Seismic performance assessment of steel moment frames with generic Locally Reinforced connections

Seismic performance of Steel Moment-Resisting Frames (SMRFs) with generic Locally-Reinforced (LR) connections has been investigated incorporating both earthquake-induced and modeling uncertainties. The LR connections were numerically modeled following the suggestions made by previous researches and were used to numerically model three buildings with various heights which were designed regarding the specifications of current standards. For reflecting the record-to-record (RTR) variability, the Incremental Dynamic Analysis (IDA) procedure was utilized within the Performance-Based Earthquake Engineering (PBEE) methodology. In this regard, the structures’ performance in a multitude of structural response levels was probabilistically assessed in “considering” and “neglecting” conditions of modeling uncertainties. For performing the modeling uncertainty analysis, a number of modeling parameters that best reflected the structures’ variability due to lack of knowledge were selected as random variables. Perturbing the modeling variables from their median values, a set of sensitivity IDAs were performed and a polynomial response surface was established. Incorporation of the response surface in conjunction with Monte Carlo simulation led, finally, to a fully probabilistic performance evaluation of the studied structures in varied structural response levels. The effect of modeling uncertainties in altering the predicted performances was found to be insignificant in the range of buildings and structural elements studied in this article.

► Locally reinforced connections are numerically modeled in OpnSees software.
► 3 SMRFs are designed and modeled using these connections and are subjected to IDA.
► The performances are quantified in terms like MAF of collapse and drift hazard curve.
► Defining 4 modeling random variables the procedure is repeated 25 times.
► Modeling uncertainties are treated by response surface based Monte-Carlo method.