CCEER-20-05: Experimental And Analytical Studies Of A Two-span Bridge System With Precast Elements Incorporating Rebar Hinge And Socket Connections

Report No. CCEER 20-05Accelerated bridge construction (ABC) has become an increasingly appealing alternative to conventional cast-in-place construction (CIP) because of the benefits it offers in reducing onsite construction time and traffic impact. Maintaining joint integrity between precast components during seismic events has been a design challenge faced by engineers. Several ABC connections have been developed and shown promise in providing joint behavior emblematic of CIP bridges. However, these projects have been limited to component tests subjected to uniaxial ground motions, leaving questions regarding the seismic performance of ABC connections when subjected to biaxial forces and system interaction as part of a bridge system. This gap in research was addressed by testing a 0.35 scale two-span ABC bridge model, Calt-Bridge 2, on the shake tables at the Earthquake Engineering Laboratory at the University of Nevada, Reno. Experimental and analytical studies were conducted to: (1) assess the performance of the ABC connections and bridge system when subjected to multiple bi- directional ground motions of varying acceleration level, (2) review the current design procedure for each connection type and revise the procedure based on findings from the experimental results to account for interaction within the bridge system or for bi-axial ground motions, (3) determine if the behavior of the bridge system under biaxial seismic loading can be captured using existing analytical modeling methods, (4) evaluate parameters for the scaled bridge model that were not tested during the shake table tests, and (5) assess the relative seismic performance of ABC connections in three bridge models (Calt-Bridge 2 and two other previously tested ABC bridges, Calt-Bridge 1 and ABC-UTC) and make recommendations based on relative connection performance. The following six ABC connections were implemented in Calt-Bridge 2: (1) a rebar hinge precast with the footing connected to the column via a pocket for the column-to-footing connection, (2) a fully precast pocket connection for the column-to-cap beam connection. (3) extended strands and headed bars enclosed in the cast-in-place portion of the cap beam for the girder-to-cap beam connection, (4) lap spliced straight bars embedded in ultra-high performance concrete (UHPC) for the deck connection over the pier, (5) precast deck panels to girder connection using deck pockets and projected steel studs from precast concrete girders, and (6) short embedment length lap spliced straight bars for female-to-female deck panel-to-panel connection. Eight bi-directional ground motions were applied to the bridge model ranging from 30% to 225% of the design level earthquake. The 1994 Northridge earthquake measured at Sylmar station was simulated as the input motion for the shake table tests. The seismic performance of the bridge model was emblematic of the behavior expected from conventional CIP bridges with ductile plastic hinges forming in the columns and capacity protected elements remaining essentially elastic. Joint integrity was maintained in the ABC connections during all earthquake runs providing satisfactory load path between joined elements. Large in-plane rotations of the superstructure were observed, which were attributed to unbalanced friction forces at the abutments. The performance of the ABC connections in Calt-Bridge 2 was compared against the connections from two other scaled ABC bridge models, Calt-Bridge 1 and ABC-UTC. Some deterioration of the grout in the rebar hinge pocket connections was observed, therefore it was recommended that socket connections be used for rebar hinge connections at the column base. Analytical studies were conducted using Opensees to evaluate ability of the model to capture the response of ABC bridge systems and connections. The calculated results were compared against the measured shake table test riesults to evaluate model accuracy. Base shear and column displacement histories were captured reasonably well in the longitudinal and transverse directions. In-plane rotation of the superstructure was not captured with roller supports at the abutments. Therefore, friction effects were incorporated at the abutments through a parametric study, which produced similar in-plane rotation response to the measured results when friction was modeled at one abutment. Overall, this study demonstrated that newly evolved ABC seismic design procedures may be utilized to expand application of ABC in moderate and high seismic regions with confidence