GENERATION OF ARTIFICIAL ACCELEROGRAMS USING STOCHASTIC METHODS

Abstract eng:
In recent years, the stochastic description of ground motion time histories used as input in the nonlinear dynamic analysis of structures has received considerable attention in the field of earthquake engineering. This is mainly due to the random nature of earthquake ground motion that may be experienced by a structure in the future as well as to the lack of recorded data fitting the specific requirements of each case (site conditions, epicentre distance etc.) In addition, a complete probabilistic description of the structural response in the framework of Monte Carlo simulation requires a large number of accelerograms with similar statistical properties. Since there are still not enough records of strong earthquakes with the above properties, this task can only be achieved through the generation of artificial accelerograms. In order to be realistic, these accelerograms have to reflect the characteristics of the ground motions expected to happen at a given site. In this paper, the issue of generation of artificial accelerograms is treated as an application of non-stationary stochastic process simulation. The non-stationarity in both ground motion amplitude and frequency content is taken into account by using the concept of evolutionary power spectrum. The correlation of the timedependent amplitude and frequency content of the signal with magnitude, distance and site conditions is deduced from recorded strong motion data. The spectral representation method along with translation process theory is used to generate artificial accelerograms that are compatible with the elastic design spectrum defined by EC8. Finally, useful conclusions are derived regarding the influence of a non-Gaussian probability distribution on the generated ground motion time histories.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket