Lee, Judong (2020-11). Development of a Truss-Arch Model Unified for Flexure and Shear-Critical Interaction of Structural Concrete Members. Doctoral Dissertation.
Thesis
Based on force-deformation behavior computed using the Compatibility Strut and Tie Model (C STM), a theoretical limit analysis model is developed to predict the ultimate shear-carrying capacity of shear-critical structural concrete beams with and without transverse reinforcement. The Truss-Arch Model Unified (TAMU) is validated based on experimental observations and from previous studies reported in the literature. The limit analysis approach assumes the failure mechanism occurs when the principal diagonal arch reaches its softened peak strength. Two truss models, with a vertical tie or an inclined tie, are used depending on whether the members are reinforced with or without shear steel, respectively. Explicit solutions for the principal strain ratio and ultimate shear strength are derived based on truss and arch contribution to flexibility and strength. A large database of test data consisting of 839 beams is assembled for the substantiation of the proposed method and also to conduct a comparative assessment of existing code-based shear analysis methods. The proposed TAMU approach predicts well the limit load capacities against the database. When compared with existing code-based shear analysis approaches, the TAMU based approach demonstrates superior accuracy with less dispersion due to modeling and aleatory uncertainty in both D-and B-regions. The TAMU approach is further extended to account for axial strain effects which may be present due to the presence of either prestress, axial load, or both. A full-scale experimental program is conducted on reinforced and prestressed concrete bent caps to investigate the effect of prestressing force, shear reinforcement spacing, interior voids, and axial loads on the bent cap. The specimens are analyzed by the TAMU method and other existing code-based analysis methods. The C-STM and TAMU analysis methods are able to accurately predict the experimental test outcomes somewhat better than existing code analysis approaches.
