Comparative seismic fragility assessment of tunnel infrastructure embedded in soil versus rock

This study presents a comprehensive framework for developing and assessing seismic fragility models for metro tunnels in one network, addressing seismic damage to both structural and non-structural components. Detailed nonlinear dynamic analyses were conducted on two representative tunnel types - D-shape tunnels embedded in rock at varying depths, and rectangular cut-and-cover tunnels in soil with different layering profiles - using seven tunnel-soil/rock configurations and 40 site-consistent ground motions. The study integrates a rigorous treatment of axial-bending interactions for tunnel responses, capturing full-time histories of the moment demand-to-capacity ratio as the structural damage index, and assesses non-structural damage using deck accelerations. Probabilistic seismic demand models were developed through linear and bilinear regressions, with peak ground acceleration (PGA) identified as the optimal intensity measure. Results show that D-shape tunnels in rock exhibit negligible probability of damage under PGAs typical of the sites, while cut-and-cover tunnels display steep fragility curves, indicating a certain level of vulnerability at low seismic intensities. Non-structural components in both tunnel types are shown to be susceptible to damage, underscoring the need for operational safety measures, such as automatic train stoppage under high deck accelerations. The proposed fragility framework provides critical insights to be applied to integrated urban tunnel networks for identifying and prioritizing weak links that facilitate post-earthquake emergency response, retrofitting decisions, and resilience-based seismic design. • Developed high-fidelity numerical models validated against centrifuge tests. • Integrated axial-bending interaction into the structural damage index computation. • Developed bilinear PSDM of D-shape tunnels in rock due to damage mode shifts. • Offered a transferable framework for portfolio-level seismic risk assessment in tunnel networks.