Machine Learning-Optimized Alkali-Activated Recycled Aggregate Concrete: Mechanical performance and Life Cycle Assessment

Cement industry is one that encourages the need to develop sustainable solutions sooner in realizing its significant carbon footprint. But mechanical characteristics and environmental performance of alkali-activated recycled aggregate concrete (AARAC) have not been yet addressed comprehensively, especially with the inclusion of the supplementary cementitious materials sourced to industrial byproducts. The present research is an in-depth analysis of alkali-activated recycled aggregate concrete (AARAC) including industrial byproducts, in which both experimental and computational modeling is performed. Nine concrete formulations with different ratios of fly ash (FA), ground granulated blast furnace slag (GGBS), and silica fume (SF) as the main binders were tested in a systematic way to determine their rheological properties, mechanical result, and environmental performance based on life cycle assessment (LCA). Three machine learning models, namely Support Vector Regression (SVR), XGBoost (XGB) and LightGBM (LGBM) are applied to predict the compressive strength and optimized with particle swarm optimization (PSO), in such a way that SHAP can be used to understand the model. It was found that the addition of GGBS significantly improved the mechanical performances and 40% replacement of GGBS special concrete showed valued compressive strength of 120.36 MPa, which means 168% performance improvement when compared to normal OPC concrete. The model showed that the XGB had the best prediction capacity with R2 value of 0.905 with curing temperature and the coarse aggregate as key factors that influence development of strength. LCA output reveals how the AARAC allowed environmental benefits including the global warming potential that was reduced by 41% in balance with FA-dominant type mixes as compared to OPC structures. In this study, a forward, aggressive challenge was made to advance self-sustainable high-performance concrete construction by synergistically coupling state-of-the-art materials characterization and machine learning parameters optimization.