Derivation of an Empirical Equation for Estimating the Ultimate Moment Capacity of Reinforced Concrete Beams Using Artificial Neural Networks (ANN) Based on ACI-318 Code Data

Abstract

This project aims to utilize Artificial Neural Networks (ANNs) to derive an empirical equation for calculating the flexural moment of reinforced concrete beams. Due to the importance of accurate flexural moment estimation in structural design, the American Concrete Institute (ACI) code equations were adopted as a reference to validate and assess the accuracy of the proposed model. A database consisting of eighty (60) reinforced concrete beams was developed, covering a wide range of geometric and mechanical properties, including beam width, effective depth, concrete compressive strength, steel yield strength, and reinforcement area. These data were used to train and test several ANN models in order to identify the most accurate and stable configuration. The results demonstrated a strong agreement between the flexural moments predicted by the selected ANN model and those calculated using the ACI code equations, as indicated by high values of the coefficient of determination (R²) and low values of the mean absolute error (MAE). Based on the final ANN model, a logarithmic empirical equation was extracted, enabling direct calculation of the flexural moment without the need to run the neural network model. The findings of this study confirm that artificial neural networks are an effective and reliable tool for modeling the flexural behavior of reinforced concrete beams. Moreover, the derived empirical equation can serve as a practical and simplified method for engineering calculations and design verification within the range of data used in this study.