Practical considerations in liquefaction numerical simulations calibrated with historical data

Abstract

The paper details some practical considerations associated with the numerical simulation of liquefaction in Wildlife site in Southern California. Two material constitutive models are implemented in the simulations: a pressure-dependent multi-yield-surface model (PDMY) and PM4Sand, both available in the OpenSees finite elements platform. Both uniaxial as well as biaxial simulations are presented in the paper. The uniaxial simulations only include the predominant horizontal shaking component while the biaxial simulations include both orthogonal horizontal shaking components. Two historical major earthquake events were simulated: the 1987 Superstition Hills and the 2010 El Mayor Cucapah earthquakes. Laboratory experimental data used for calibration of the material models was obtained from historical data published in the 1980's. This data is particularly valuable since they correspond to intact (undisturbed) samples extracted from Wildlife site about two years before occurrence of the 1987 earthquake. In all the simulations, the models were able to capture salient features of the deposit's behavior, such as the magnitude of surface accelerations, and dilative behavior of the soil. However, the excess pore water pressure rises earlier than what the site recordings indicates. This may be attributed to the fact that the constitutive models do not consider the concept of volumetric threshold shear strain below which no excess pore water pressure is generated during cyclic shear loading. It was also found that there was a significant overestimation of the excess pore pressure for the 2010 El Mayor Cucapah earthquake simulations. This may be attributed to the effect of the site shaking history which increased the site resistance to liquefaction. This added resistance was not reflected by the numerical model since it was calibrated with samples extracted about 25 years before the 2010 earthquake.