Efficient simulation of stochastic seismic response of long-span bridges in river valleys using hybrid BEM-FEM

Abstract

Determining ground motions is essential for seismic analysis of long-span bridges located at complex sites. However, there is a lack of earthquake records for local site conditions. Thus, this study focused on the effect of local site conditions on the stochastic response of long-span high-pier rigid frame bridges crossing a valley with an overlying water layer. For this purpose, a hybrid boundary element and finite element method (BEM-FEM) is proposed to investigate the stochastic seismic response of long-span bridges at such a complex site. First, a simulation method for 3-D spatially varying ground motions ascribed to wave scattering from irregular surface topography, sedimentary soil, and water layers was established based on the BEM combined with the spectral representation method. Furthermore, the statistical properties of the bridge response were explored based on the FEM and Monte Carlo simulations. The advantage of this work is that it reveals how the seismic wave propagation in this particular valley topography and water layer affects the stochastic response of the bridge considered. Moreover, the results indicate that seismic wave scattering leads to a significant energy redistribution of the input waves in the tri-directions related to the surface location, and a decrease in the coherency loss functions at all frequencies. Owing to this change, the peak ground accelerations (PGAs), bridge response means, and response variability are doubled. The beneficial effect of the water layer on the PGAs is weakened, while the adverse effect of hydrodynamic pressure dominates. The parameter analysis results demonstrate that a 10-m-deep sedimentary soil layer produces a 47% increase in the internal force of the pier bottom. The solid-liquid dynamic coupling effect should be accurately quantified for bridges located at saturated sites. Considering the local site conditions, the coefficient of variation (COV) of the response of the pier top exceeded 0.3, indicating strong randomness.