Strategies for Inclusion of Uncertainties in Modeling Techniques for Enhancement of Capabilities of Pushover Analysis

Abstract

Pushover analysis is an extensively used tool for performance evaluation of structures under seismic loads. Continuous efforts are on for enhancement of capabilities of the same. Refinements to geometric and material modeling have led to better understanding of structural performance. Notwithstanding the attempts, though analytical predictions of strengths have been in close agreement with experimental, displacements predicted have differences. Attempts to close this gap between predicted and observed displacement characteristics have always been centered around geometric and material modeling. The sequence of plastic hinge formation and its influence on displacement characteristics needs very serious consideration. The present study highlights this issue with illustrations by suggesting strategies to reduce tremendous computational efforts, required for considering plastic hinge formation sequence in performance appraisal. Strategy 1 considers 15% of potential plastic hinge locations and allows variations leading to different sequences. Two sub strategies have been proposed by way of allocation of defective plastic hinge locations to horizontal and vertical planes. When hinges with uncertainties are restricted to horizontal planes, variations in base shear values are between 1.4% to about 1.7% in comparison with experimental study. Whereas displacements are lower by 10%. When hinges with uncertainties are restricted to vertical planes, the difference in both base shear and displacements show the range of variations of 7% and 15% with respect to experimental results suggesting assignment of plastic hinges with uncertainties distributed in horizontal planes for better results. Strategy 2 adopts randomization of plastic hinge locations with uncertainties associated, distributed throughout the 3D frame. The analysis results indicate such a consideration is superior then strategy 1 to get analytical results almost perfectly matching with experimental results. Strategy 1, finds application where defect and deficiency features are known a priori. Whereas strategy 2 can be employed in all situations.