Prediction of airborne sound transmission across a timber-concrete composite floor using Statistical Energy Analysis

This paper concerns the development and experimental validation of prediction models using Statistical Energy Analysis (SEA) to calculate the airborne sound insulation of a timber-concrete composite floor. The complexity in modelling this floor is due to it having (1) a multilayer upper plate formed from concrete and Oriented Strand Board (OSB), (2) multiple types of rigid connector between the upper plate and the timber joists and (3) a resiliently suspended ceiling. A six-subsystem model treats the concrete-OSB plate as a single subsystem and three different five-subsystem models treat the combination of concrete, OSB and timber joists as a single orthotropic plate subsystem. For the orthotropic plate it is suggested that bending stiffnesses predicted using the theories of Huffington and Troitsky provide a more suitable and flexible approach than that of Kimura and Inoue. All SEA models are able to predict the weighted sound reduction index to within 2 dB of the measurement. The average difference (magnitude) between measurements and predictions in one-third octave bands is up to 4 dB. These results confirm that SEA can be used to model direct transmission across relatively complex floor constructions. However, this requires the inclusion of measured data in the SEA model, namely the dynamic stiffness of the resilient isolators and the cavity reverberation time.