Experimental and machine learning-based analysis of red mud influence on recycled aggregate concrete properties

Area of Science:

• Materials Science and Engineering
• Sustainable Construction Materials
• Concrete Technology

Background:

• Growing environmental concerns necessitate sustainable alternatives in concrete production.
• Red mud (RM), an industrial byproduct, and recycled concrete aggregate (RCA) offer potential for eco-friendly concrete.
• Understanding the mechanical and durability properties of RM-RCA concrete is crucial for its practical application.

Purpose of the Study:

• To experimentally investigate the compressive strength and acid resistance of concrete incorporating varying percentages of RM and RCA.
• To evaluate the effectiveness of machine learning (ML) models in predicting the compressive strength of RM-RCA concrete.
• To identify key parameters influencing ML model predictions using SHAP analysis.

Main Methods:

• Systematic replacement of cement with RM (0-15%) and natural coarse aggregate with RCA (0-100%).
• Experimental testing of compressive strength and acid resistance for all concrete mixes.
• Training and validation of four ensemble ML models (RF, KNN, AdaBoost, XGBoost) using 251 datasets.

Main Results:

• Compressive strength increased up to 10% RM replacement, with significant gains observed at higher RCA percentages.
• Acid resistance initially improved with RM addition up to 10% before declining.
• XGBoost demonstrated the highest accuracy in predicting concrete compressive strength, with curing age identified as the most influential parameter.

Conclusions:

• Incorporating up to 10% RM and significant amounts of RCA can enhance sustainable concrete properties.
• Machine learning, particularly XGBoost, provides a reliable method for predicting the performance of RM-RCA concrete.
• Curing age, RCA, silica fume (SF), and natural coarse aggregate (NCA) are critical factors for accurate strength prediction.