Seismic Site Response Analysis with Porewater Pressure Generation
Advisor: Professor Youssef M. A. Hashash
Microsoft Teams Link
The generation, buildup, and dissipation of excess porewater pressure in saturated sandy soil during earthquakes can substantially influence the seismic response of the ground and the seismic design of structures. Excess porewater pressure can cause a decrease in soil stiffness and strength, which can modify the ground response during earthquake loading. When the excess porewater pressure builds up to a high level, soil liquefaction might be triggered, which is one of the most catastrophic results of earthquakes. When liquefaction triggers, soil can lose stiffness and strength significantly and flow like a viscous fluid. Critical infrastructures, such as buildings, bridges, earth dams, harbor retaining structures, manmade islands, and pipelines, might be damaged. Therefore, understanding the effects of porewater pressure on the ground seismic response plays an essential role in performance-based seismic design, assessment of infrastructure resiliency, and seismic hazard mitigation. The cyclic stress approach is the most widely used analysis method to evaluate the potential of liquefaction initiation in the field. However, this empirical approach involves significant uncertainties in quantifications and correction factors of soil liquefaction resistance and seismic demands. One-dimensional site response analysis with empirical porewater pressure is also widely used for effective stress analysis. But the calibration of the empirical porewater pressure models requires advanced laboratory test data, which might be biased to certain soils. Besides, models in this method have limitations in simulating the soil dilative behavior and volumetric response. The state-of-the-art liquefaction analysis method uses a solid fluid coupled finite element framework with a nonlinear effective stress soil model. Evaluation of the performance of a nonlinear effective stress soil model for liquefaction modeling remains a challenge in numerical analysis.
A simplified three-dimensional nonlinear effective stress material model, I-soil, was recently developed and implemented in LS-DYNA, a commercial finite element package. It has been successfully applied in multiple total stress soil-structure interaction problems, for instance, concrete-faced rockfill dams, nuclear power plants, tunnels, highrise buildings, and buried underground reservoirs. However, using it for effective stress analyses, especially for liquefaction problems, has not been validated. Therefore, this study evaluates the applicability and performance of I-soil/LS-DYNA to liquefaction analysis from an element to a system level, then applies it to a
large-scale parametric study that focuses on liquefaction triggering. Key implementation points are identified and highlighted throughout this study to ensure the proper application of I-soil. These points include (1) calibrating the model parameter controlling dilatancy behavior based on constant volume friction angle, (2) maintaining the shear strength ratio when subdividing soil layers, (3) ensuring that the stress-strain pairs of the backbone curve have positive and monotonically decreasing tangent shear modulus, (4) applying correct initial vertical and horizontal stress state in total and effective stress analyses, and (5) modeling water table as a phreatic (zero-pressure) boundary instead of a non-flow boundary. Besides, I-soil material parameters can easily be calibrated based on field measurements, empirical correlations, and laboratory tests. The generation of model parameters and one-dimensional shear beam models are automized using coding scripts.
Element level evaluation indicates that the model can capture the empirical correlations and laboratory observed soil behaviors with some noticed deficiencies that are found not to affect the model's performance for liquefaction triggering analysis at the system level. At the system level, the evaluation indicates that I-soil can capture the characteristics of recorded accelerometer and porewater pressure transducer measurements in liquefaction centrifuge tests and field cases at various depths. Two- and three-dimensional models are more suitable for problems with geometric and boundary effects. Finally, a large-scale liquefaction parametric study is conducted while utilizing High-Performance Computing resources provided by Stampede2. The seed soil profiles are based on real liquefiable sites in the United States, Japan, and New Zealand. A broad range of recorded motions is selected from the NGA-WEST2 database to capture the uncertainties in earthquake motion intensity and frequency components. Randomization in shear wave velocity profiles is applied with compatible shear strength profiles to capture the variabilities in soil profiles. Variations in water table depths are considered to account for the effects of water table fluctuation. The results show that excess porewater pressure can amplify the impact of site resonance. The difference between total and effective stress analysis becomes significant when excess porewater pressure is considerable. In addition, using the I-soil model on the LS-DYNA platform has proven to be computationally efficient.