Artificial Neural Networks for Predicting Mechanical Properties of Reinforced Concrete: A Comparative Study with Experimental Data

Abstract

This study explores the use of Artificial Neural Networks (ANNs) for predicting key mechanical properties of reinforced concrete, including compressive, tensile, and flexural strength, based on mix parameters such as water-cement ratio, aggregate content, cement type, and curing time. Traditional testing methods are time-intensive and costly, underscoring the need for predictive models that offer rapid, accurate insights into concrete performance. Here, an ANN model was developed and trained using experimental data, achieving strong correlation with physical testing results and capturing the nonlinear interactions between input parameters. Sensitivity analysis identified the water-cement ratio, cement content, and aggregate proportions as critical factors influencing strength, with the ANN demonstrating high sensitivity to these variables. Comparisons with traditional methods highlight the ANN model’s advantages in speed, cost-efficiency, and predictive accuracy, making it a practical tool for construction quality control. This study suggests that ANN models can be integrated into the construction workflow for quick, data-driven decision-making in mix design adjustments. Future work could expand the model's applicability by incorporating a wider range of concrete types and exploring hybrid machine learning approaches to further enhance accuracy and generalizability in diverse construction applications.