Simulation of Grain Fracture-Induced Pile Creep and Setup in Sands

Abstract

Pile setup is now generally recognized as a characteristic feature of displacement pile axial response in sand. However, a clear understanding of the phenomena underlying pile setup is still missing because reliable and practical methodologies to study the problem are lacking. It is very difficult to measure and control all possible contributing factors in field testing experiments, whereas physical laboratory tests often fail to replicate field observations. The resulting ambiguities undermine proposals to incorporate setup in design procedures. A new methodology to study pile setup based on the discrete-element method (DEM) is proposed here. We build a virtual centrifuge chamber using a granular material DEM model calibrated to represent Fontainebleau sand, a quartz sand. The DEM material model, which incorporates delayed grain fracture, was previously verified by reproducing creep and relaxation in element tests, but relies on much simpler measurements for material calibration. Model piles are installed in the chamber by jacking, performing long (up to 1 month) creep stages and measuring shaft resistance at regular intervals by means of pullout tests. The simulated pile creep and pullout responses were well-aligned with previous experimental observations. The simulations presented strongly support the hypothesis that delayed grain fracture is a significant contributor to pile setup. The methodology presented opens the way to quantify this and other physically based explanations of pile setup and creep.