Characterizing the micro-mechanical triggering of liquefaction via the Seismic Persistence Index

Abstract

Predicting seismic liquefaction requires identifying micro-mechanical warning signs that appear before macroscopic failure. The number of particle contacts is described by conventional metrics like the coordination number (Z), but they frequently miss the discrete onset of instability. To address this, the study introduces the Seismic Persistence Index (SPI), a new micro-mechanical descriptor designed to quantify the structural memory of the granular contact network. The SPI is evaluated using a benchmark Discrete Element Method (DEM) dataset for Toyoura sand under both drained and undrained loading conditions. The analysis reveals a striking micro-mechanical paradox. Under undrained loading, the SPI drops abruptly from approximately 0.56 to a brief minimum near 0.05 before stabilizing between 0.30 and 0.35, while global connectivity remains close to its peak, with Z ≈ 9.73 at collapse and rising slightly to ≈ 9.82 afterward. This collapse is accompanied by a sharp instability impulse in the stepwise SPI decay (ΔSPI ≈ − 0.51) and a simultaneous spike in the skewness of the probability density function (PDF) of normalized contact force distributions (γ > 200), reflecting extreme force localization within the granular skeleton. Drained loading, by contrast, shows gradual fabric renewal with no sudden persistence loss. Within the scope of this dataset, SPI appears to offer a high-resolution micro-mechanical signature of the liquefaction trigger. These findings motivate the hypothesis that persistence-based metrics may serve as physically meaningful indicators of seismic instability, pending validation across broader materials, loading paths, and drainage conditions.