Conventional methods for testing non-structural concrete properties, particularly compressive strength, are often hindered by inefficiencies, variability, and time-consuming processes. These approaches require extensive manual labor and physical testing, contributing to significant resource consumption and waste. In response to growing demands for more sustainable construction practices, this study introduces advanced machine learning models to predict concrete properties with greater accuracy and efficiency. By employing Decision Trees, Artificial Neural Networks, and Random Forest Regressors, predictive models were trained on a large, diverse dataset of concrete mixes. These models optimize the prediction of compressive strength, reducing the time and cost associated with traditional testing techniques while improving the precision of predictions. Furthermore, the integration of machine learning reduces material waste by predicting optimal concrete compositions, minimizing the use of raw materials like cement, a major contributor to global CO₂ emissions. A web-based tool was developed to allow civil engineers and construction professionals to input concrete compositions and receive real-time strength predictions, ensuring better quality control, resource efficiency, and environmental sustainability in construction projects. This innovative approach highlights the potential of machine learning not only to enhance the accuracy and speed of concrete testing but also to reduce the environmental impact of construction activities. Ultimately, this project contributes to more sustainable and cost-effective solutions for the construction industry.

Keywords

Machine Learning Models, Concrete Compressive Strength, Decision Trees, Artificial Neural Networks (ANNs), Sustainability.