KSCE JOURNAL OF CIVIL ENGINEERING, v.26, no.10, pp.4235 - 4241
Abstract
The effect of uncertainty on the shear wave velocity (V-S) profiles from the Korean soil conditions on the seismic site response analysis (SRA) was investigated by comparing the amplification factors (AFs) computed from 170 V-S profiles to those from three (representative) statistically derived V-S profiles (Approach 1: a constant standard deviation of V-S, sigma(ln(Vs)), Approach 2: a depth-dependent, sigma(ln(Vs)), and Approach 3: 5% and 95% percentiles for generating lower and upper bounds). All of the V-S profiles are divided into three site classes (B, C, and D) by the National Earthquake Hazards Reduction Program (NHERP) site classification system. Seven pairs of ground motion records (two horizontal components) that represent Korean bedrock conditions are entered into all of the three types of V-S profiles to run the SRA with a single type of modulus reduction and damping curves. The root-mean squared difference (RMSD) of the mean and standard deviation, (mu(ln(AF)), and sigma(ln(AF))) of the AFs in natural log-scale between the three types of V-S profiles is estimated, which results in Approach 3 with the smallest RMSD measure for all the site classes. Conventionally, characterizing the uncertainty of the V-S profiles is typically assessed through the representative profiles of Approach 1; however, this practice suggests that the inclusion of the percentile bounds (Approach 3) can be supplemental to better describe the V-S uncertainty while taking the Approach 1 as primary representative profiles.