Effect of Uncertainty in Vs with Inter- and Intralayer Variability and Nonlinear Soil Properties on Equivalent Linear GRA Using a Developed Code in the Frequency Domain

The study attempts to assess the effect of uncertainty in shear wave velocity (Vs) and nonlinear dynamic soil properties on equivalent linear ground response analysis (GRA). The Vs uncertainty is considered through the Toro (1995) model for interlayer variability and a one-dimensional (1D) random field for intralayer variability. To account for the uncertainty in nonlinear dynamic soil properties, randomized modulus reduction (G/Gmax) and damping curves (ξ) are generated using a previously developed model. Then a Monte Carlo type simulation is conducted to examine the effect of both types of uncertainties on GRA. A code in the MATLAB version 7.5 (R2007b) environment is developed for the entire analyses. The study considers four different levels of Vs uncertainties, i.e., σ(lnVs)=0.10, 0.15, 0.20, and 0.30. To examine the impact of shaking intensity, Loma Prieta input motion is scaled to four different peak ground acceleration (PGA) values: 0.05g, 0.10g, 0.20g, and 0.40g. As the σ(lnVs) increases, the surface mean peak transfer function (TF), spectral acceleration (SA), and PGA tend to be underestimated compared to the baseline responses, with the scenario that combines both the uncertainties, i.e., Vs and nonlinear dynamic soil properties, exhibiting the most pronounced underestimation. The intensity of input motions also significantly affects the GRA outcomes. The surface mean peak SA is found to be overestimated than the baseline responses at a lower intensity of shaking, whereas, a higher intensity of shaking (i.e., 0.40g PGA) leads to underestimation of the same. The standard deviations of surface TF, SA, and PGA increase with both σ(lnVs) and intensity of shaking. The maximum standard deviations are observed in the third scenario, in which uncertainties in Vs and the G/Gmax and ξ curves are combined.