Particle contact evolution: A precursor mechanism for earthquake-induced liquefaction prior to pore-pressure buildup

Dear Editor,

Earthquake-induced soil liquefaction is a destructive geohazard that repeatedly damages buildings, lifelines, and critical infrastructure during strong shaking. It has traditionally been attributed to cyclic loading that elevates pore-water pressure, reduces effective stress, and destabilizes the load-bearing particle skeleton. Previous works have suggested that liquefaction is closely associated with changes in the internal particle structure, particularly the breakdown of interparticle contact networks. However, these insights are often based on indirect observations or numerical assumptions and lack direct experimental validation of the underlying mechanism.

In this letter, we establish an experimentally decoupled validation framework for investigating the sequence of microstructural and hydraulic events in earthquake-induced liquefaction of saturated loose sand. Building on important insights from recent discrete-element simulations and recognizing that grain-scale physical experiments that resolve contact-network evolution and hydraulic response remain relatively scarce, we combine complementary physical experiments with simplified numerical modeling. This framework enables systematic examination of the interplay among contact degradation, contractive volumetric change, and excess pore-pressure development. Specifically, computed tomography (CT) reconstruction with three-dimensional (3D) printing isolates hydraulic effects, photoelastic measurements capture real-time interparticle force evolution, and computational fluid dynamics-discrete element method‌ (CFD-DEM) modeling probes post-instability fluid dynamics. Together, these approaches provide a particle-scale route for studying liquefaction, with potential for future integration into unified 3D measurements of contact evolution and pore pressure.

We first designed an idealized limiting-case experiment in which contact evolution is mechanically suppressed while the pore geometry is kept fixed so that the hydraulic response can be isolated. Although this setup cannot fully reproduce the complete behavior of natural sand, it provides a controlled way to isolate the hydraulic response from structural evolution. A loose quartz-sand specimen was scanned by X-ray CT, and the 3D pore architecture was reconstructed at a representative-elementary-volume (REV) scale (800 × 800 × 800 pixels). This REV was then used to compare and analyze the microstructures of the natural sand and the 3D printed replica. Using stereolithography, we printed a rigid, monolithic replica that preserved the specimen’s overall dimensions (180 mm in diameter and 100 mm in height), porosity, coordination number, and tortuous flow paths but mechanically bonded the grains into a continuous solid (Figures 1A–1G). We then conducted one-degree-of-freedom horizontal shaking tests on both saturated natural sand (Figure 1H) and the saturated printed analog (Figure 1I) under identical boundary conditions. The horizontal displacement amplitude was fixed at mm. For the natural quartz-sand specimen tested in this study, no liquefaction and no measurable excess pore-pressure rise were observed at 4 Hz. When the frequency reached 5 Hz and above, liquefaction was consistently triggered within a few loading cycles, as evidenced by a rapid increase in excess pore pressure followed by gradual dissipation toward the initial level (Figure 1J). In sharp contrast, under the same loading history and identical hydraulic boundary conditions, the saturated 3D-printed specimen exhibited no significant positive excess pore-pressure buildup across the tested frequency range, with the pore-pressure signal remaining close to the initial baseline throughout the excitation (Figure 1K). Because contact rearrangement is precluded in the printed material, these observations demonstrate that cyclic loading does not generate excess pore pressure in the absence of contact failure. In the natural sand, cyclic shear leads to contact loss and irreversible contractive compaction. As the pore volume decreases, the nearly incompressible pore fluid resists volume change and an excess pore pressure is generated. In the shaking-table liquefaction tests on saturated loose sand conducted in this study, the excess pore pressure observed in liquefying sand therefore accompanies, rather than precedes, the structural collapse of the contact network.