Best estimate seismic analysis are generally based on time-domain simulations of structural responses. The seismic load is then modeled by a stochastic process representing ground motion. For this purpose, the analyst can use recorded accelerograms or work with synthetically generated ones. When dealing with structures located in very weakly seismic regions, then synthetic accelerograms are often used since no or nor enough strong motion data is available. It is then necessary to have at our disposal methods that allow for generating synthetic accelerograms that realistically characterise earthquake ground motions while respecting requirements of seismic codes and regulations. However, most of the methods for proposed in literature for generating synthetic accelerograms do not accurately reproduce the natural variability of ground motion parameters (such as PGA, CAV, Arias intensity) observed for recorded time histories. In this paper, we introduce a new method for generating synthetic ground motion, based on Karhunen-Loeve decomposition and a non-Gaussian stochastic model. The proposed method enables the structural analysists to simulate ground motion time histories featuring all the required properties mentioned above. In order to demonstrate its capability, we study the influence of the simulation method on the different ground motion parameters and on soil response spectra. We finally compute fragility curves in order to illustrate the advantages of the proposed method.
|Title of host publication||11th International Conference on Applications of Statistics and Probability in Civil Engineering ICASP11|
|Publication status||Published - 2011|
|Event||11th International Conference on Applications of Statistics and Probability in Civil Engineering ICASP11 - Zurich, Switzerland|
Duration: 1 Jan 2011 → …
|Conference||11th International Conference on Applications of Statistics and Probability in Civil Engineering ICASP11|
|Period||1/01/11 → …|
Zentner, I., Poirion, F., & Cacciola, P. (2011). Simulation of seismic ground motion time histories from data using a non-Gaussian stochastic model. In 11th International Conference on Applications of Statistics and Probability in Civil Engineering ICASP11