Optimization of recycled rubber self-compacting concrete: Experimental findings and machine learning-based evaluation

Area of Science:

• Materials Science
• Civil Engineering
• Sustainable Construction

Background:

• Standard concrete modeling faces challenges with nonlinear and environmentally sensitive properties.
• Waste tire rubber aggregates (WRTA) offer a potential sustainable alternative for coarse aggregates in concrete.
• Self-compacting concrete (SCC) requires precise characterization for optimal performance.

Purpose of the Study:

• To evaluate the rheological and mechanical properties of SCC with varying percentages of WRTA as a partial replacement for coarse aggregates.
• To determine the optimal WRTA substitution level for balancing performance and waste management.
• To compare the predictive accuracy of linear regression (LR) and extreme gradient boosting (XGBoost) for rubberized SCC characteristics.

Main Methods:

• Incorporated WRTA at 0%, 5%, 10%, and 20% as a substitute for coarse aggregates in SCC.
• Assessed fresh properties using slump flow, J-ring, and V-funnel tests.
• Evaluated hardened properties through compressive and splitting tensile strength tests.
• Utilized LR and XGBoost machine learning models for property prediction.

Main Results:

• Increased WRTA content led to a decrease in workability and hardened properties.
• A 5% WRTA substitution was identified as optimal for reducing environmental impact and managing waste.
• A 10% WRTA substitution resulted in a 34% decrease in compressive strength and 28% in tensile strength after 28 days.
• XGBoost outperformed LR, achieving higher R² values for property prediction.

Conclusions:

• A 5% WRTA substitution is feasible for producing self-compacting rubberized concrete (SCRC) with improved sustainability.
• While a 10% substitution is possible, it significantly reduces mechanical strength.
• XGBoost is an effective machine learning tool for accurately predicting the behavior of rubberized SCC.