Aiming at the non-isothermal seepage phenomena in fractured rock mass, this paper conducts nonlinear dynamic research on the coupled seepage problem. Based on solid–fluid heat conduction energy equations and the mutual coupling of temperature and seepage fields, the non-isothermal seepage constitutive relation of fractured rock is derived, and a one-dimensional nonlinear dynamic governing model is established. Theoretical analysis indicates the equilibrium solution of non-isothermal seepage is more complex than that under the isothermal condition. Numerical calculations reveal that temperature variation shifts equilibrium positions and alters the occurrence conditions of hysteresis bifurcation, verifying temperature as a core regulatory factor for seepage dynamic responses. Successive sub-relaxation iteration stability analysis demonstrates obvious differentiated convergence speeds: the seepage field converges markedly faster than the temperature field when the coupled system reaches steady state. Compared with the isothermal seepage, the temperature effect changes the location of abrupt transition points and critical threshold of control parameters, rendering fractured rock seepage systems easier to trigger abrupt structural mutation even at low rock fragmentation degrees. This study clarifies the internal nonlinear dynamic mechanism of thermal–fluid coupled seepage, identifies potential mutation risks in petroleum exploitation and geothermal development, and supplies essential theoretical support for related engineering applications.

This paper investigates the dynamic behavior of non-Darcy seepage systems in post-failure rock. A one-dimensional non-Darcy seepage evolution equation is established, and a 6-dimensional nonlinear ordinary differential system is derived via the spectral truncation method. Eigenvalue analysis is adopted to determine the instability and bifurcation conditions, with the bifurcation diagram plotted. The fourth-order Runge–Kutta method is used to obtain phase trajectory patterns under different initial values. The results confirm the existence of transcritical bifurcations and fold bifurcations. The dynamic response of the system is discontinuous with control parameters, and phase trajectory symmetry breaking occurs with the increase in nonlinear terms. The reduced-order model shows diverse phase trajectories including equilibrium, periodic, chaotic attractors and unstable states. The system is sensitive to initial values, which significantly affect phase trajectory behaviors. The system may lose stability and trigger water inrush hazards under critical conditions. The bifurcation diagram and critical parameters obtained can provide a theoretical basis for the early warning, risk assessment and prevention of coal mine water inrush hazards.

Kongzhuang coal mine of Shanghai Datun Energy Resources Co., Ltd. is a typical deep coal mine in eastern China. The mining disturbance of working face in deep coal mine leads to the fracture and movement of overlying strata and the damage of the aquifer in overlying strata, thereby causing water inrush disaster and posing a serious threat to the safety production in coal mine. Therefore, the temporal and spatial evolution mechanism of the water-conducting fractured zone of overlying strata in the Kongzhuang coal mine is researched systematically in this paper. Especially, the hydro-geological conditions and mining conditions in the Kongzhuang coal mine are analyzed; on this basis, the fracture and movement of overlying strata are simulated by physical similarity simulation test, and the temporal and spatial evolution rule of overlying strata in the Kongzhuang coal mine is obtained. Besides, the development height of “two zones” is measured by double-end water shutoff detection method, and the risk assessment for the water inrush disaster of the coal seam roof is carried out in the Kongzhuang coal mine. The research achievements in this paper indicate that the water-conducting channel formed in the mining process of #7 coal seam is the most important water-filling channel. Quaternary aquifer water recharges the mine indirectly through the bedrock aquifer, and the sandy mudstone with large thickness is the key layer to control the development of a water-conducting fracture zone. Meanwhile, the height of the caving zone is 26.7 m, about 8.3 times of mining height, and the height of the fracture zone is 68.3 m, approximately equal to 16.26 of the mining height. The results of #1 and #2 borehole leakages in the double-end water shutoff detection method show that the height of the water-conducting fractured zone is 63.70 m-65.27 m, and the split-to-mining ratio is 15.17-15.54. The water inrush risk of the coal seam roof shows that most of the 7436 working face is in the transition zone, and a small area around the cutting hole of the working face is in the relatively dangerous zone. Therefore, the innovation of this paper is that the temporal and spatial evolution mechanism of the water-conducting fractured zone of overlying strata in the Kongzhuang coal mine is revealed, which provides the theoretical guidance for the prediction and prevention of water inrush disaster in the coal mine with the similar mining conditions.

When the collapse column of overburden is disturbed by the working face, the grain loss in the karst collapse column occurs by the dissolution and corrosion of groundwater, thereby inducing the water inrush disaster. The test samples are prepared based on the fractal theory and the Talbol grading theory, and the seepage evolution law of fractured rock in collapse column under triaxial stress is studied, by employing the triaxial seepage test equipment. Besides, the seepage mechanics model of broken rock is established and calculated in the COMSOL Multiphysics, and the water-conducting channel under mass loss condition in the collapse column is further elucidated. The research results indicate that the loss ratio of mass is inversely proportional to the Talbol power index, and grain mass loss rate increases with the decrease of the Talbol power index. During the infiltration process, the evolution of pore structure is related with grain size distribution. With the increase of the Talbol power index, the overall porosity increases. Grain loss is an internal factor in seepage loss stability. Flow speed is accelerating, and seepage pathways are communicated with each other to induce the water inrush disaster.

In order to reveal the mechanical response of surrounding rock of karst tunnel under stress-damage-seepage coupling effect, a new damage constitutive mechanical model of surrounding rock of karst tunnel under stress-damage-seepage coupling effect is established, which is calculated by the COMSOL Multiphysics in this paper. When the mechanical parameters are assigned to microelements of rock mass by the Weibull distribution function, the larger the value is, the more homogeneous the rock mass is. The variation trend of strain energy density with is similar to equivalent stress, which increases firstly and then decreases. The number of damage points increases with the increase in loading step and decreases rapidly after reaching the peak value and then remains a small number in the later loading stage. With the increase in , the stress range expands to the rock mass above the vault and below the floor; the stress value increases significantly, and the surrounding rock of karst tunnel is closer to strength limit, leading to the damage of rock mass. With the increase in , the area of the damage area in the upper part of the vault becomes larger, and most of the rock mass below the bottom plate is damaged; the damage area is semicircular, which indicates that both places are damaged by shearing action, resulting in the developed fissures. Besides, there are the distribution characteristics of “high value on both sides with a peak value and low value in the middle position” in the permeability distribution, and high permeability is located at the arch foot, and the low permeability is located at the floor. The larger the value of , the larger the permeability. The research achievements provide an important theoretical basis for prediction and treatment for dynamical disaster of karst tunnel.

Geothermal energy and underground thermal storage (UTES) are vital to the low-carbon energy transition, yet their optimization is bottlenecked by multi-scale heterogeneity, coupled thermal–hydraulic–mechanical–chemical (THMC) processes, and the high computational cost of full-physics simulations. This review systematically evaluates machine learning (ML) as a foundational paradigm for overcoming these computational and scale-bridging challenges. We categorize current advances into three key functional roles. First, data-driven upscaling directly maps pore-scale features to macro-scale effective properties, replacing traditional empirical homogenization. Second, deep surrogate models mimic high-fidelity THMC simulations at a fraction of the computational cost, enabling real-time prediction and uncertainty quantification. Third, physics-informed digital twins integrate real-time sensor streams with cloud architectures for dynamic reservoir management. Furthermore, we address the generalization limits of purely data-driven approaches, highlighting physics-informed machine learning (PIML) and hybrid architectures that embed conservation laws as strict constraints. Finally, we outline future pathways toward multimodal data fusion and edge-cloud deployment, marking a shift from static offline modeling to dynamic, physics-safeguarded real-time reservoir optimization.