Modelling the 3D brittle response of masonry buildings to tunnelling

Accurate simulation of the building response to soil settlements is useful for the risk assessment of underground excavations in urban areas. Two-dimensional numerical models with coupling between the soil and structure offer a relatively flexible means of investigation for large scale projects, as complete 3D simulation is time consuming. However, the material parameters of the 2D building models need to be adjusted to reproduce the correct building stiffness relative to the soil. Furthermore, 2D models neglect potential torsional effects,which can be relevant during the tunnel advance and due to building orientation with respect to the tunnel. In this work, the responses of 2D and 3D models are compared to evaluate the effect of the dimensional assumption on the accuracy of the prediction of cracking in masonry structures. First, a 2D finite element simulation, which includes both the soil and the structure, is compared to the results of a centrifuge test on a scaled 3D structure subjected to tunnelling in sand. The assumptions employed to calculate the stiffness of the "equivalent" 2D model are highlighted and discussed. Second, a semi-coupled model, where the soil is not modelled directly but the soil displacements are directly applied to an interface which accounts for the soilstructure interaction, is presented. Finally, a 3D semi-coupled model of the structure is used to investigate the 3D boundary effect in the centrifuge sand box. In all simulations, a cracking model is used to account for the structural damage experimentally observed. Comparison of the 3D model and experimental results demonstrates the ability of the model to reproduce the torsional building response.The modelling of cracking in the masonry makes it possible to assess the development of structural damage and its influence on the building response. The comparison between the 2D and 3D results enables quantification of the impact of 2D simplification on the final prediction.