Application of Machine Learning Techniques to Predict Rheology and Compressive Strength of Steel Fibre Reinforced Concrete

Steel fiber reinforced concrete (SFRC) is increasingly employed in civil engineering projects, including the construction of buildings, bridges, and rigid pavements. This growing interest is attributed to the notable advantages SFRC offers in terms of strength and durability, compared to traditional concrete and asphalt pavements. Key factors like the rheological properties, exemplified by the concrete slump, and strength characteristics, such as compressive strength, are crucial in both the design and practical application phases of these projects. Rheological and strength properties of SFRC depend mainly upon the fiber type, fiber geometry while compressive strength of normal concrete basically governs by its water to cement ratio and aggregate characteristics. Having all these various parameters makes the design and planning of SFRC structures complicated for engineers due to the dependency of these aspects towards the final output. In this context, machine learning (ML) techniques have gain huge attention due to their capability in predicting and forecasting interdependency of different variables accurately. In this study, machine learning (ML) technique named artificial neural networks (ANN) is used to predict the rheological and strength properties of SFRC. It was evident that ANN when properly optimized is an effective and robust tool to predict rheological and strength properties of SFRC. These predictions allow planners and engineers to predict performance for civil designs and compromise cost substantially while saving a lot of time and effort.