Utilisation of hardwood in cross-laminated timber - A numerical study on vibrations in floors

Abstract: Cross-laminated timber (CLT) is an engineered wood product consisting of several layers of laminations bonded together with adhesives. In contrast to heavier construction materials, timber and other lightweight building materials are more prone to serviceability issues due to vibrations. In Sweden, the raw material used for CLT is mainly pine and spruce, which are softwoods. However, different wood species are available, such as birch and beech, which are the most common hardwood species in Sweden and Europe, respectively. In addition, previous research indicates potential advantages with utilisation of hardwood in CLT with regard to vibroacoustic performance. Thus, the present work aims to further examine if benefits with regard to vibrations in CLT floor panels can be obtained by utilising hardwood. Initially, three different numerical models were created to determine a suitable modelling approach for CLT. A high-fidelity model where each lamination was modelled individually, a layered model where each layer was modelled individually, and a twodimensional composite model were created. In addition, experimental testing was carried out on a real CLT panel to obtain data which the numerical models could be compared to. The layered model matched the experimental results the best and was thus considered for further validation. The material parameters of the layered model were calibrated with Newton optimisation, where the most influential parameters of the model were altered. Several different calibrations were performed, where the most accurate calibration with regard to the normalised relative frequency difference (NRFD) yielded an average NRFD of less than one percent considering the first nine modes. Consequently, this modelling approach was considered valid. A case study was conducted to determine the effect utilisation of different wood species has on the dynamic behaviour of CLT floor panels. Three differently sized panels with different lay-ups were tested, where all three panels fulfil the current design criteria to ensure practical relevance. The dynamic behaviour was evaluated by the means of modal properties, frequency response functions and acceleration response from footfall loading. The analysis on the modal properties showed that changing material from spruce to birch and beech had little effect on the modal order for all tested panels. The material change did however affect the natural frequencies of the tested floor panels to various degrees, and the fundamental frequency was generally lowered. The effect on the frequency response functions showed that the natural frequencies were altered, and that the fundamental frequency was generally lowered when utilising birch and beech instead of spruce. In addition, the magnitude of the accelerations was observed to be lowered when utilising birch and beech, where the larger panels were affected slightly more by material change than the smaller panels. The acceleration response from footfalls was analysed by simulating five consecutive steps walking straight across the length of the CLT panels. For each panel and for each wood species, ten different walking frequencies were tested. The walking frequency was shown to have a large impact on the acceleration response of the panels. By examining the frequency content of the acceleration response with a fast Fourier transformation, it was observed that the response was amplified if a harmonic of the walking frequency matched the fundamental frequency of the panel. In addition, up to the ninth harmonic of the walking frequency amplifies the acceleration response to a large extent. Consequently, design methods considering only a single walking frequency might not yield the largest acceleration response. Utilising hardwood yielded a significant reduction in acceleration response due to footfalls. The RMS values of the acceleration responses was reduced by up to 70% when using birch and beech instead of spruce. In addition, the reduction was more significant for the larger panels.