**Wooden Houses Collapse Simulation during Earthquake and Tsunami at a City Level base on the ASI-Gauss Finite Element Method**

Abstract

In the 2011 Tohoku earthquake, many wooden houses in coastal area collapsed by tsunami and became debris after the fractures. The total amount of debris during tsunami was estimated about 22 million tons. These debris increased power of tsunami and it induced more human and economic losses. The flooding with debris gave a huge delay behind recovery from this tsunami disaster. It is, therefore, necessary to evaluate not only the tsunami run-up phenomenon but also the total amount of possible debris and its diffusion and distribution. In this study, we aim at a numerical estimation of the total amount of possible debris and the diffusion and distribution given by tsunami run-up. As preliminary step, we examined a method of modeling wooden houses, which are likely to collapse by tsunami, at a city scale. The number of wooden houses targeted at a city level may reach about a half million. We should evaluate the collapse of a huge number of these houses at the same time as much as accurately. The modelling for collapse of houses is, therefore, simplified with low computational costs. Then, the ASI-Gauss code, which is a branch of finite element method and can analyze progressive collapse phenomenon of frame and beam structures, is selected as a simulator for each house. Wooden houses have been modelled with beam element. The main advantage of the ASI-Gauss code is its robustness in progressive collapse simulations, since it can reasonably simulate these framed structure collapse phenomena including a material non-linearity such as elasto-plastic behavior and a geometrical non-linearity for large deformation. In addition, the original shifted integration technique reduces the computational cost by using only two linear beam elements for each framed member. The fracture conditions for the framed member are considered by examining the curvatures, axial tensile strains, and shear strains in the element. Once a rupture of member is judged in the element, it is expressed by reducing the sectional forces of the element immediately after the occurrence of the fractured section. The whole city model for the ASI-Gauss code is automatically generated from a Geographic Information System data; the SHP formatted data. The beam element model for each houses is generated from the outline of houses, building age and its floor levels defined in the SHP formatted data. The huge finite element models at a city level are parallel solved with a multi-node PC cluster and supercomputer in this study.