Updated robust reliability: considering record-to-record variability and uncertainty in modelling parameters

Abstract

Modeling parameter uncertainty can affect significantly the structural seismic reliability assessment. However, this type of uncertainty may not be described by a Poisson process as it lacks renewal properties with the occurrence of each earthquake event. Furthermore, considering uncertainties related to ground motion representation by employing as-recorded ground motions together with modeling parameter uncertainties can prove quite challenging. Robust fragility assessment, proposed previously by the authors, employs the structural response to recorded ground motion as data in order to update prescribed seismic fragility models. Robust fragility can be extremely efficient for considering also the structural modelling uncertainties by creating a dataset of one-to-one assignments of structural model realizations and as-recorded ground motions. This can reduce the computational effort by about two orders of magnitude. However, it should be kept in mind that the fragility concept is based on the underlying assumption of Poisson-type renewal. Using the concept of updated robust reliability, considering both uncertainty in ground motion representation based on as-recorded ground motion and non-ergodic uncertainties, the conservative error introduced by adopting Poissonian structural reliability assessment based on the robust fragility concept can be quantified.