Optimizing Punching Shear Strength Assessment in CFRP-Reinforced Concrete Slabs Through Machine Learning

Tensile behavior and punching shear are prime hindrances for steel-reinforced slabs with heavy loading. To alleviate those properties, fiber-reinforced polymer (FRP) bars (made from glass, carbon, and basalt) are essential and can be used as an efficient reinforcement material in slabs. However, estimation of punching shear, which helps to understand shear deformation, is a challenging task. This paper presents prediction models using machine learning techniques to estimate the punching shear capacity of carbon fiber-reinforced polymer (CFRP) reinforced concrete slabs. For developing the prediction model, input variables such as column area, punching perimeter, effective depth, elasticity module of the bars, compressive strength of the concrete, and FRP reinforcement ratio were considered to predict the output variable punching shear capacity. The prediction results found that the machine learning techniques used in this study such as artificial neural networks (ANN) and multilinear regression (MLR), both performed well with more than 95% accuracy with less scatter. In this sense, the novel prediction models could be helpful as the best tool to predict shear behavior for structural engineers.