Innovative sustainable concrete with waste glass materials: an explainable machine learning for compressive strength...

Area of Science:

• Materials Science
• Civil Engineering
• Environmental Science

Background:

• Sustainable concrete utilizes waste glass to mitigate cement production impacts and aggregate depletion.
• Accurate compressive strength (CS) prediction is vital for optimizing sustainable concrete mixtures and ensuring structural integrity.

Purpose of the Study:

• To develop and optimize machine learning models for predicting the CS of concrete containing glass powder (GP) and glass sand (GS).
• To identify key parameters influencing CS and provide a practical decision-support tool for sustainable concrete design.

Main Methods:

• Five machine learning models (RF, KNN, AdaBoost, LightGBM, XGBoost) were trained and optimized using Grid Search on a dataset of 270 experimental samples.
• Input parameters included curing duration, cement content, GP, GS, water, density, sand, and basalt.
• SHAP and Partial Dependence Plot (PDP) analyses were used for parameter influence assessment.

Main Results:

• LightGBM achieved the highest predictive performance (R² = 0.964, RMSE = 2.05 MPa), outperforming XGBoost and RF.
• KNN and AdaBoost showed lower predictive accuracy.
• Curing duration and cement content were identified as positively influencing CS, while water and GP showed negative effects.

Conclusions:

• The optimized LightGBM model provides accurate and interpretable predictions for sustainable concrete CS.
• A user-friendly GUI was developed for the LightGBM model, enhancing its practical applicability in concrete mixture design.
• This tool supports the development of reliable and sustainable concrete.