IMPACT OF SEISMIC HAZARD CHARACTERIZATION ON PROBABILITY OF LIQUEFACTION

摘要

The objective of this paper is to assess the sensitivity of safety factor against the onset of liquefaction to the characteristics of the seismic hazard data defined for the site. The site considered is a deep soil site, approximately 1000 feet thick, for which hazard, laboratory and field geophysical data are available. The uniform hazard spectra (UHS) defined at the rock outcrop at a given return period are used to define the characteristics of the ground motion input to the site. The spectra are defined in terms of their median values and variabilities at each frequency. Using a Monte Carlo procedure, a large number of input ground motions (500 cases) are defined at the rock outcrop. For each case, ground motion time histories are developed to fit the UHS randomly selected from the hazard data, assuming that the UHS is an appropriate definition of the seismic input. The time duration of the ground motion is selected to be consistent with the mean magnitude associated with the hazard at the return period considered. For each soil column analysis, peak induced shear stresses are then calculated throughout the soil overburden using the concept of upward propagating shear waves through the one dimensional soil column. Shear modulus and hysteretic damping characteristics used in these calculations are consistent with the experimental data available for the site. Corresponding soil shear strengths are estimated from Standard Penetration Test (SPT) blow count data determined from the site field investigation program. The Seed empirical correlations are used to estimate the soil shear strength or capacity needed to define the onset of liquefaction. Since the SPT blow counts are found to be essentially independent of shear wave velocity measurements at the site, blow counts for strength estimates are randomly selected from the available data for each soil layer to estimate shear strength. The median values of safety factors and their variability are then found for each layer by analyzing the computed safety factors from each of the convolution calculations.