Progressive or disproportionate collapse occurs when localized structural damage leads to widespread collapse or failure of a structure. Although the loss of any structural component in a building has the potential to initiate progressive collapse, structural columns in steel buildings are particularly susceptible to initiating this behavior if their load-carrying capacity is compromised. Steel-framed buildings can possess the capacity to bridge over a single lost column and arrest collapse, but the dynamic and three-dimensional nature of this event prevents simple design-based analysis approaches from providing accurate assessments of collapse resistance.
This research employed a set of two prototype steel moment-framed buildings to study dynamic ground-level column-loss scenarios for a variety of column locations within the structures. One building contained three stories while the other had ten. Both were intended to be representative of typical perimeter moment-frame office buildings built in a low-seismic region of the United States. Three-dimensional finite element models were constructed to model the buildings using shell elements and incorporating the steel deck and composite concrete slab floor system. Nonlinear material models were used along with simplified component models for beam and girder connections. Accurate structural and non-structural masses were used to capture realistic inertial effects. The models were then analyzed using the Abaqus/Explicit finite element analysis engine to simulate instantaneous structural loss of a single ground-level column. This analysis was carried out for twelve individual columns in the three-story building and four individual columns in the ten-story building. Analysis was conducted for a sufficient time following column loss to assess structural collapse or obtain the peak vertical displacement if collapse was arrested. The output was then post-processed to obtain stresses in the steel deck and concrete slab as well as resultant connection forces and load-redistribution behavior.
The three and ten-story building were found to be capable of arresting collapse following the loss of an individual ground-level supporting column for most column locations. Demands were the least severe for perimeter columns within a moment frame, but the structures were also able to bridge over lost interior columns that had no moment connectivity. Connection demands were significant in most column-loss scenarios and adequate moment connection strength and ductility was found to be necessary to ensure successful collapse arrest.

Unreinforced masonry (URM) structures represent a significant portion of the residential building stock of the Central and Eastern United States (CEUSA), accounting for 15% of homes in the 8-state region impacted by the New-Madrid Seismic Zone and an even greater portion of the building stock in most other regions of the world. In addition to significant population, the brittle nature of URM buildings further supports a thorough consideration of seismic response given the susceptibility to severe failure modes. Currently, there is a pressing need for analytically based fragility curves for URM buildings. In order to improve the estimation of damage state probabilities through the development of simulation-based masonry fragilities, an extensive literature survey is conducted on pushover analysis of URM structures. Using this data, capacity diagrams are generated, from which damage exceedance limit states are defined.

Demand is simulated using synthetically derived accelerograms representative of the CEUSA. Structural response is evaluated using an advanced capacity spectrum method developed at the Mid-America Earthquake Center. Capacity, demand, and response are thus derived analytically and response data is used to generate an improved and uniform set of fragility curves for use in loss-assessment software via a framework amenable to rapid expansion of pushover database and variation of ground motion records. A set of best practices is hereby developed for selection and use of experimental and analytical data, input ground motions, analysis of structural capacity, limit state definitions, seismic design categories, and probabilistic analysis methods. Curves are expressed in multiple forms for wide range of use in loss-assessment applications. Results are discussed and compared with other relationships developed in the literature, along with those generated using HAZUS opinion-based capacity data. The parameters of the improved fragility relationships developed and presented in this thesis are provided and suggested for reliable use in seismic loss assessment software.