Predicting saturated and unsaturated hydraulic conductivity of bentonite with data-driven and physics-guided machine learning models

Abstract Accurate prediction of saturated and unsaturated hydraulic conductivity of bentonite is critical for assessing the long-term safety of high-level radioactive waste repositories, where barrier efficiency depends on coupled processes. This study develops a data-driven machine learning model to predict saturated hydraulic conductivity and a physics-guided machine learning model to predict unsaturated hydraulic conductivity of bentonite. For the saturated hydraulic conductivity prediction, a dataset of 215 experimental measurements was compiled, incorporating key soil properties such as montmorillonite content, specific gravity, liquid limit, plastic limit, initial water content, dry density, and temperature. To predict unsaturated hydraulic conductivity, the study integrates experimental data, synthetic data generated using the Van Genuchten model, and outputs from the developed machine learning model for saturated hydraulic conductivity. This dataset includes specific gravity, montmorillonite content, initial dry density, initial water content, initial void ratio, plasticity index, and suction. AdaBoost, CatBoost, and XGBoost algorithms were used to train the machine learning models, while the Whale Optimization Algorithm was applied for hyperparameter tuning. The developed models are publicly available via web-based interfaces for estimating saturated and unsaturated hydraulic conductivity of bentonite from soil parameters.