A Systematic Analysis of Travel Time Randomization Used to Account for Vs Uncertainty in 1D Ground Response Analyses

A new statistical framework has been recently proposed to account for shear wave velocity (V-s) variability in seismic ground response analyses (GRA). This new framework operates on the cumulative shear wave travel time (tt(s)) instead of directly on, V-s which has been hypothesized to allow for better management of subsurface variability. However, this framework has only been implemented in a few case studies, and more validations are needed. In this study, we perform a systematic analysis to assess the influence of two input parameters on tt(s) randomization, and consequently, on GRA predictions. Particularly, we assess the influence of material property variability as quantified using the lognormal standard deviation on tt(s) (sigma(lntts)) and the interlayer correlation (rho). The assessments are conducted at two downhole array sites so that the GRA predictions can be compared to actual ground motion observations. The results show that the recommended sigma(lntts) values of 0.01-0.02 might underestimate variability, but nevertheless, using values greater than 0.05 might yield unrealistic V-s profiles with excessively high or even negative, V-s which can be mitigated by using higher rho.