Mechanical properties and microcracking behaviour of heterogeneous rocks after high-temperature treatment

The construction of large-scale deep underground projects, such as deep geothermal energy exploitation and the geological disposal of high-level radioactive nuclear waste, plays a critical role in ensuring national energy security, optimizing resource utilization, and addressing environmental protection challenges. These projects contribute to the sustainable and rapid development of the national economy. However, deep rock masses are typically subjected to complex environmental conditions, including high temperatures, and exhibit inherent heterogeneity. These factors significantly influence their strength characteristics and damage evolution mechanisms. In deep underground engineering, accurately predicting the mechanical properties, microcracking behavior, and strength evolution of heterogeneous rock under high-temperature conditions remains a key scientific challenge. This study investigates the mechanical behavior and microcrack evolution of heterogeneous rock, particularly granite, subjected to high-temperature treatment. A comprehensive approach integrating laboratory experiments, numerical simulations, theoretical modeling, and machine learning techniques is employed to gain deeper insights into the thermo-mechanical response of such rock materials. The main research contents and conclusions are as follows: (1) In terms of laboratory tests, the digital image correlation technology is used to quantitatively analyze the influence of temperature on the evolution of the fracture process zone under mode I/II loading conditions. The results show that with the increase of temperature, the mechanical properties of rock (including peak load, generalized stiffness, mode I/II fracture toughness and absorbed energy) gradually decreased. The load-displacement curve and macroscopic fracture characteristics of rock after high-temperature treatment showed that the rock gradually changed from brittle to ductile. The length and width of the fracture process zone increased with the increase of temperature. The length-to-width ratio of the fracture process zone was concentrated in 2~3 at lower temperatures (25~400 °C), while it increased to 3~4 at higher temperatures (600~1000 °C). Subsequently, the effects of grain size and mineralogical composition on the fracture toughness, fracture path, tortuosity and fracture width of granite under mode I loading conditions after high-temperature treatment were further studied. The results show that granite with larger average grain size and more heterogeneous grain size distribution is more sensitive to high-temperature treatment. Coarse-grained granite is more likely to produce macroscopic fractures after high-temperature treatment. And the linear crack density increases with the increase of temperature and grain size. The fracture surface topography varies with the mineralogical composition and treatment temperature, such as conchoidal fracture and lamellar tearing fracture. Grain size and mineralogical composition significantly affect the load-displacement curves and mode I fracture toughness of rocks. Between 400 °C and 600 °C, a clear brittle-ductile transition was observed in coarse-grained granite, but fine-grained granite still showed strong brittle characteristics after treatment at 1000 °C. The mode I fracture toughness decreases with increasing temperature and grain size. In addition, grain size affects the tortuosity of the fracture path under mode I loading, which is particularly evident in coarse-grained and medium-grained granite. In medium-grained granite, the tortuosity and width of the fracture path increase with increasing treatment temperature, while the effect of temperature on the tortuosity of the fracture path in fine-grained granite is negligible. (2) In terms of numerical simulation, a grain-based model based on modeling the real microstructure of rocks is proposed. Five granite specimen models with different grain size heterogeneity indices H (H ranging from 0.19 to 1.22) are established. The effects of grain size heterogeneity on the mechanical properties, microcrack behavior, and multi-scale progressive damage evolution of granite treated at high-temperature under uniaxial compressive loading were quantitatively analyzed. The results show that grain size heterogeneity and treatment temperature have a key influence on the mechanical properties of granite. The mechanical properties show a decreasing trend with the increase of temperature and grain size heterogeneity index. Subsequently, several empirical formulations are proposed to predict the temperature-related mechanical properties of granite with different grain size heterogeneity index. In general, granite specimens with greater grain size heterogeneity index have lower critical temperatures for brittle-ductile transition. Both temperature and grain size heterogeneity affect the microcracking behavior of rocks, including the number, density, and orientation. With the increase of temperature, the number and density of thermally induced microcracks also increase. In general, the thermal stress of granite during heating increases with the increase of grain size heterogeneity index. When the granite is relatively homogeneous (H ranges from 0.19 to 0.76), the failure pattern is mainly controlled by thermal stress-induced cracks; when the granite is relatively heterogeneous (H ranges from 0.93 to 1.22), the controlling factor of the failure pattern becomes the heterogeneity of the grain size distribution. (3) In terms of theoretical model, based on the sliding wing crack model, it is proposed to use temperature-dependent initial damage to characterize the microstructural damage of rock during high-temperature treatment. The mathematical expression of the proposed rock nonlinear failure criterion is consistent with the Hoek-Brown failure criterion for intact rock. Then, five sets of experimental data were used to verify the accuracy of the micromechanics-based failure criterion proposed in this work. The results show that the failure criterion considering temperature-dependent initial damage can well capture the strength characteristics of rock after high-temperature treatment. The temperature-dependent initial damage parameter well describes the influence of thermal damage on the strength evolution of rock after high-temperature treatment. (4) In terms of machine learning, five predictive models for the triaxial compressive strength of rock subjected to high-temperature treatment were developed using different machine learning techniques, all of which achieved relatively accurate predictions. Utilizing seven input parameters (specimen diameter and height, rock density, temperature, confining pressure, crack damage stress, and elastic modulus), the back-propagation neural network (BPNN) demonstrated strong predictive capability. The findings indicate that machine learning techniques enable geotechnical engineers to effectively utilize extensive field and experimental data to achieve accurate and efficient predictions of rock properties, including the temperature-dependent triaxial compressive strength.