A dependence-aware probabilistic seismic fragility framework for adjacent pile-supported wharf-shield tunnel systems in liquefiable ground

Adjacent pile-supported wharves and shield tunnels in liquefiable coastal ground may experience coupled seismic responses. Conventional fragility frameworks, however, neglect statistical dependence between adjacent infrastructure components, potentially biasing seismic risk estimates. This study develops a system-level seismic fragility framework for an adjacent wharf-tunnel system considering inter-component demand dependence. An OpenSees finite-element model is established and validated against centrifuge test results. Based on 160 ground-motion records, tunnel joint opening and maximum wharf deck displacement are selected as demand parameters. Twenty-five intensity measures (IMs) are evaluated using multiple criteria, and the preferred component-level IMs differ: ASI performs best for the tunnel (0.89), whereas PGV performs best for the wharf (0.93). To develop a consistent system-level probabilistic demand model, PGV is selected as a balanced system-level IM through Pareto screening. The PGV -based marginal PSDMs are re-established and coupled using a Gaussian copula. Component demands show positive dependence, with Spearman's ρ = 0.5671 and Pearson's r = 0.4206, indicating that independence may oversimplify system response. Fragility curves show increasing damage probabilities with PGV . At PGV = 1.2 m/s, moderate and severe damage probabilities are 0.83 and 0.25, respectively. These findings highlight unified IM selection and dependence-aware modeling for adjacent port infrastructure in liquefiable ground.