Machine Learning-Based Prediction of Load Carrying Capacity of Reinforced Concrete Elements

Reinforced Concrete (RC) elements such as beams and columns play a fundamental role in ensuring structural safety and stability in civil engineering structures. Accurate prediction of their load-carrying capacity is essential for safe and economical design. Traditional design approaches, based on codal provisions such as IS 456:2000 and ACI 318, rely on simplified analytical expressions derived from experimental studies. Although reliable, these methods may not fully capture the complex nonlinear interaction between material properties, geometric parameters, and reinforcement characteristics. This study presents a machine learning-based predictive framework for estimating the ultimate load-carrying capacity of reinforced concrete elements. A structured dataset comprising key structural parameters—including concrete compressive strength, steel yield strength, reinforcement ratio, section dimensions, and loading conditions—was compiled and preprocessed. Supervised learning algorithms, particularly Random Forest Regression, were implemented to model the nonlinear relationships between input parameters and structural capacity. The developed model was trained and validated using statistical performance indicators such as the coefficient of determination (R²), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). The predicted capacities were compared with traditional code-based calculations as per IS 456:2000 to evaluate consistency and reliability. The results demonstrate that the machine learning model provides accurate and efficient predictions, with strong correlation to codal values. A web-based application was developed to enable real-time capacity prediction and comparative analysis between machine learning and traditional methods. The proposed hybrid framework enhances structural design efficiency, reduces computational effort, and supports data-driven decision-making in reinforced concrete design.