Probabilistic Assessment of Collapse Limit- State in Steel Frames by Simulating Failure Modes Using Bayesian Probability Network

Due to complex nature of the collapse phenomena, along with its important role in structural performance, collapse assessment has recently gained a great attention in performance based earthquake engineering. Furthermore, uncertain sensitive parameters on seismic collapse performance have intensified the complexity. In this paper, a novel probabilistic approach based on simulating failure modes within IDA framework with application of Bayesian Probability Network (BPN) is proposed to evaluate collapse of structures. Applying pushover analysis, likely Failure Modes (FMs) are recognized first, and then along with sensitive Random Variables (RVs), these FMs are incorporated in BPN. Conditional probability of these FMs related to the incorporated RVs, are calculated within IDA agendum, which directly considers record-to-record uncertainty.
Some of the results investigated in this research are: Comparing mean annual rate of collapse derived from the proposed method with the conventional limit-states considered in IDA approach, detection of structural response in a probabilistic framework and updating through BPN. Totally, it can be claimed that via the newly proposed methodology not only the safety index of structures are calculated in an effective way, but also structural response in the collapse limit state is detected probabilistically.