Accelerated Large-Scale Seismic Damage Simulation With a Bimodal Sampling Approach

Regional damage simulation is a promising method to prepare organizations for the unforeseeable impact of probable seismic natural hazard. Nonlinear time history analysis (NLTHA) finite element models (FEM) buildings in region can provide resembling results actual buildings’ damages and responses. This approach requires large-scale computational resources, improve efficiency, parallel processing representing building FEM with lumped mass are proposed. However, computing complexity still far-reaching when high-performance not available. The inventory consists numerous similar limited number distinct structures. In this paper, we propose data-driven that runs NLTHA structures exclusively infers responses other using surrogate model. Considering skewed distribution region, novel informative sample selection proposed designed bimodal sampling input domain. We use Gaussian process regression as model compare performance different methods. able approximate regional regarding total economic loss estimation 98.99% accuracy while reducing demand about 1/7th time.