Nonlinear dynamic v elastic analysis for seismic deformation demands in concrete bridges having deck integral with the piers

Abstract

Seismic deformations in concrete bridges having fairly long continuous deck integral with the piers and restrained transversely at the abutments but free to slide there longitudinally, are estimated via nonlinear dynamic analysis and compared to the corresponding outcomes of modal response spectrum analysis with 5% damping. Eight bridge configurations are studied, all with a box girder deck prestressed with bonded tendons, having three or five spans and piers of various cross-sections and about equal or very different heights. Two versions of each bridge are considered: one of conventional, force-based design and another having much less overstrength in the piers and developing less inelastic action in the deck. The piers are taken in the analysis with effective elastic stiffness equal to the secant stiffness to the yield point of their end section(s). The deck is discretized longitudinally into a string of nonlinear elements, whose moment-rotation relations are derived from moment-curvature analysis of the section in the two main directions of bending. Seven spectrum-compatible motions with Peak Ground Acceleration (PGA) of 0.25, 0.35 and 0.45 g on rock are applied in the longitudinal or transverse direction. Modal response spectrum analysis with 5% damping gives on average good—or at least safe-sided—predictions of inelastic deformation demands at the deck and the piers of regular bridges, especially under longitudinal earthquake. However, it underestimates inelastic deformation demands in bridges having piers of very different height, even when design measures are taken to harmonize stiffness across the piers. The stronger the ground motion, the larger are on average the elastic predictions of deformations relative to those from nonlinear dynamic analysis.