11.19: Seismic fragility assessment of steel frames: Application of advanced reliability analysis

Abstract

ABSTRACTDuring the application of advanced design procedures (e.g. performance based design) engineers may need to compute the actual risk of the structures. The fragility of structures exposed to seismic effects is often characterized by the fragility curves that show the failure (or collapse) probability under an earthquake characterized by a given intensity measure. Since the accurate calculation of such curves can be computationally demanding and extremely complicated due to the high nonlinearity of the problem and the difficulties in describing the seismic effect, common methodologies often invoke significant simplifications: neglecting certain types of uncertainties, assuming certain distribution model, using simplified analysis procedures to calculate the mechanical response of the structure, etc. It is emphasized that uncertainties in the resistance are often handled in a vague manner in the calculations, since their relevance is thought to be secondary in comparison to the uncertainties in the seismic effect.Current paper illustrates a comprehensive methodology bridging over the above gaps. In order to accomplish this goal, fragility curves are determined on the basis of the full probabilistic approach. The stochastic model properly incorporates all the relevant random variables with proper distribution types and parameters. The mechanical response of the structure is computed by nonlinear dynamic (time‐history) analysis in OpenSees finite element software. Reliability analysis is completed using First Order Reliability Method in FERUM. In order to perform the iterative calculations, the coupling of the stochastic and the mechanical model had to be performed, which required the synchronization of the above mentioned software.The main steps and results of the current framework are illustrated through an example of a) low‐rise and b) three‐storey steel portal frames. According to the results, the uncertainties in the resistance may have a significant effect on the probability of failure. Considering, that the results of the approximate methods found in the literature may significantly differ from the results obtained by reliability analysis, there is an understandable need for the more frequent usage of the full probabilistic approach as a background for practical applications, hence all types of uncertainties can be handled elaborately.