Early-age stabilization of kaolin soil using fly ash and nanographene: a hybrid experimental and machine learning investigation

Abstract

Soil stabilization remains a critical challenge in geotechnical engineering, particularly in achieving sustainable and high-performance improvements for problematic clays. This study investigates the synergistic effects of high fly ash (FA) and low carbon-based nanographene (NG) additives on the geotechnical and microstructural properties of kaolin clay. Beyond conventional stabilization approaches, the novelty of this work lies in integrating sustainable industrial by-products (FA) with advanced nanomaterials (NG) to enhance soil performance. Experimental analyzes were performed to evaluate changes in unconfined compressive strength (UCS), physical properties, consolidation, and microstructural behavior. The innovative aspect of this study is the identification of optimum FA-NG combinations that significantly improve short-term mechanical behavior while promoting environmentally friendly soil stabilization. Furthermore, machine learning models, including multiple linear regression (MLR), support vector regression (SVR), decision tree regression, random forest regression, and extreme gradient boosting (XGBoost), were used to predict UCS. The integration of interpretable AI techniques (SHAP and PDP) provides a novel framework for understanding the contribution of each parameter to soil strength. Results revealed that optimized FA-NG mixtures not only enhance mechanical performance but also demonstrate the potential of combining sustainable materials with machine learning to establish innovative methodologies in geotechnical engineering. This dual contribution (material innovation and data-driven modeling) represents the main achievement of the study and offers a new perspective for future soil stabilization practices.