A comparison of numerical approaches to quantity sound insulation of lightweight wooden floor structures

Abstract

Quantifying air-borne and structure-borne sound insulation is an important design consideration for the indoor comfort of a building. Although sound insulation performance is commonly measured experimentally, numerical methods can have time-saving and economic benefits. Further, numerical methods can be incorporated within building simulations to provide an estimate of the acoustic environment. In response, this paper evaluates three different computational approaches for quantifying sound insulation in one-third octave bands (50-5000 Hz) of a lightweight floor including: an artificial neural network (ANN) model, an analytical (theoretical) model, and a finite element model (FEM). The three numerical methods are tested on the sound insulation of a cross laminated timber floor. The results of this study show that there are advantages for using each approach. The ANN model is able to accurately predict the sound insulation performance at high frequencies, but over-predicts the performance at low frequencies. Inversely, the analytical and FEM strategies provide closer estimates of low frequency sound insulation performance but overpredict the performance at high frequencies. While no model is able to accurately represent acoustic behavior across all frequencies, this work provides numerical approaches to quantify sound insulation performance. Keywords: sound insulation, artificial neural networks, building acoustics, numerical analysis, floor structures