Capacity Spectrum Seismic Design Methodology for Bridges with Hybrid Sliding-Rocking Columns

Bridges with hybrid sliding-rocking (HSR) columns offer an attractive alternative for accelerated construction in seismic regions. HSR columns include internal unbonded post-tensioning, end rocking joints, and intermediate sliding joints over the column length. As such, HSR columns introduce new design variables, such as the number and distribution of sliding joints, the coefficient of friction at the sliding joints, and the joint-sliding amplitude, compared to conventional cast-in-place systems. Force-based seismic design methodologies are difficult to apply because of the lack of response modification factors—R-factors—accounting for the various combinations of these design variables and the corresponding nonlinear responses. To address this challenge, this paper introduces a capacity spectrum seismic design (CSSD) methodology for bridges with HSR columns that holistically accounts for all design variables and directly utilizes the resulting nonlinear system properties. A key component of the proposed CSSD methodology is the development of a simplified model for the monotonic and cyclic quasi-static pushover analysis of HSR columns. The simplified analysis model was validated through experimental data, and was further used to explore the effect of major design variables on the response of HSR columns. Furthermore, this paper explores the existence of multiple performance points (i.e., multiple solutions) predicted by the capacity spectrum method for given ground motions and provides a tracking algorithm to identify them. The proposed CSSD methodology was validated through both experimental data and computational studies. The simplified analysis model was capable of capturing the fundamental response mechanisms of HSR columns. The proposed CSSD methodology provided reasonably accurate predictions of peak displacements and base shear. Larger discrepancies were observed in the comparisons with experimental data (by overpredicting the response by 25%) due to the existing damage to the test specimen from prior testing, which could not be accurately captured by the simplified model, and because the simplified model does not account for variations of the vertical load due to vertical excitation.