3-D finite element modeling of reinforced concrete beam-column connections: development and comparison to NCHRP Project 12-74

This report, 3-D Finite Element Modeling of Reinforced Concrete Beam-Column Connections – Development and Comparison to NCHRP 12-74, investigates the use of finite element modeling (FEM) to predict the structural response of the cast-in-place (CIP) reinforced concrete bent cap-column test specimen reported in NCHRP Report 681 – Development of a Precast Bent Cap System for Seismic Regions. Analysis was performed using LS-DYNA as the finite element processor. The Karagozian & Case Damaged Concrete model, material MAT_072, was used as the constitutive model for all concrete elements and material MAT_003, a plastic kinematic model, was used as the constitutive model for the reinforcing steel. Strain-hardening effects of steel were neglected for this analysis. Boundary conditions on the FE model were identical to the vertical and horizontal restraints used on the CIP specimen during testing. The FE model only considered a monotonic push loading sequence, whereas the CIP specimen was subjected to reverse cyclic loading. To account for the difference in loading, the FE model results were compared to the hysteretic envelope from the CIP specimen. The lateral load-lateral displacement response of the FE model (Model 1) compared reasonably well to the actual and theoretically predicted response of the CIP specimen. For lateral displacements less than that corresponding to a displacement ductility of 4.1, the FE model showed a larger stiffness than the actual CIP response. The model stiffness degraded as a greater number of concrete elements in the column plastic hinging region accumulated damage. The degradation and lateral load-displacement response matched the predicted response within 5% for a displacement ductility larger than 2.0; however, the model degradation was not as severe as that observed for the CIP specimen. Concrete damage in the FE model correlated reasonably well with observed cracking and spalling of the CIP specimen. Significant damage was observed in the column of the FE model, near the joint, reflecting flexural cracking. Initial yielding of column longitudinal bars in the FE model occurred at a displacement ductility 26% larger than the CIP specimen. Based on contours of concrete damage and principal stress vectors, the primary shear crack formed diagonally through the joint of the FE model at a lateral load 6% higher than that of the CIP specimen. Joint rotation for the FE model was significantly less than that of the CIP specimen, approximately half of the specimen values. Conclusions include: 1) finite element modeling using appropriate constitutive models and element formulation can accurately capture the nonlinear behavior of reinforced concrete beam-column connections, including flexural cracking, joint shear cracking, steel reinforcement yielding and overall stress distribution; 2) element size for concrete and steel reinforcement significantly impacts the overall response and accuracy of results and therefore must be carefully selected for convergence; 3) the Karagozian & Case damaged concrete model, material MAT_072, can accurately capture the cracking of concrete using limited inputs (f 'c and aggregate size). Recommendations include: 1) additional analysis should be performed to appropriately incorporate a strain hardening model for the reinforcing steel; 2) strain distribution of the steel reinforcement in the joint (longitudinal reinforcement, joint hoops, and joint stirrups) should be further investigated as well as the hoop strain distribution in the column plastic hinge region; 3) a concrete constitutive model capable of reverse cyclic loading should be investigated; 4) a bar slip model for bond between the concrete and reinforcing steel should be investigated.