Modelling the sound insulation of mass timber floors using the finite transfer matrix method

Cross Laminated Timber (CLT) technology has revolutionized the use of timber in construction in just 20 years. However, the development of mid- and high-rise CLT buildings has raised concerns about the sound insulation provided by these structural elements and about the reliability of simulation tools and models which are currently used. The mechanical characteristics of mass timber elements do not allow simplifications such as infinite out-of-plane stiffness, diffuseness of the vibrational field and perfectly elastic behavior upon impact, to mention a few, which are commonly assumed for traditional structures. The availability of modelling/simulation tools (and the relative input data) that provide accurate predictions of airborne and structure-borne sound insulation is therefore a current and relevant topic. This work presents an investigation into the input parameters to use for modelling CLT elements using the Finite Transfer Matrix Method (FTMM). The results of laboratory measurements on two CLT floors are compared to results obtained using two FTMM-based software packages in a three-step procedure. First, measurements were performed on two timber floor solutions. Following this, two operators working with different FTMM-based software packages performed blind simulations, based upon the information shared on the materials' characteristics. Finally, the input data were modified in order to return the best fit to the experimental data. The aim of the work is twofold: (1) to verify the degree of accuracy of the software and (2) following a reverse-engineering process, to retrieve the properties of the materials that need to be modelled through equivalent physical dimensions. The results for the bare CLT floor show that using dynamic E value for the plate modelling returns slightly more accurate results. Conversely, the question of modelling of a complete floor, including a floating floor, deserves greater attention, as modelling the resilient underlay using static values of dynamic stiffness can alter the results to a great extent