Nonlinear seismic response of asymmetric structures

The purpose of this study was to investigate the nonlinear, inelastic response of one-story, symmetric- and asymmetric-plan structures to uniaxial and biaxial lateral earthquake ground motions. The investigation is a combined experimental and analytical program. Through the study, the lateral-torsional response of the system was studied for a range of system parameters with the goals of examining the adequacy of current building code torsional design assumptions and the ability of analytical software to predict inelastic response.; The experimental part of the investigation involved subjecting a single-story steel moment-frame to a series of uniaxial and biaxial earthquake ground motions on the U.S. Army CERL shaketable. The structure used in this study was a rigid diaphragm, approximately eight feet on a side, supported by four steel pipe columns, five feet in length. Sixteen steel masses were attached to the structure in various configurations in order to provide dead load and mass asymmetry. In addition, different types of pipe columns and steel strap braces were used to create eight different configurations of mass, strength, and stiffness eccentricity.; The analytical part of the investigation involved the nonlinear finite element programs Abaqus and Drain-3DX to analyze the response of the model during the earthquake simulations. Analytical models were created using the experimentally measured dynamic properties of the test structure and material properties of the steel columns. The elastic and inelastic response of the structure during each configuration was then predicted using both Abaqus and Drain-3DX. Subsequently, analytical models were created based on typical design assumptions, without the benefit of experimentally measured structural properties. The inelastic response of the structure was then predicted using Drain-3DX.; Implications of the experimental results on the adequacy of seismic design provisions are discussed. In addition, the ability of analytical software to predict inelastic torsional response is discussed for both a model tuned to the measured dynamic properties of the actual structure and a model based on common modeling assumptions. Further, the effects on prediction accuracy of different analytical modeling parameters and assumptions are discussed.