Seismic failure mechanisms of soil-structure-tunnel systems in liquefiable urban deposits: vector and scalar fragility assessment

Abstract

This study presents a comprehensive investigation into the seismic response and failure mechanisms of soil-structure-tunnel interaction (SSTI) systems in liquefiable urban deposits through rigorously validated finite element simulations. A two-dimensional plane-strain model representing a shallow rectangular tunnel beneath adjacent mid-rise buildings is developed in Midas GTS NX and calibrated against centrifuge test data to ensure high-fidelity reproduction of liquefaction-induced behavior. The modified UBCSAND constitutive model is employed to capture the nonlinear effective-stress response of loose saturated sand, while adaptive meshing and Rayleigh damping enhance numerical stability during strong shaking. A suite of ten real earthquake ground motions with diverse spectral characteristics is applied to examine the evolution of excess pore-water pressure, deformations, ground-surface heave, and building settlement. The results show that surface buildings substantially constrain liquefaction and deformation beneath their foundations, whereas free-field soils and zones surrounding the tunnel undergo extensive strength degradation and upward displacement. A transient soil-arching mechanism develops above the tunnel crown, delaying uplift until near-complete loss of shear resistance. Strong seismic excitations lead to pronounced heave in the street corridor and differential settlement and tilting of the buildings. Correlation analyses between deformation demands and seismic intensity measures (IMs) reveal that cumulative and duration-sensitive IMs exhibit stronger predictive capability for SSTI responses than peak parameters. Scalar-and vector-valued fragility functions are developed using incremental dynamic analysis (IDA) to quantify the probability of exceeding specified damage states. Results demonstrate that scalar fragility estimates vary substantially depending on the selected IM, whereas vector-valued fragilities significantly reduce epistemic uncertainty by capturing the combined effects of complementary IM pairs. Vector IMs provide the most reliable predictions of deformation demands. Overall, the study highlights that liquefaction-induced SSTI behavior is governed primarily by seismic energy and displacement demands, underscoring the critical importance of vector-valued fragility models for realistic seismic vulnerability assessment and resilience-based design in liquefaction prone urban environments.