UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Optimal design of duct silencers in naturally-ventilated buildings Vasudevan Shankar, Vivek


Natural-ventilation systems in sustainable buildings rely on wind and buoyancy effects to ensure sufficient airflow though the building. To promote airflow, these buildings have openings that connect rooms and common spaces. These openings are usually made as large as possible to maximize airflow across connected spaces. Large openings result in noise ingress between these spaces, leading to poor noise isolation and speech privacy. The performance of these openings needs to be optimized -- meaning the sound isolation of these openings has to be improved, with minimal impact on the airflow performance. Previous work by Bibby investigated openings in thin partitions. This study addresses the problem of openings in thick partitions, which essentially are short ducts, as well as long ducts that deliver air-flow to different building levels. Simple surrogate models were developed from symbolic regression; they can be used as a provisional tool for predicting the sound and airflow performance of ducts with absorptive lining and baffle inserts. This tool can be used in the design phase to build optimal ducts, or for retrofit noise control. Predicted results from 2D Finite Element Analysis were used as the input data-set to build these surrogate models. Novel methods were used to extend the surrogate model into the 3D domain and to model complex duct configurations. Different configurations of sound absorbers (duct lining and baffles) were tested to identify optimal designs. The use of scale-models of ducts to study their real-life performance was also investigated. The surrogate models were validated experimentally in a 1:4-scale-model facility, by applying appropriate scaling laws, and were found to capture the general trend of the acoustics and airflow performance within acceptable accuracy. Finally guidelines for designers were proposed based on results from the surrogate models, experiments and FEA.

Item Citations and Data


Attribution-NonCommercial-NoDerivs 2.5 Canada