Fragility analysis of nonproportionally damped inelastic MDOF structural systems exposed to stochastic seismic excitation

A novel methodology for conducting efficiently fragility analysis considering nonproportionally damped inelastic multi degree-of-freedom (MDOF) structural systems subject to stochastic seismic excitations defined by an advanced stochastic model consistent with magnitude-epicentral distance earthquake properties is developed. To this aim, an approximate stochastic dynamics technique for determining the system response amplitude probability density functions (PDFs) is developed. Firstly, relying on statistical linearization and state-variable formulation the complex eigenvalue problem is addressed through the time-domain. Secondly, utilizing the forced vibrational modal properties of the linearized MDOF system in conjunction with a combination of deterministic and stochastic averaging treatment, the MDOF system modal response amplitude process PDFs are determined. The modal participation factors are then defined for the complex-valued mode shapes and the total response amplitude process PDFs are provided in physical coordinates. Subsequently, appropriate limit states are related with the higher order statistics of the engineering demand parameters (i.e. that of the PDF) for quantifying structural system related fragilities. Nevertheless, due to the vector-valued nature of the adopted intensity measure, depicting system fragilities takes the form of three-dimensional fragility surfaces. The associated low computational cost renders the proposed methodology particularly useful for efficient structural system fragility analysis and related performance-based engineering design applications. A multi-storey building structure comprising the bilinear hysteretic model serves as a numerical example for demonstrating the reliability of the proposed fragility analysis methodology. Nonlinear response time-history analysis involving a large ensemble of compatible accelerograms is conducted to assess the accuracy of the proposed methodology in a Monte Carlo-based context.