Knowledge of water and sediment flow characteristics in tunnels is essential for structural safety and post-disaster rescue operations. In this research, we developed a numerical model based on the Euler–Euler approach to simulate the three-phase flow of air, water, and sediment, aiming to predict the spatiotemporal flow characteristics in tunnels following the inrush. The model assumes that air, water, and sediment constitute a continuous medium through each other. The interaction drag between phases was taken into account. A standard k-ε turbulence model was used to simulate the irregular fluid flow. The spatiotemporal fluid behavior within the tunnel was investigated by analyzing the volume fraction, fluid velocity, and fluid pressure. In addition, the effects of the support condition, the machinery obstruction, and the curve radius on the three-phase flow characteristics were discussed. Numerical simulation results indicate that in the spatial dimension, the volume fractions, velocity, and pressure all display gradient distribution; in the temporal dimension, these three parameters all show phased features. Moreover, fluid behavior is greatly influenced by the support condition, the machinery obstruction, and the curve radius. The results establish a fundamental basis for additional investigations concerning spatiotemporal characteristics prediction of water and sediment inrush.

Accurate prediction of tunnel water inflow (WI) is critical for preventing construction hazards and supporting engineering decision-making. This study proposes an enhanced stacking machine learning framework to achieve reliable WI prediction. SMOGN technology was used to address the issue of imbalanced data distribution. On this basis, the Bayesian optimization tool Optuna was utilized for automated hyperparameter tuning of various base learners, including elastic net (Enet), multilayer perceptron (MLP), support vector regression (SVR), and extreme gradient boosting (XGBoost). Base learner outputs were combined with original features to form augmented inputs, with ridge regression as the meta-learner. Gaussian process regression (GPR) modeled residuals for uncertainty quantification, while the Sobol method assessed feature importance. Results show that feature augmentation and residual modeling improve prediction performance. The proposed stacking model outperforms individual base models and achieves state-of-the-art results. From an engineering perspective, the framework can be used to forecast potential high-inflow zones, guide advance grouting or drainage measures.

Temperature-dependent mechanical behavior and wetting-induced swelling affect the stability of underground structures in expansive mudstone. Three modified constitutive models were evaluated against temperature-controlled triaxial compression and laterally constrained swelling tests. The Mohr–Coulomb strain-softening and Hoek–Brown-based unified hardening–softening models incorporated temperature effects and equivalent swelling strains. The unsaturated unified hardening model was extended through equivalent temperature–stress mapping. Mechanical calibration used data from 12 triaxial conditions, supplemented by evaluation at 40 °C after recalibration at 20 and 60 °C. Swelling parameters calibrated at 20 °C were held constant at higher temperatures. For the 12 calibration conditions, the Mohr–Coulomb model yielded the lowest errors in peak deviatoric stress and deviatoric stress at 5% axial strain. The hardening–softening model improved pre-peak response and peak-strain estimates relative to Mohr–Coulomb. The unsaturated model achieved the lowest pre-peak normalized root-mean-square error of 10.10% and a swelling mean absolute error of 0.22 percentage points across 40–80 °C, but overestimated deviatoric stress at 5% axial strain. The sensitivity results support a staged calibration procedure involving initial parameter estimation followed by joint refinement using the designated calibration data. These findings provide a basis for model selection and parameter calibration in analyses of temperature-dependent mechanical and swelling responses of expansive mudstone.