-based methods are used to predict the attenuation of sound in long tunnels. However, these models do not address the interference effects of the sound fields generated by all image sources. A numerical model has been developed, in which the total sound field is computed by summing contributions from all image sources coherently. This numerical model also incorporates a correction term for calculating the atmospheric absorption of sound in air. To validate the numerical models in practical situations, two road traffic tunnels have been chosen for extensive measurements. The levels of the transmitted noise have been recorded in one-third octave band frequencies at various separations up to a maximum of 400 m. The predictions using the coherent model agree reasonably well with the measured data at all frequencies. The agreements between the field data and the theoretical predictions using the energy-based model are tolerable at high frequency, but less so at low frequency. In most cases, the predictions of the coherent model give the best results, with an accuracy to within 3 dB. On the other hand, the energy-based models are not able to predict the peaks and dips across the frequency spectra, the variance with the measurement results being up to 7 dB at low-frequency bands.

The sound generated by the unsteady motion of a vortex filament moving over a flat boundary with a sharp flow impedance discontinuity is studied theoretically. Theoretical results show that the vortex filament undergoes significant accelerating or decelerating motions and radiates sound at the instant when it moves across the plane of impedance discontinuity. The accelerations and decelerations of the vortex filament are shown to be the major mechanisms of sound generation. The sound so produced has a large low-frequency content such that the change in the flow impedance affects only the sound generation process but not the subsequent sound propagation to the far field.