Stochastic characterization of strong ground motion and applications to structural response

This study addresses the problem of characterizing strong ground motion for the purpose of computing the dynamic response of structures to earthquakes. A new probabilistic ground motion model is proposed which can act as an interface between ground motion prediction studies and structural response studies. The model is capable of capturing, with at most nine parameters, all those features of the ground acceleration history which have an important influence on the dynamic response of linear and nonlinear structures, including the amplitude and frequency content nonstationarities of the shaking. Using a Bayesian probabilistic framework, a simple and effective statistical method is developed for extracting the "optimal" model from an actual accelerogram. The proposed ground motion model can be efficiently applied in simulations as well as analytical response and reliability studies of linear and inelastic structures. The random response of linear and nonlinear oscillators subjected to the proposed stochastic excitation is considered. The nonlinearity of the oscillator is accounted for by equivalent linearization. A formulation is developed which approximates the original lengthy expressions for the second-moment statistics of the transient response by much simpler expressions. The results provide insight into the characteristics of the nonstationary response and the effect of the ground motion nonstationarities. It is found that the temporal nonstationarity in the frequency content of the ground motion significantly influences the response of both linear and nonlinear structural models. Simulations are also used to study the sensitivity of inelastic structural response parameters to the details of the ground motion which are left "random" by the model. The results can also be used to provide a quantitative assessment of the expected structural damage associated with the ground motion described by the model