Evaluation of the performance of piled bridge abutments affected by liquefaction-induced ground deformations through centrifuge tests and numerical analysis tools

Deformations of a bridge approach embankment due to earthquake-induced liquefaction in the underlying soil can impose large loads or demands on the piled abutment and pile foundations that are embedded through the embankment. However, because the piled foundation is embedded through an embankment of finite width, the foundation may actually restrain the movement of the embankment and thus create smaller displacements and loads on the piled foundation than would be expected based on analyses that uncouple the embankment deformations from the bridge response.;To investigate this phenomenon, three centrifuge tests were conducted. Each centrifuge test modeled two identical bridge approach embankments underlain by liquefied soil, with a total of six embankments modeled. The embankments were constructed of dry sand and underlain by saturated loose sand, which liquefied during shaking. Each centrifuge test included one embankment with a pile group at its crest and one embankment without a pile group. The results from the centrifuge tests provided physical evidence that the resisting forces from piles can reduce the embankment displacements that result from earthquake-induced liquefaction in the underlying soils. This beneficial mechanism, known as pile pinning, should be included in practice when analyzing or designing piled bridge abutments in liquefiable ground, as it could result in significant economic savings (e.g., soil remediation or foundation retrofit).;Two options to account for this pile pinning effect in design are (1) equivalent static analyses (ESA) and (2) non-linear deformation analyses (NDA). ESA involves three parts: (1) calculating embankment displacements for a range of restraining forces from the piles and bridge superstructure through slope stability and Newmark sliding block analyses, (2) calculating the restraining forces from the pile foundation and bridge superstructure for a range of possible embankment displacements through pushover analysis with imposed free-field soil displacement profile, and (3) calculating the compatible displacement and pile restraining force from the first two steps. The NDA involves using the finite element or finite difference method with material constitutive models to calculate the dynamic response. Some of the key steps in the NDA are generating the finite element or finite difference mesh, calibrating the material models, establishing the pre-shaking stress condition, and updating the geometry and stresses.;Both the ESA and NDA were used to analyze these centrifuge tests, with the purpose of evaluating these tools' predictive capabilities. For these centrifuge tests, the ESA significantly over-predicted displacements and demands on the piles; whereas the NDA provided a better comparison (although still with a tendency to over-predicted). The better predictive capability of the NDA was attributed to several modeling capabilities not possible with the ESA. The relative merits of ESA and NDA for current practice depend on how decisions in design may be affected. Future advances can be expected with NDA methods, including improvements in the constitutive models, numerical methods, and software usability. Future advances are unlikely to be realized with ESA methods given their inherent simplifications of the physical processes.