Experimental P-Y Curves from Centrifuge Tests on Pile Foundations Subjected to Liquefaction and Lateral Spreading

The results of five centrifuge models were used to evaluate the response of pile-supported wharves subjected to inertial and liquefaction-induced lateral spreading loads. The centrifuge models contained pile groups that were embedded in rockfill dikes over layers of loose to dense sand and were shaken by a series of ground motions. The p-y curves were back-calculated for both dynamic and static loading from centrifuge data and were compared against commonly used API p-y relationships. It was found that a significant reduction in ultimate soil resistance occurred in dynamic p-y curves in partially/fully liquefied soils as compared to static p-y curves. It was also found that incorporating p-multipliers that are proportional to the pore water pressure ratio in granular materials is adequate for estimating pile demands in pseudo static analysis. 1. INTRODUCTION Liquefaction-induced ground deformations can cause severe damage to pile-supported wharves and other waterfront structures. A common approach in analyzing the lateral behavior of piles against seismic loads is using the beam on nonlinear Winkler foundation (BNWF) or p-y spring analysis. One common p-y relationship for sand is the one proposed by the American Petroleum Institute, also known as the API sand model (API 1993). While the API sand model was originally developed for the static loading conditions, it is common to modify the API sand curves for the effects of cyclic loading. However, there is no consensus on how to modify the static p-y curves for the effects of liquefaction and pore water pressure generation in loose granular soils. In previous studies, p-y springs of piles in liquefying soils were back-calculated from case histories, centrifuge model studies (e.g., Wilson et al. 2000; Brandenberg et al. 2005; Abdoun et al. 2003), full-scale tests (e.g., Rollins et al. 2005; Chang and Hutchinson 2013), and numerical analyses (e.g., McGann et al. 2011). In general, these studies found the softening effects of soil liquefaction on p-y curves, as well as a hardening behavior attributed to the dilative response of the liquefied soil around the pile. However, these studies provide varying recommendations on how to modify the static p-y curves to capture this complex behavior. Some design guidelines propose softening the static p-y curves using p-multipliers (e.g., Caltrans 2012). Other studies propose an upward concave shape for p-y curves in liquefied soils (e.g., Franke and Rollins 2013; Chang and Hutchinson 2013).