Evaluating soil pressures on a deep foundation element due to liquefaction-induced lateral spreading

The behavior of bridge foundations during earthquakes includes the inertial response of the foundation and superstructure system, and the kinematic forces induced by the lateral spreading ground. Both must be analyzed in practice to design a bridge foundation, or other foundation type, subject to seismic forces and potential lateral spreading-induced pressures. The kinematic loadings on a large, rigid caisson due to an all-sand profile undergoing liquefaction and subsequent lateral spreading are evaluated in this thesis. While a relatively large volume of work has been completed pertaining to the lateral loads imposed on flexible foundation elements, such as relatively small diameter pile foundations, there is a paucity of work pertaining to lateral pressures of larger, more rigid, foundation elements. Fourteen centrifuge tests were conducted at the NEES facility at Rensselaer Polytechnic Institute (RPI) to investigate the lateral spreading-induced kinematic pressures on a relatively large deep foundation (caisson) and to observe the related ground movement of a lateral spread. Both “unprotected” and “protected” caissons were modeled. The protected caissons included the addition of a ground deflection element, or wall, on the upslope side of the caisson intended to limit net pressures transferred to the caisson by the moving ground. The unprotected caissons had no upslope ground deflection wall. Tactile pressure sensors were used to directly record these increases in lateral pressure on the caisson during the shaking events, as well as the initial geostatic pressures prior to liquefaction and the final geostatic pressures after shaking and excess pore water pressure dissipation. The tactile pressure sensors: (1) provided good hydrostatic pressures under most any testing condition, and geostatic pressures consistent with theoretical at-rest pressures provided the model container was rigid, (2) exhibited dynamic pressures (minima and maxima pressure spikes) lower than actual pressures as independently measured with pore water pressure transducers (PPT) and, uncorrected, they were considered unreliable, (3) yielded reasonable average (kinematic) pressures during lateral spreading, and (4) showed that the normal pressures measured declined substantially under some conditions when shearing forces were transmitted to the pressure sensors as in lateral spreading conditions. A pressure correction was developed to account for the relatively low sampling rate of the tactile pressure sensors in this work and to correct for the low dynamic pressure measurements obtained in the raw data. The collective input energies, delivered to the models at the base of the model container directly from the electro hydraulic shaker, exhibited a coefficient of variation (COV) of about 0.25, chiefly as a result of variations in the shaker hydraulic system. The input motion energy affected the maximum depth of liquefaction, the shape of the PWP and acceleration records, and the magnitude of lateral displacement. However, even with these input motion variations, ground behavior characteristic of lateral spreading on a gentle slope was evident in the test results including negative dips in the pore water pressure (PWP) records coinciding with transient downward spikes in the acceleration-time records. This response was consistent with observations from previous centrifuge studies with similar configurations. The COV for PWP response, ground acceleration response, and lateral displacement magnitudes ranged from about 0.05 to 0.30, indicating generally reproducible model preparation techniques and COV values consistent with other geotechnical measurements reported in the literature. However, based on these findings, the input motion energy must be carefully monitored and controlled in a centrifuge testing program because it appears that variations in input energy are responsible for much of the variation in soil response observed among the tests. During lateral spreading, as the ground moved toward the foundation, a passive wedge of soil formed upslope of the model foundation. The size of this passive wedge controls the net pressure exerted on the deep foundation element. The passive wedge was observed directly by monitoring surface displacement measurements and by examining ground movement patterns with depth in the centrifuge tests. From these experiments and simulations, parameters were extracted (passive wedge angle, ; passive wedge depth, h; and pore water pressure ratio, ru; all occurring at time steps of maximum moment at the caisson base) that describe the size, shape, and behavior of the passive wedge. The  value was 16º ± 2º, h was between 4 m and 7 m, and the ru values varied with depths from about 0.2 near the surface, to 0.8 at a depth of 10 m. Analytical approaches to modeling the results [the strain wedge model (SWM), and a hybrid Rankine-Broms-liquefied strength ratio approach (hybrid approach)], were found to be viable for estimating lateral pressures under these conditions. Another approach, often used in practice, for estimating the bending moments for design include determining the bending moments on the caisson inferred the assumption of: (1) a linearly increasing net pressure, and back-calculating an equivalent ‘K’ value; and (2) doing the same, but assuming that the moment was caused by a constant net pressure on the caisson and back-calculating an equivalent constant pressure (p) value. This approach was also pursued and the resulting K and p values were 0.61 and 40 kPa, respectively. When these values and approaches were compared with existing results by others, it was found that there was a large amount of scatter, and there was no immediate and clear correlation between larger diameter, stiffer foundation elements and those foundation elements that are much smaller diameter and more flexible. The efficacy of ground deflection walls positioned upslope of the model caisson in mitigating the effects of the laterally spreading-induced pressures was also explored. The applicable centrifuge tests included four primary centrifuge tests; two “unprotected” caisson tests lacking a ground deflection wall, and two “protected” caisson tests, with a deflection wall. The deflection walls proved successful in that: (1) the ground displacements were significantly greater for the protected caisson tests than the unprotected caisson tests, suggesting that the laterally-spreading soil was readily advancing around the deflection walls relatively unimpeded; (2) the size of the passive wedge that formed upslope of the protected caissons was considerably smaller and less developed than the passive wedges upslope of the unprotected caissons; and (3) the lateral pressures generated on the upslope face of the protected caissons were substantially lower than those of the unprotected caissons. These observations suggest that ground deflection walls may be installed for seismic retrofits of existing bridges, while large diameter foundations may be constructed with a diamond or circular shape, potentially with a similar reduction in bending moment, to mitigate the consequences of lateral spreading for new construction