Reliability-based seismic performance evaluation of steel frame buildings using nonlinear static analysis methods

Performance-based seismic design enables the structural engineer to quantify the probability that a building design will satisfy specified seismic performance objectives. A probabilistic assessment of seismic building performance requires the quantification of all of the uncertainties associated with the building design. The research in this dissertation focuses on quantifying the uncertainty in the nonlinear static analysis method to predict building demands. The nonlinear static analysis method was investigated using three different methods for calculating the target roof displacement: (1) Coefficient Method, (2) Capacity Spectrum Method, and (3) Equivalent System Method. These methods were used to predict seismic demands on 3-, 9-, and 20-story Pre-Northridge welded steel moment frame buildings using sixty earthquake ground motion records, corresponding to 50%, 10%, and 2% in 50 years seismic hazard levels. The uncertainty in the nonlinear static analysis method was investigated for the following response quantities: (1) maximum roof displacement, (2) maximum interstory drift, (3) maximum beam rotation, and (4) maximum panel zone rotation. The nonlinear time history analysis method was used as the benchmark for evaluating the uncertainty in the nonlinear static analysis method because it does not require the assumptions involved with linear analytical models and the static application of force. In addition, the accuracy of the nonlinear time history analysis method itself was evaluated by studying the measured full-scale response of two instrumented buildings. The uncertainty in these methods was incorporated into a performance-based seismic design framework.; In general, it was found that estimates of building response from the nonlinear static analysis are generally insensitive to the pattern of lateral load used to perform the pushover analysis and the assumptions made in the analytical model of the beam-column joint panel zone region. On the other hand, it was found that the accuracy of the nonlinear static analysis method tends to decrease with an increase in the level of seismic hazard. It was found that the maximum displacement response was generally overestimated by the nonlinear time history analysis method.