Optimizing flexural strength of RC beams with recycled aggregates and CFRP using machine learning models

Area of Science:

• Civil Engineering
• Materials Science
• Computational Mechanics

Background:

• Structural performance of concrete beams is crucial for infrastructure.
• Sustainable construction materials like recycled aggregates and CFRP offer potential benefits.
• Understanding the flexural behavior of beams with these materials requires advanced analysis.

Purpose of the Study:

• To investigate the flexural bearing behavior of reinforced concrete beams incorporating recycled aggregates and CFRP.
• To develop and evaluate advanced machine learning predictive models for structural performance.
• To quantify the impact of sustainable materials on beam strength and load-bearing capacity.

Main Methods:

• Experimental testing of eight concrete beams with varying material compositions.
• Development of machine learning models (Random Forest Regressor, XGBoost, LightGBM) using experimental data.
• Data preprocessing, feature selection, hyperparameter tuning via Pareto optimization, and model evaluation using MSE, MAE, and R².

Main Results:

• Optimized machine learning models demonstrated excellent predictive performance.
• Beams with 70% recycled aggregate and 10% silica fume showed a 53.03% increase in compressive strength.
• These beams also exhibited a 7% increase in load-bearing capacity compared to control beams.

Conclusions:

• Integrating recycled aggregates and CFRP significantly enhances concrete beam performance.
• Advanced machine learning models effectively predict the structural behavior of beams with sustainable materials.
• This research contributes to the development of eco-friendly and high-performance construction materials.