A new urban earthquake disaster simulation method: City-scale nonlinear time-history analysis

Earthquakes can cause severe damage and huge economic loss to modern cities. Accurate and efficient seismic damage simulations of buildings have therefore become an indispensable part of earthquake hazard mitigation effort worldwide. Rational simulation results can help to illustrate potential seismic damage and can further provide an essential reference for city planning and post-earthquake rescue work. Motivated by the above, a new urban earthquake disaster simulation method, i.e., city-scale nonlinear time-history analysis (THA), is performed in this research work. The key scientific outcomes achieved are addressed below: (1) Physics-based multi-scale modeling method for urban buildings. Urban buildings can be classified into three types: Ordinary multi-story buildings, ordinary tall buildings, and special buildings. Corresponding analysis methods are proposed: Multi-story buildings can be simulated using nonlinear multiple-degree-of-freedom (MDOF) shear model; tall buildings can be simulated using nonlinear MDOF flexural-shear model; special buildings, such as large-span spatial structures and super tall buildings, can be established using the refined finite element models based on the fiber beam-column and multi-layered shell elements. (2) CPU/GPU collaborative parallel computing method. Seismic damage simulations often result in excessive computational workload. Therefore, a CPU/GPU collaborative parallel computing method is proposed: CPU will assign tasks to GPU; GPU will perform the nonlinear THA of each building; CPU will then copy result data from GPU for post-processing. Case study results show that, the proposed method can achieve a speed up of 39, compared with traditional CPU computing method with similar hardware cost. (3) Numerical coupling scheme for site-city interaction simulations. Interactions among densely distributed buildings in a city and the site, i.e., the site-city interaction (SCI) effects, can significantly influence the seismic behavior of buildings. Therefore, a numerical coupling scheme is proposed: Nonlinear MDOF models are used to represent different buildings above the ground, while an open-source spectral element program, SPEED, is used to simulate wave propagation in soil layers. The proposed numerical scheme is validated and proved useful for investigating the SCI effects. (4) High-fidelity visualization method using 3D urban polygon models and oblique aerial photography. Seismic simulation of urban buildings often produces a large number of computational results, which would be difficult to comprehend without appropriate visualizations. Thus, a high-fidelity visualization method using 3D urban polygon model is proposed, which covers three major components: Building identification, floor plan generation and displacement interpolation. In addition, to achieve a more realistic seismic damage scenario, a photo-realistic visualization method based on oblique aerial photography is proposed, including lightweight modeling, building segmentation, and dynamic visualization. Using these methods, high-fidelity visualization of seismic simulation results can be achieved. (5) Seismic loss prediction and secondary disaster simulation methods for urban buildings. Based on FEMA P-58 (next generation seismic performance assessment method of buildings) and the city-scale nonlinear THA, a practical approach for refined regional seismic loss prediction is proposed. In addition, the secondary disaster simulation for falling debris and fire following earthquake are proposed. Such a simulation will provide a useful reference for seismic loss assessment, site selection of emergency shelters, evacuation plan design, urban fire safety planning, and post-disaster emergency planning. The proposed disaster simulation method has been adopted by the SimCenter supported by the National Science Foundation of the United States to perform a seismic damage simulation and loss prediction for 1.8 million buildings in the San Francisco Bay Area. The proposed method has also been successfully applied to the seismic damage prediction of other cities, such as Beijing and Tangshan. Combined with the real-time recorded ground motions, the proposed method has been successfully used for the prompt assessment of seismic damage in 32 domestic and 42 international earthquakes. The assessment results (e.g., Jiuzhaigou Earthquake) agreed very well with the actual building damage. In summary, the city-scale nonlinear THA takes accurate consideration of the characteristics of ground motions and those of buildings, and the analysis outcomes will provide an important reference for city planning, post-earthquake rescue work, seismic damage assessment, and earthquake scenario simulation.