Modeling and Response Prediction in Performance-Based Seismic Evaluation: Case Studies of Instrumented Steel Moment-Frame Buildings

An evaluation of the four analytical methods recommended in FEMA-356 for the estimation of seismic demands is carried out on two instrumented steel buildings. Prior to conducting the FEMA-356 analyses, computer models of each structural system were calibrated to observed response data. One of the two buildings that experienced no visible damage during Northridge was modeled as a three-dimensional system, while the second building that suffered moderate damage during the Northridge earthquake was further tuned to yield a simplified two-dimensional model so that detailed inelastic evaluations could be performed on the model. Dozens of linear and nonlinear computer analyses under both static and seismic loads were carried out on each building, and typical results representing a summary of the findings are presented in this paper. Results of the evaluations provide insight into both modeling issues and the validity of the four analytical procedures outlined in FEMA-356. It is found that calibrating structural models to observed response is sensitive to mass and stiffness modeling assumptions and indirectly to the intensity of the ground motion that has an effect on the participation of nonstructural components. Linear and nonlinear static procedures do not consistently predict interstory drift estimates, a critical parameter in seismic evaluation and design. The linear static approach was found to be conservative in one case study while the findings from the second case study were less conclusive about the validity of each analytical procedure. This suggests that significant changes are warranted in the proposed FEMA-356 Prestandard if it is to be adopted in future design codes or in seismic evaluation studies of existing buildings.