Two-dimensional (2D) frame analysis has been used extensively to predict the seismic response (e.g., peak story drifts and collapse) of buildings subjected to earthquake excitation. In this type of analysis, the vertical lateral force resisting systems (LFRS) in the building (i.e., braced frames or moment frames) are modeled assuming the horizontal diaphragm system is rigid and that the vertical LFRS responds independent of the rest of the building. In an actual building, the diaphragm system deforms contributing to lateral drifts, and interacts dynamically with the vertical LFRS. Furthermore, inelasticity and failure of the diaphragm can have substantial effect on the seismic behavior and collapse of the overall building. It might be expected, therefore, that 2D frame analysis would result in smaller estimates of seismic collapse potential than three-dimensional (3D) analysis. For this reason, FEMA P695 specifies that the collapse margin ratio obtained from 3D analyses should be amplified by a factor of 1.2, irrespective of diaphragm modeling approach, to account for larger collapse potential found in 3D analyses compared to 2D analyses.To understand the effects of 3D analysis, different diaphragm modeling assumptions, and unidirectional vs. bidirectional seismic excitation on the seismic collapse response of buildings, a computational study was conducted with models varying in complexity from those equivalent to a nonlinear 2D frame analysis to a full 3D building analysis with nonlinear behavior in both the vertical LFRS and the diaphragm system. Buckling restrained braced frames were used as the vertical LFRS and the diaphragm system was concrete-filled steel deck at the floors and bare steel deck at the roof. Beams and columns were modeled using nonlinear beam-column elements, while the diaphragm was modeled as a system of diagonal truss elements that were either rigid, elastic, or nonlinear. Nonlinear response history analyses were performed with the far field ground motion set from FEMA P695 and resulting performance in terms of drifts and collapse were evaluated.Results showed that diaphragm elasticity led to an increase in the period for the first four modes of between 5% and 34%. Diaphragm deformations led to similar collapse probabilities, but as much as 20% larger story drift for the design earthquake and maximum considered earthquake hazard levels. However, when the hazard level was increased to a level called the ACMR_(10%) in FEMA P695, the building behavior changed and diaphragm inelasticity shared some of the displacement demands, thereby reducing the deformation demands in the BRBF and preventing some of the collapses associated with BRB fracture. At this larger hazard level, the building models that included diaphragm inelasticity had up to 43% fewer collapses than those with rigid or elastic diaphragms. It was also found that subjecting the models to bidirectional ground motion pairs as compared to single unidirectional ground motions, led to 11% to 16% larger peak in-plane BRBF drift demands at the MCE hazard level, and as much as 80% more collapses at the ACMR_(10%) hazard level. This implies that the FEMA P695 factor of 1.2 on collapse margin ratio for 3D models subjected to bidirectional ground motion pairs may warrant additional investigation.
展开▼