Bounds of reliability function for structural systems subjected to a set of recorded accelerograms

The present study focuses on reliability analysis of linear discretized structures subjected to ground motion acceleration modeled as a zero-mean Gaussian stationary random process fully characterized by an imprecise power spectral density (PSD) function i.e. with interval parameters. The bounds of such interval parameters are determined by analyzing a large set of accelerograms recorded on rigid soil deposits. To discard outliers from the set of accelerograms, the Chauvenet's Criterion is applied. Then, to assess structural safety, the imprecise PSD function of ground motion acceleration is incorporated into the formulation of the classical first-passage problem. Due to imprecision of the excitation, the reliability function of the selected extreme value response process turns out to have an interval nature. It is shown that the bounds of the interval reliability function can be readily evaluated by exploring suitable combinations of the endpoints of the interval spectral moments of the selected response process identified relying on structural dynamic properties and taking advantage of the dependency of the proposed imprecise PSD function on three interval parameters only.