Surrogate‐Based Time‐Continuous Seismic Fragility Assessment of Corrosion‐Deteriorated Reinforced Concrete Bridges Under Climate Change

ABSTRACT Seismic fragility assessment of reinforced concrete (RC) bridges exposed to corrosive environments must reliably capture time‐dependent deterioration impacts. Climate change compounds this challenge through nonstationary temperature and humidity variations that accelerate bridge corrosion. Due to the embedded significant computational cost, prior studies typically develop corrosion‐based seismic fragility models at coarse, discrete time intervals (e.g., every 10 years). In contrast, this study proposes a surrogate‐based, two‐stage analysis framework that enables high‐resolution, time‐continuous seismic fragility assessment of columns and bearings in RC bridges under corrosion deterioration and climate change. In Stage I, normalized deterioration factors are treated as explicit stochastic inputs that modify bridge attributes (e.g., material properties and cross‐sectional dimensions), which are fed into numerical simulations to obtain corresponding bridge seismic demands and capacities. These bridge demand/capacity data are then used to train Gaussian‐process surrogates conditioned directly on the deterioration factors. In Stage II, climate‐change and corrosion scenarios are separately modeled to generate nonstationary, time‐dependent trajectories of deterioration factors, which are propagated through the surrogates to produce continuous, time‐PGA fragility surfaces over the bridge lifetime. Monte Carlo simulations are also integrated to quantify and propagate uncertainties across different modeling steps. The proposed methodology is demonstrated through three‐span continuous RC bridges in Eastern Canada, a region characterized by moderate seismicity and pervasive chloride‐induced corrosion due to heavy use of de‐icing salts. Results show that the framework captures how ground motion intensity, climate change, bridge service life, construction year, and geometric attributes jointly influence high‐resolution evolutions of fragility estimates. The methodology provides an extensible platform for time‐dependent, multi‐hazard fragility and risk assessment and offers valuable insights to support bridge design, inspection, maintenance, and retrofit decision‐making.