Simplicity versus accuracy trade-off in estimating seismic fragility of existing reinforced concrete buildings

摘要

This paper investigates the trade-off between simplicity (modelling effort and computational time) and result accuracy in seismic fragility analysis of reinforced concrete (RC) frames. For many applications, simplified methods focusing on “archetype” structural models are often the state-of-practice. These simplified approaches may provide a rapid-yet-accurate estimation of seismic fragility, requiring a relatively small amount of input data and computational resources. However, such approaches often fail to capture specific structural deficiencies and/or failure mechanisms that might significantly affect the final assessment outcomes (e.g. shear failure in beam-column joints, in-plane and out-of-plane failure of infill walls, among others). To overcome these shortcomings, the alternative response analysis methods considered in this paper are all characterised by a mechanics-based approach and the explicit consideration of record-to-record variability in modelling seismic input/demands. Specifically, this paper compares three different seismic response analysis approaches, each characterised by a different refinement: 1) low refinement - non-linear static analysis (either analytical SLaMA or pushover analysis), coupled with the capacity spectrum method; 2) medium refinement - non-linear time-history analysis of equivalent single degree of freedom (SDoF) systems calibrated based on either the SLaMA-based or the pushover-based force-displacement curves; 3) high refinement - non-linear time-history analysis of multi-degree of freedom (MDoF) numerical models. In all cases, fragility curves are derived through a cloud-based approach employing unscaled real (i.e. recorded) ground motions. 14 four- or eight-storey RC frames showing different plastic mechanisms and distribution of the infills are analysed using each method. The results show that non-linear time-history analysis of equivalent SDoF systems is not substantially superior with respect to a non-linear static analysis coupled with the capacity spectrum method. The estimated median fragility (for different damage states) of the simplified methods generally falls within ±20% (generally as an under-estimation) of the corresponding estimates from the MDoF non-linear time-history analysis, with slightly-higher errors for the uniformly-infilled frames. In this latter cases, such error range increases up to ±32%. The fragility dispersion is generally over-estimated up to 30%. Although such bias levels are generally non-negligible, their rigorous characterisation can potentially guide an analyst to select/use a specific fragility derivation approach, depending on their needs and context, or to calibrate appropriate correction factors for the more simplified methods.