# Investigation of cyclic liquefaction with discrete element simulations

## TL;DR

This study uses discrete-element simulations to model sand behavior under cyclic loading, proposing new measures to predict liquefaction severity and analyzing the effects of irregular seismic sequences.

## Contribution

It introduces a calibrated DEM model for sand, develops a methodology to quantify seismic sequence severity, and proposes scalar measures for predicting liquefaction.

## Key findings

- A DEM model accurately replicates sand behavior under cyclic loads.
- A new methodology ranks seismic sequences by severity.
- A stress-based scalar measure effectively predicts liquefaction.

## Abstract

A discrete-element method (DEM) assembly of virtual particles is calibrated to approximate the behavior of a natural sand in undrained loading. The particles are octahedral, bumpy clusters of spheres that are compacted into assemblies of different densities. The contact model is a Jager generalization of the Hertz contact, which yields a small-strain shear modulus that is proportional to the square root of confining stress. Simulations made of triaxial extension and compression loading conditions and of simple shear produce behaviors that are similar to sand. Undrained cyclic shearing simulations are performed with nonuniform amplitudes of shearing pulses and with 24 irregular seismic shearing sequences. A methodology is proposed for quantifying the severities of such irregular shearing records, allowing the 24 sequences to be ranked in severity. The relative severities of the 24 seismic sequences show an anomalous dependence on sampling density. Four scalar measures are proposed for predicting the severity of a particular loading sequence. A stress-based scalar measure shows superior efficiency in predicting initial liquefaction and pore pressure rise.