Bayesian Earthquake Vulnerability Modeling for Concrete Buildings

Accurate assessment of seismic vulnerability and fragility is essential for robust risk evaluation. However, uncertainty in the estimated ground-motion intensity measure often complicates this task. Traditional modeling approaches frequently rely on fixed, best-estimate intensity measures. As a result, these methods may over- or underestimate damage probabilities at both lower and higher ground-motion intensities, leading to unreliable risk estimates. This study presents a Bayesian framework for modeling the seismic vulnerability of concrete buildings, leveraging empirical loss data from the South Iceland Seismic Zone. The methodology is grounded in zero-inflated beta regression and Markov Chain Monte Carlo simulations, enabling robust uncertainty quantification in model parameters and predictions. Ground-motion intensity measures (IM) are treated probabilistically to address inherent uncertainties. The findings underscore the importance of accounting for both zero-loss structures and ground-motion IM uncertainty when calibrating vulnerability models, which in turn informs the development of seismic design guidelines, enhances damage prediction for future events, and supports informed policymaking for disaster preparedness and resource allocation.

کلیدواژه ها:

Bayesian inference ، seismic vulnerability ، earthquake fragility model ، uncertainty modeling