Seismic response of piles in layered soils: Performance of pseudostatic Winkler models against centrifuge data

Abstract

In this study, the suitability of the pseudostatic approach for the seismic analysis of pile foundations in layered soils is explored by means of experimental data from centrifuge tests performed at 60g. A free-head single pile and a capped (1 × 3) pile group, embedded in a two-layered soil comprising a soft clay layer underlain by dense sand, are tested in the centrifuge under sinusoidal and earthquake excitations. For the pseudostatic analysis, a one-dimensional Winkler model is developed using hyperbolic p-y curves from design codes. The kinematic and inertial loads on the pile foundations are derived using the experimentally measured free-field soil displacements and accelerations, respectively. Different approaches of modifying the p-y relationship to account for soil layering are compared. The importance of considering peak spectral acceleration in lieu of peak ground acceleration at the soil surface to compute the inertial force for the pseudostatic analysis is highlighted. Pile group effects are investigated by considering p-multipliers from literature to account for pile-soil-pile interaction. Results reveal that: (i) for low-intensity seismic motions, the pseudostatic approach with inertial pile-head loading stemming from peak ground acceleration (PGA) at soil surface led to a reasonable agreement of the maximum bending moment with experimental data for both single pile and pile group, (ii) for high-intensity base excitations, the use of the peak spectral acceleration, instead of PGA, at soil surface with suitable damping considerations to derive the inertial load in the pseudostatic model provided a maximum bending moment prediction that was acceptable for the single pile but conservative for the piles in the group compared to the centrifuge records.