PDEM-based stochastic seismic response analysis of sites with spatially variable soil properties

The spatial variability of soil properties has a significant effect on the seismic responses of sites. Considering that the probability density evolution method (PDEM), an efficient probabilistic methodology, has not been used in the probabilistic analysis with spatially variable soil properties before, the random field theory is incorporated to the traditional framework of PDEM to perform stochastic seismic response analysis of a soil profile. The stochastic seismic responses, including seismic deformation, surface ground motion, amplification factor and excess pore water pressure ratio, are obtained to quantify the effect of spatially variable soil properties. Probabilistic sensitivity analyses are also carried out to study the influence of spatial variability of different soil parameters and coefficients of variation on the ground motion in terms of probability density function. The results indicate that the random field model of shear modulus has a higher possibility of experiencing higher ground motion intensity than does friction angle. •The random field theory is incorporated into the traditional framework of PDEM.•The effect of spatially variable soil properties is quantified in terms of PDF and CDF.•The propagation of soil uncertainties in the dynamic system of soil profile is revealed by instantaneous PDF.•A refined probabilistic sensitivity analysis is conducted in terms of PDF and CDF.