Large diameter shield tunnels traversing liquefiable soil-rock strata are highly susceptible to seismic hazards, as earthquake-induced soil liquefaction significantly reduces soil strength and stiffness. Therefore, it is crucial to accurately assess the seismic performance of these tunnels. This study first establishes a numerical model for tunnel seismic response analysis, considering soil liquefaction, segment nonlinearity, and joint deformation. The validity of the model is affirmed through experimental, theoretical, and additional numerical simulations. The probabilistic seismic demand models are established employing the seismic database consisting of 120 ground motion records. Subsequently, a quantitative selection method for the optimal Intensity Measure (IM) based on fuzzy comprehensive evaluation is proposed, identifying Velocity Spectrum Intensity (VSI) as the most suitable among 29 commonly used IMs, and the IMs related to duration exhibit poor performance. The study then categorizes tunnel damage into three states: minor, moderate, and extensive, using joint opening as the damage measure. Finally, seismic fragility analysis is employed to assess seismic performance of tunnel, and fragility curves derived using VSI and Peak Ground Acceleration (PGA) is compared. The results indicate that PGA, a commonly used IM, significantly underestimates the probability of damage to the tunnel, with a maximum underestimation of 22.4%.

The seismic damage investigations indicated the high probability for the pile-supported structure to suffer further destruction due to the effects of strong aftershocks, but the seismic fragility assessment of pile-supported structures considering aftershocks rarely attracted notice. In this study, the seismic fragility of a pilesupported structure under the mainshock-aftershock sequence is systematically assessed through the simulation results based on a solid-fluid coupling finite element model. Firstly, the numerical model is validated by the test date got from a centrifuge model in order to check its effectiveness. Then the strong motion records recorded at liquefied sites are selected for synthesizing the seismic sequences through scaling and combination. And the Engineering Demand Parameter (EDP) is dertemined as the residual displacement, representing the cumulative damage caused by seismic sequences. Additionally, the quantitative limit states of EDP corresponding to four damage states considering pile-soil interaction is defined by a push-over analysis. Furthermore, since the choice of optimal Intensity Measures for seismic sequences is beneficial for enhancing the credibility, 24 sets of Intensity Measures are assessed from three sides including efficiency, practicality, and proficiency. Finally, the mainshockaftershock fragility surfaces with two Intensity Measures are proposed, considering the random uncertainty. And the results indicate that mainshock aftershock sequences could induce higher exceedance probabilities of a limit state when comparing to under the excitation of mainshock only. The research results could offer a basis for the seismic performance evaluation of the kind of pile-supported structure.