Purpose: Rapid industrialization and construction generate substantial concrete waste, leading to significant environmental issues. Nearly 10 billion metric tonnes of concrete waste are produced globally per year. In addition, concrete also accelerates the consumption of natural resources, leading to the depletion of these natural resources. Therefore, this study uses artificial intelligence (AI) to examine the utilization of recycled concrete aggregate (RCA) in concrete.Design/methodology/approach: An extensive database of 583 data points are collected from the literature for predictive modeling. Four machine learning algorithms, namely artificial neural network (ANN), random forest (RF), ridge regression (RR) and least adjacent shrinkage and selection operator (LASSO) regression (LR), in predicting simultaneously concrete compressive and tensile strength were evaluated. The dataset contains 10 independent variables and two dependent variables. Statistical parameters, including coefficient of determination (R²), mean square error (MSE), mean absolute error (MAE) and root mean square error (RMSE), were employed to assess the accuracy of the algorithms. In addition, K-fold cross-validation was employed to validate the obtained results, and SHapley Additive exPlanations (SHAP) analysis was applied to identify the most sensitive parameters out of the 10 input parameters.Findings: The results indicate that the RF prediction model performance is better and more satisfactory than other algorithms. Furthermore, the ANN algorithm ranks as the second most accurate algorithm. However, RR and LR exhibit poor findings with low accuracy. K-fold cross-validation was successfully applied to validate the obtained results and SHAP analysis indicates that cement content and recycled aggregate percentages are the effective input parameter. Therefore, special attention should be given to sensitive parameters to enhance the concrete performance.Originality/value: This study uniquely applies AI to optimize the use of RCA in concrete production. By evaluating four machine learning algorithms, ANN, RF, RR and LR on a comprehensive dataset, this study identities the most effective predictive models for concrete compressive and tensile strength. The use of SHAP analysis to determine key input parameters and K-fold cross-validation for result validation adds to the study robustness. The findings highlight the superior performance of the RF model and provide actionable insights into enhancing concrete performance with RCA, contributing to sustainable construction practice.

In this study, a comprehensive dataset containing 358 data points was collected from the literature, focusing on the compressive strength, split tensile strength, and modulus of elasticity of concrete made with recycled concrete aggregate (RCA). An Artificial Neural Network was used machine to predict mechanical properties of RCA concrete. Furthermore, K-fold cross validation was utilized to validate the model’s reliability, and sensitivity analysis was performed to identify the most influential input parameters among the independent variables. The model demonstrated strong performance during training, achieving R² values of 0.93 for compressive strength, 0.92 for split tensile strength, and 0.99 for modulus of elasticity with corresponding RMSE of 2.55, 3.85, and 0.37, respectively. The MAE and MAPE values during training were 0.68 and 0.03 for compressive strength, 0.71 and 0.03 for split tensile strength, and 0.08 and 0.01 for modulus of elasticity, respectively. Testing results revealed R² values of 0.75 for compressive strength, 0.78 for split tensile strength, and 0.67 for modulus of elasticity, with RMSE values of 8.57, 5.03, and 3.83, respectively. Moreover, the sensitivity analysis indicated that the cement percentage and water-to-cement ratio were the main input parameters which significantly influence RCA concrete strength.

The production of cement, which is the key ingredient of concrete, leads to environmental pollution by releasing massive amounts of CO₂ and using significant natural resources. Therefore, shifting towards sustainable and greener materials is essential for mitigating these challenges. In this study, recycled concrete powder (RCP) was used as a cement replacement (0%, 5.0%, 10%, and 15%), solving the waste dumps issue and promoting sustainability. Furthermore, the concrete is also reinforced with steel fibers which were obtained from waste rubber tires to improve concrete tensile strength. The concrete properties were evaluated through slump cone test, compressive strength, failure patterns, tensile strength, scanning electronic microscopy, and FTIR analysis. The results indicate that the concrete strength properties improved with the substitution of RCP. The compressive and tensile strength of the optimum mix (10% RCP and 2.0% addition of steel fibers) are 15.8% and 23% more than those of reference concrete. However, the concrete flow is adversely impacted due to RCP angular particle shapes. Failure patterns indicate that RCP and steel fibers improved concrete ductility. SEM and FTIR analysis indicate microstructural improvement with RCP and steel fibers. Finally, the analysis concluded that the developed concrete showed better performance, solved waste dumps issues, and promoted sustainability.

A machine learning based hybrid modelling and experimental study is presented to advance the design of sustainable engineering cementitious composites (ECC) incorporating recycled constituents. An experimental program was first carried out on ECC mixtures incorporating recycled concrete powder (RCP) as a partial cement replacement and waste tire steel fiber (WTSF) as a reinforcement alternative, as a result indicated a peak compressive strength of 128.5 MPa. After that, a proposed hybrid AI model and benchmark algorithms such as, Artificial Neural Networks, Random Forest, and Gradient Boosting, was trained using 403 experimental data points to predict compressive strengths. The hybrid AI model achieved the highest predictive performance, with coefficient of determination values of 0.96 during training and 0.87 during testing. Overall, the combined experimental and modeling findings demonstrate that the proposed hybrid AI approach can reliably predict compressive performance and accelerate the development of low-carbon cementitious composite.

Efforts to reduce concrete’s embodied carbon and the environmental impacts of construction-and-demolition (C&D) waste have increasingly shifted toward using recycled-aggregate concrete (RAC) to replace natural aggregate, either partially or entirely. Over the past two decades, dozens of studies have been conducted on the properties of RAC mixes. These studies provide a solid foundation for prediction models that eliminate the reliance on actual tests, which can be time-consuming. Machine learning (ML) models predict concrete properties from interacting variables, such as water-to-cement ratio, cement content, RAC replacement level, aggregate density, etc. These data-driven predictions are more reliable than traditional empirical formulas. ML model can capture nonlinear relationships among diverse input features, providing a framework for optimizing mix design and evaluating mechanical performance. In this study, a database of 358 mix designs from the literature was used to predict three key mechanical properties (compressive strength, split-tensile strength, and modulus of elasticity) of RAC-based concrete. The decision tree was used as the principal predictive model, and its robustness was verified through K-fold cross-validation. Model performance was high in training (R² = 0.93, 0.92, and 0.94, respectively) and remained acceptable in testing (R² = 0.75, 0.78, and 0.76). SHAP (SHapley Additive exPlanations) Analysis shows cement content and aggregate density are the most influential features across all three properties, followed by water-to-cement ratio and RAC replacement level. This interpretable ML framework streamlines mix-design optimization, reduces laboratory work, and guides production of low-carbon RAC with dependable mechanical performance.