Seismic Fragility Curve Assessment Based on Synthetic Ground Motions with Conditional Spectra

Fragility curve evaluation using ground motions compatible with conditional spectra allows for ground motion selection with good consistency with the hazard. However, this evaluation requires selecting ground motions from a database, which is not straightforward. Moreover, it relies on a fragility curve fitting procedure, which may fail to give a solution. Lastly, it estimates the statistical uncertainty of the curve based on simple bootstrap resampling. This paper proposes a novel framework for estimating fragility curves with two major improvements. First, the proposed fragility curve fitting procedure is not iterative and does not require an initial estimation of the curve parameters, unlike the existing curve fitting procedure. Second, the statistical uncertainty of fragility curves is estimated based on the Fisher's information matrix, which is a rigorous alternative to bootstrap resampling. Moreover, stochastic ground motions compatible with the conditional spectra are generated and used as excitations. This is a practical alternative, as it does not require a ground motion database for scenario-specific selection. Results obtained with the developed framework are compared to results based on existing procedures in a case study of in-structure components in an industrial building. We analyze cases in which our curve fitting procedure gives a solution more reliably than the existing. Moreover, the developed procedure for estimating the uncertainty of fragility curves leads, in this case study, to better estimations for the statistical uncertainty of the fragility curves.