Prediction models and computer simulations have become indispensable tools for designing and Consulting in acoustics. The reliability of thè results obtained from these tools is dependent on different causes. Low frequencies, for example, proved to be less accurate both in thè spread sheet and computer models.

The main subject of this dissertation is to study five small music halls that Arup has completed in the last few years and to compare their measured performances at low frequencies with different prediction methods. This is owing to thè fact that at low frequencies all thè investigated concert halls show consistent differences between predicted and measured results. The aim of this dissertation is to determine whether this difference is a result of prediction errors or if it is a measurement error (lab/site). Considering thè uncertainties that come from thè analysis of all thè data collected during thè researches done in Arup, this thesis develops a statistical method, Monte Carlo, to estimate thè uncertainties that can provide possible reasons for thè different results obtained between predictions and measurements.