|
کلیدواژهها
|
Deep Beam, Shear Strength, ACI code, Strut and Tie Model, Finite Element Method, Artificial Neural Network, ABAQUS Programming, and WEKA Programming
|
|
چکیده
|
Reinforced concrete deep beams are crucial structural components widely used in civil infrastructure because of their capacity to support heavy loads over short spans. Unlike traditional slender beams, their behavior involves complex nonlinear stress distributions and direct strut mechanisms, requiring advanced analytical and computational tools for accurate shear strength evaluation. This thesis offers a comprehensive assessment of deep beam shear strength using several approaches: The Strut- and- Tie Model (STM), Finite Element Method (FEM) via ABAQUS, ACI, and Artificial Neural Networks (ANN) implemented in WEKA. The study comprises two parts. The first part begins with an extensive literature review, followed by a detailed experimental program testing twelve reinforced concrete deep beams with varying shear span-to-depth ratios, reinforcement ratios, and concrete strengths. Analytical calculations are based on STM and FEM simulations, accounting for material nonlinearities and post-cracking behavior. Results indicate that FEM provides highly accurate predictions, especially for small to medium-sized beams, with an average prediction ratio (FEM to experimental results) of 0.92 Additionally, STM performs reliably, with an average prediction ratio (STM to experimental results) of 0.96, though it occasionally produces unconservative estimates for larger beams, as four out of twelve predictions exceeded 1. The second part involves analyzing 233 beams from literature, considering parameters such as span-to-depth ratio, shear span-to-depth ratio, concrete compressive strength, reinforcement amounts, depth-to-width ratio, and longitudinal bar diameter. Both the ACI and Neural Network (WEKA) methods were applied. The neural network's prediction of shear strength closely matched experimental results, with a coefficient of determination of 0.838, while the ACI code prediction yielded a coefficient of 0.394. This thesis concludes with a proposal for an integrated predictive framework combining the interpretability of STM, the accuracy of FEM, and the adaptability of ANN. Recommendations are made for improving design codes and guiding future research
|