Assessing uncertainty in the computation of seismic failure rates due to record-selection

Current seismic codes suggest selecting earthquake records close to uniform hazard spectrum (UHS) estimates in the frame of non-linear time history analysis of structural systems subject to earthquakes. However, different results can be achieved using the same number of records but with different strong motion data, and also the number of records itself can impact results of the seismic fragility estimation process. The present work aims therefore to highlight how the process of record selection can significantly influence the results, being a source of relevant uncertainty. Hence, a methodology for stochastically define a hazard-consistent record selection is defined with the aim to assess its influence on the uncertainty in quantifying a seismic failure rate.