School of Civil and Environmental Engineering

Ph.D. Thesis Defense Announcement

New Approach to Evaluate Soil Behavior during/after Liquefaction from Particle Level Responses

By

Tianlong Xu

Advisors:

Dr. J. David Frost

Committee Members:

Dr. Arun M. Gokhale (MSE), Dr. Paul W. Mayne (CEE), Dr. Susan E. Burns (CEE), Dr. Gregory Hebeler (Golder Associates, Inc.)

Date & Time: Tuesday, March 12, 7:30am

Location: SEB 122

As a worldwide major hazard, soil liquefaction has drawn the attention of the geotechnical researchers in the past several decades. To date, the nature of soil liquefaction has not yet been fully understood due to the high complexities of interpreting its triggering processes as well as estimating its consequences. The mainstream approaches to evaluating soil liquefaction rely heavily on the penetration data collected by several in-situ testing probes (e.g., CPT). The acquisition process of these in-situ data may cause some disturbance on the original soil state. In addition, the causality of soil liquefaction exhibits multivariate controlled behavior so that analyses on a subset of the factors or on an incomplete dataset may result in significantly scattered results. Motivated by these two concerns, the on-going research primarily focuses on two parts: examining the probe-soil interaction models of multiple in-situ testing approaches and exploring the multi-stage and multi-factor governed soil liquefaction from particle level responses (DEM analyses).
The research suggests that inserting probes into the soil can significantly disturb the original soil state. Appropriate data corrections as well as device modifications taking into account the insertion effect are shown to be desirable in yielding a better prediction on the original soil state. The dimensions of the probes are proven to be highly correlated to the severity of the insertion effect. The numerical results also suggest that despite significant insertion effects, the measurements collected during the initial probe-soil interaction stage are still more reliable than those collected during the subsequent stages where additional soil disturbances may occur. The research also discusses establishing links in between the horizontal soil stress and the penetration resistance readings, which may yield more robust in-situ data interpretations. 
In order to explore the particulate soil behavior during and after liquefaction, two DEM model systems (Model A and Model B) were developed and tested. Model A simulates the process of liquefaction by controlling the confining stress in the boundaries of the soil mass, while Model B is controlling the pressure in a novel pore structure system (based on graph theory), which is integrated in the DEM soil particle system. Model A qualitatively evaluates soil liquefaction by parametrically testing a series of numerical samples that are configured with different combinations of parameters such as the confining stress, seismic demand, soil density, and soil gradation. Model B quantitatively evaluates soil liquefaction by reproducing the experimental results of a group of centrifuge tests and cyclic simple shear tests in DEM and discussing the alteration of both the soil skeleton and pore structure. The parametric analyses with Model A identify confining stress as the predominant factor that influences the liquefaction process, which is consistent with historical studies. Contact reformation and different soil behaviors for clean sand sample and well-graded sample are other major findings from Model A. Both model systems suggest that the liquefaction process evolves with significant alteration of the soil skeleton. Model B further suggests that the alteration of the pore structure is another critical fact that leads to a new framework to understand the liquefaction triggering mechanism.