Artificial neural network modeling to predict compressive strength and static modulus for self-compacting concrete...

This study uses recycled concrete aggregate (RCA) available in Tripura, India, as a coarse aggregate for self-compacting concrete (SCC). The region suffers from the scarcity of normal stone aggregate (NSA) and using RCA decreases the demand-supply gap and promotes sustainability. The specimens were cast using different percentages of RCA (0, 25, 50, 75, and 100%) in replacement of NSA. The study compares different models and their abilities to forecast concrete's compressive strength and static modulus using both artificial neural networks (ANN) and normal linear regression. The developed model incorporated age (14, 28, and 90 days), grade of concrete (M25, M30, and M35), and type of aggregate (RCA and NSA) into consideration. This study shows the effect of using RCA on the workability (slump test, J-ring test and L box test) property of SCC. The study also shows that incorporating the initial sorptivity coefficient in SonReb (ultrasonic pulse velocity and rebound number) models, and the use of ANN increases the prediction of both compressive strength and static modulus of SCC. The test results demonstrate that ANN-based models significantly enhance the prediction of compressive strength and static modulus compared to traditional models. Incorporating RCA in SCC maintains workability within acceptable limits while promoting sustainability by reducing construction waste and lowering carbon emissions. The study also presents the reduction in carbon emission in Tripura, India, due to the application of recycled concrete as a coarse aggregate. These insights provide a strong basis for adopting RCA in practical construction applications, addressing both environmental and material scarcity challenges.