UC Berkeley

The California Department of Transportation (CalTrans) is urging researchers and contractors to develop the next generation highway bridge design. New design solutions should favor the use of modular construction techniques over conventional cast-in-place reinforced concrete in order to reduce the cost of the projects and the amount of constructions on site. Earthquake resistant bridges are designed such that the columns are monolithically connected to the girder and the foundations. Hence, despite the great improvements recently made in modular bridge construction, a large amount of concrete is still cast in place to properly splice the reinforcements between the segments.

Instead of designing earthquake resistant bridges with monolithic joints, it is proposed to use discontinuous connections in this thesis. The segments may rock at their interface during a severe earthquake and, if rocking rotation is too large, the structure may collapse. However, if a bridge is allowed to rock moderately, it may modify the earthquake response and drastically reduce the resisting forces within the structure.

The research presented in this dissertation focuses on the modeling of rocking connections. First, the behavior of rocking rigid blocks under earthquake excitation is studied. It is proposed to restrain the rigid blocks with an unbonded post-tensioning cable in order to allow rocking but prevent overturning. The findings made on rigid blocks, however, cannot be applied to deformable structures because of the limitations of the model. Therefore, a completely different approach is proposed. Instead of modeling the behavior of an entire block, it is proposed to model only the rocking surface. A zero-length finite element is developed, allowing to represent the in-plane rocking rotation between two frame elements. It allows to investigate the behavior of a deformable column rocking freely on its base as well as the stability of a rocking column restrained with a cable and subjected to a large earthquake excitation. The consequences of a post-tensioning cable failure and yielding of the column are also investigated. It is proposed to add a dissipative fuse between the base of the column and its footing in order to enhance the stability of the structure. Finally, the behavior of a conventional monolithic bridge is compared with a bridge allowed to rock at the columns joints.

The results obtained with the rigid block model show that, when columns are allowed to rock under earthquake excitation, it is possible to adjust the response and preserve stability with a Post-Tensioning (PT) cable. The implementation of the zero-length rocking element permits to study the behavior of deformable structures that are allowed to rock. At first, this element is used in combination with a very stiff elastic element and the results are consistent with the response of a rigid block. This element shows that the free rocking response of an elastic column may stop rocking and start to oscillate in flexure. The dissipative fuse in combination with a long unbonded PT cable proves to be effective. However, it is shown that, if the dissipative fuse is too large, it may prevent the column from returning to its initial position. At last, it is shown that a bridge structure allowed to rock and restrained with cables can sustain a large earthquake. The resisting moments within the columns are greatly reduced when compared with a conventional bridge while the drift ratio remains moderate.\

Several subjects are left for further research. First, the zero-length rocking element represents rocking only in the plane of the frame. The development of a 3D rocking element is challenging because the column may rock and roll and may also twist around one corner. The design of the rocking surface is not investigated; a bridge prototype should be designed and tested experimentally to validate the feasibility of the solution proposed in this study.