Evolution Characteristics of Coal Fractures under High-Temperature Coupling with Triaxial Stress

Abstract

Underground coal fires pose severe hazards to safe mining and environmental sustainability, with fracture development in coal serving as critical pathways for oxygen supply and heat transfer. However, the dynamic evolution of fractures under coupled high-temperature and triaxial stress conditions remains poorly understood. In this study, bituminous coal samples from a fire-affected mining area in Xinjiang, China, were subjected to high-temperature treatments (200 and 400 °C) and subsequently loaded under triaxial compression (confining pressure 5 MPa) while being monitored by an industrial CT scanning system. Full stress–strain curves were obtained, and VG Studio MAX software was used to quantitatively analyze fracture morphology, volume, density, porosity, and 3D fractal dimension across five successive loading stages. The results reveal that temperature governs the mechanical behavior of thermally treated coal. With increasing temperature, the compaction stage shortens, elastic modulus decreases, and peak stress first increases then decreases, while the durations of plastic deformation and postpeak failure stages first shorten then lengthen. The porosities of RC, C200, and C400 coal samples at the first scan were 0.07%, 0.28%, and 0.70%, respectively, indicating that high-temperature treatment increased the initial porosity of the coal samples. Based on fracture density, porosity, and fractal dimension, fracture evolution is divided into three stages: smooth development, slow growth, and rapid growth. The 3D fractal dimension further confirms this staged evolution. These findings provide a theoretical basis for predicting fracture network evolution in coal fire zones and offer practical guidance for fire prevention strategies.

1. Introduction

As mining operations extend into deeper and more geologically complex conditions, surrounding rock masses in roadways exhibit increasingly developed joints and fractures, leading to rapid deformation and damage. Among various geotechnical hazards, underground coal fires pose a particularly severe challenge to safe and efficient mining. , These fires typically originate from the spontaneous combustion of residual coal in mined-out areas, with core temperatures potentially exceeding 1000 °C. Such fires not only consume vast coal resources but also cause serious environmental pollution on a global scale. − A critical factor exacerbating the hazard of underground coal fires is the development and propagation of fractures in coal and rock masses, which serve as pathways for oxygen supply and heat transfer. Therefore, understanding the evolution of fractures under coupled high-temperature and stress conditions is essential for effective coal fire prevention and control.

Extensive studies have characterized the fracture and pore structures of coal. − Zhou et al. used NMR to investigate pore-fracture evolution during dry-wet cycles and revealed how these changes regulate coalbed methane storage. Similarly, Wang et al. employed NMR to investigate the pore structure evolution and mechanical performance of coal gangue-fly ash cemented backfill under different aggregate gradations, demonstrating a strong negative exponential correlation between porosity and uniaxial compressive strength, with harmful pores (>100 nm) acting as the primary strength-limiting factor. Yu et al. quantitatively characterized dry-wet cycle-induced porosity increase and physical deterioration of bituminous coal using micro-CT. Wan et al. integrated CT-based 3D reconstruction with COMSOL simulations to reveal that preinjection blocking fluid suppresses pore structure development, reduces pore throat abundance, and impairs connectivity, with such structural alterations dictating fluid seepage characteristics. Wu et al. derived Df3 = C × Df2 + 1 from CT-based fractal analysis of coal microfractures and established an exponential porosity–fractal dimension relationship. Li et al. applied μCT scanning and 3D reconstruction techniques to reveal the spatial distribution and dynamic evolution of fractures in coal under triaxial compression.

Despite these advances, most of these studies were conducted under room-temperature conditions or focused on single-factor effects (temperature or stress alone). Few have systematically investigated the coupled effect of high-temperature pretreatment and subsequent triaxial loading on the dynamic evolution of coal fracturesa scenario directly relevant to underground coal fires, where coal is first subjected to prolonged high temperatures followed by mechanical stress. Moreover, quantitative characterization of fracture evolution across the full stress–strain process under such coupled conditions remains limited. The characterization of pore size distribution in low-rank coals has been further refined by integrating multiple experimental techniques. Tao et al. constructed a continuous distribution model of gas storing space in low-rank coals by combining mercury intrusion porosimetry (MIP) and low-temperature nitrogen adsorption (LTNA) data, successfully obtaining the complete pore size distribution from nanopores to microfractures. This approach provides a methodological foundation for quantitatively evaluating the contribution of different pore sizes to total porosity and gas storage capacity, which is directly relevant to the porosity characterization in our study.

Researchers have also examined the mechanical behavior and damage mechanisms of coal under different loading conditions. Liu et al. showed how fractures grow and mesopores collapse in deep coal seams when stress is released. They also found that controlling the speed of cavity creation is key to protecting gas migration paths. Wang et al. used industrial CT to observe spatial fracture distribution after triaxial tests, identifying compression-shear damage as the dominant failure mode. Zhang et al. conducted in situ X-ray tomography using a novel X-ray-transparent triaxial core holder, providing direct evidence that increased effective stress compresses large fractures into smaller isolated ones, leading to a dramatic reduction in fracture porosity. The coupled effects of temperature, confining pressure, and pore structure on rock mechanical behavior have been systematically reviewed by Ma et al., who emphasized that triaxial testing remains the most direct and effective method for investigating rock mechanics under deep geological conditions. Furthermore, few experimental studies have directly observed the dynamic evolution of fractures in coal under triaxial stress following high-temperature exposure by using real-time CT scanning.

The effects of thermal treatment on coal structure and properties have garnered increasing attention. Xue et al. investigated the effects of heating–cooling cycles on the mechanical properties and damage evolution of granite for underground cryogenic storage, revealing that thermal shock significantly promotes crack propagation and structural degradation, with crack width increasing exponentially with temperature. While their study focused on granite in the context of cryogenic storage, the thermal damage mechanisms identifiedincluding thermal stress-induced microcracking and progressive mechanical degradationare directly relevant to coal under high-temperature conditions. Zhao et al. conducted high-temperature water vapor pyrolysis experiments to reveal the evolution of permeability and pore-fracture structure in oil shale during heating and identified the key temperature thresholds. Wu et al. found that the number of pore fractures and throats reached a maximum at 200 °C, after which thermal rupture became the dominant damage mechanism. Shi et al. used in situ CT to study bituminous coal during steam pyrolysis, showing significant pore-fracture evolution at 300–400 °C, with “V”-shaped permeability and increased porosity after high-temperature treatment. Yang et al. reported that preoxidation at 140 °C optimizes coal’s mechanical properties, with maximum elastic modulus, compressive strength, and thermal conductivity. Liang et al. observed surges in pore volume, specific surface area, and surface cracks after coal heating, and established a basis for zoning oxidation-prone seams based on pore-crack size and oxidation kinetic parameters. The thermal decomposition of coal and its effect on pore structure have also been investigated from a coalification perspective. Tao et al. studied the material composition, pore structure, and adsorption capacity of low-rank coals around the first coalification jump (FCJ) in the eastern Junggar Basina region adjacent to our sampling site (Sulfur Gorge mine, southern Junggar Basin). This coalification-induced evolution is analogous to the thermal treatment-induced changes observed in our study: as treatment temperature increases from 200 to 400 °C, the coal matrix undergoes organic matter consumption, porosity evolution, and subsequent mechanical property degradation. However, unlike coalification which occurs over geological time scales (millions of years), our study investigates the immediate effects of short-term high-temperature treatment (hours) on the mechanical behavior of coal under triaxial stressa process that has received limited attention in the literature.

Recent technological advances have enabled more sophisticated investigations of coupled processes in coal and rock masses. Tao G et al. employed multiscale characterization methods to systematically reveal the development characteristics and controlling factors of the pore-fracture structure in deep coal seams, and elucidated its influence on permeability. Li T et al. employed a multiscale approach combining experiments and numerical simulations to reveal the influence mechanism of microscopic characteristics, including mineral composition and pore structure, on hydraulic fracture propagation in shale. Yan F et al. proposed and validated a novel acidification-assisted controlled electric pulse (ACEP) method for synergistically improving the pore-fracture structure of coal reservoirs and significantly enhancing permeability. Recent advances in CT-based characterization have enabled more sophisticated investigations of fracture evolution in coal and rock. Xue et al. employed X-ray CT combined with gray-level co-occurrence matrix (GLCM) texture analysis to quantify mesoscopic damage in granite under cyclic liquid nitrogen shocks, demonstrating that cyclic thermal-cryogenic treatment significantly exacerbates cracking and increases fracture node numbers. Li et al. integrated micro-CT and three-dimensional laser profilometry to characterize fracture network evolution under cyclic thermal and cryogenic impact, revealing progressive mechanical deterioration with enhanced fracture connectivity. Tang et al. employed a self-developed THM-CT coupled test system to investigate the mesostructure evolution and fracture mechanism of shale under real-time thermal-mechanical coupling conditions, establishing a quantitative relationship between microscopic damage factor and applied stress. While these studies have advanced our understanding of fracture evolution under thermal-mechanical conditions, they either focused on granite or shale, used cyclic thermal shock rather than sustained high-temperature pretreatment, or employed relatively low temperature ranges (20–180 °C). Moreover, the dynamic evolution of fractures in coal under coupled high-temperature (200–400 °C) and triaxial stress conditionsa scenario directly relevant to underground coal firesremains poorly understood.

While existing research has extensively examined the static characteristics of pore-fracture structures, mechanical properties, and thermal damage of coal, limited attention has been paid to the dynamic evolution of fractures under coupled high-temperature and high-stress conditions. This scenario is directly relevant to underground coal fires. Specifically, how preexisting thermally induced fractures respond to subsequent mechanical loading and how this coupled process influences overall fracture network development remain poorly understood.

To address this knowledge gap, this study investigates the dynamic evolution characteristics of fractures in high-temperature-treated coal under triaxial stress. Bituminous coal samples from the Sulfur Gorge mine in Xinjiang Uygur Autonomous Region, China, an area affected by coal fires, were subjected to high-temperature treatments at 200 and 400 °C. The treated samples were then loaded under triaxial compression and monitored using an industrial CT scanning system specifically designed for loaded coal. Through mechanical analysis and 2D/3D image processing using VG Studio MAX software, the evolution of fracture morphology, volume, density, porosity, and fractal dimension was quantitatively characterized across different compaction stages. By systematically analyzing the full stress–strain process combined with real-time CT imaging, this study reveals the characteristic changes and evolutionary stages of fractures inside high-temperature coal bodies during mechanical loading. The findings provide a theoretical foundation for understanding fracture behavior in underground coal fire zones and offer practical guidance for fire prevention and control strategies. Compared with previous studies, this work makes three new contributions: (i) it systematically investigates the coupled effect of high-temperature pretreatment (200 and 400 °C) and subsequent triaxial loading using real-time CT scanning; (ii) it quantitatively characterizes fracture evolution across the full stress–strain process using porosity, fracture density, and 3D fractal dimension; (iii) it identifies three temperature-dependent fracture evolution stages (smooth development, slow growth, and rapid growth) correlated with mechanical deformation stages.

2. Experimental Methods

2.1. Coal Samples

The experimental coal samples were bituminous coal collected from the working face of the coal seam in the Xishanyao Formation (Middle Jurassic) at the Sulfur Gorge mine in Xinjiang Uygur Autonomous Region, China. This mine is currently threatened by coal fires. According to the GB/T23561.7–2009 standard, a cylindrical coal sample (Φ25 mm × 50 mm), suitable for industrial CT scanning of loaded coal, was drilled from the working face to serve as the original coal sample, as shown in Figure (a).

1.

(a, b) Test coal sample.

Based on the relationship between index gas and temperature, the coal transitions from slow to rapid oxidation at temperatures below 200 °C. According to the relationship between oxygen consumption rate and temperature, the complete combustion stage is at 400 °C. Therefore, the RC coal sample after the screening were processed through a muffle furnace to obtain 200 and 400 °C coal samples. The coal samples were heated in the muffle furnace at a rate of 5 °C/min and held at the target temperature (200 or 400 °C) for 2 h. To prevent the oxidation and spontaneous combustion of the coal samples at high temperatures after being removed from the muffle furnace, the heated coal samples were quickly wrapped with high-temperature-resistant asbestos and then placed in a nitrogen bag for storage. The samples were allowed to cool naturally to room temperature (approximately 30 °C) inside the nitrogen-filled bag. The cooling rate followed a natural decay: the temperature dropped from 400 °C/200 °C to below 100 °C in approximately 45 min, and to room temperature after about 3 h. Under this inert, slow-cooling condition, additional microcracking caused by thermal stress during cooling is considered minimal. Thus, the fracture evolution observed in subsequent tests primarily reflects the coupled effect of high-temperature treatment and mechanical loading, rather than cooling-induced damage. It should be noted that the present study focuses on the residual effects of high-temperature treatment on the subsequent mechanical behavior and fracture evolution of coal. Therefore, the coal samples were tested after cooling to room temperature. Although this approach does not fully reproduce the coupled thermo-mechanical conditions in underground coal fire zones, it allows the temperature-induced structural damage to be preserved and quantitatively evaluated during subsequent triaxial loading. Therefore, the conclusions of this study are primarily applicable to thermally treated coal and the assessment of residual damage caused by high-temperature exposure. The fracture evolution behavior under simultaneous heating and in situ stress conditions warrants further investigation in future studies. The RC coal sample and the treated coal samples at 200 and 400 °C were noted as RC, C200, and C400, respectively, as shown in Figure (b).

The mass changes of the experimental coal samples before and after the high-temperature treatment were recorded using an electronic balance, and the dimensional changes were recorded using vernier calipers. A coal sample with a relatively flat top and bottom section and a diameter close to 25 mm was selected as the industrial CT experimental coal sample, and its basic parameters are listed in Table .

1. Basic Parameters of Industrial CT Coal Samples before and after High-Temperature Heating.

	Before high-temperature heating			After high-temperature heating
Coal samples	Diameter (mm)	Height (mm)	Weight (g)	Diameter (mm)	Height (mm)	Weight (g)
RC	25.01	50.02	31.87	-	-	-
C200	25.00	50.01	31.16	25.01	50.07	26.95
C400	25.01	50.01	32.19	25.04	50.16	26.74

2.2. Test Apparatus

The experiments were conducted using a loaded coal-rock industrial CT scanning system developed by the GENERTEC Electric Group, USA. The system consists of two devices: an industrial CT scanner and a TRS10030–50/25 triaxial stress loading tester, as shown in Figure .

2.

(a-f) Loaded coal-rock industrial CT scanning system.

The industrial CT scanner was operated at a tube voltage of 180 kV in order to obtain high X-ray penetration of coal-rock masses after treatment at different temperatures. The principle is shown in Figure . The tube current was set to 240 μA to obtain higher contrast. A scanning setting with an exposure time of 800 ms was used to obtain higher quality images with less smearing. Each scan acquires 1200 images/time to get a finer fracture volume.

3.

Industrial CT system working principle.

The triaxial stress loading test model TRS10030–50/25 can apply a maximum axial stress of 100 kN, a maximum displacement of 20 mm, and a maximum axial pressure of 30 MPa. Confining pressure is provided by the confining pressure fluid, and the maximum temperature can reach 80 ± 0.5 °C, and the maximum confining pressure is 30 MPa. The test sample turntable can withstand up to 30 kg of coal sample.

The VG Studio MAX software can be used to observe and obtain the pattern of coal-rock fracture evolution more intuitively and comprehensively, as shown in Figure (e) and (f).

2.3. Test Program

For each temperature condition (RC, 200 °C, 400 °C), three parallel specimens were prepared and subjected to the same heating, loading, and CT scanning procedures. The quantitative results reported in this study represent the average values obtained from the parallel tests, while the CT images and 3D fracture evolution figures correspond to representative specimens showing typical fracture development characteristics. Based on the estimated in situ stress conditions of the Xinjiang Sulfur Gorge coal fire area from which the coal samples were collected, a confining pressure of 5 MPa was selected for the triaxial tests. This estimation is supported by detailed exploration of this area, which revealed that the deepest combustion zone reaches 215 m, confirming that the coal fire primarily affects shallow coal seams at burial depths of less than 300 m. Using the empirical formula σ_v = γH, where γ is the average unit weight of overlying strata (typically taken as 25 kN/m³ in rock mechanics), the vertical stress at a depth of 200 m is calculated as 5 MPa. Accordingly, a confining pressure of 5 MPa was adopted to represent the actual geostress conditions of the shallow coal fire zone. This selection is also consistent with the confining pressure ranges commonly used in previous studies on coal mechanics under similar depth conditions. − To prevent the coal sample from shaking under triaxial stress loading, a downward force of 0.8 kN was applied to the sample in advance to keep it stable. Each experimental coal sample was scanned five times successively for five stages: the unapplied confining pressure, the fracture compaction stage, the linear elastic deformation stage, the plastic deformation stage, and the complete failure stage.

2.3.1. Preheat Startup

Due to the high precision and ionizing radiation characteristics of industrial CT equipment, the system must be preheated under ambient conditions with a temperature below 30 °C and humidity below 85% before operation. To ensure the proper operation of the ray tube, the vacuum level inside the tube was set to <20 nbar. The experiment would start after the indicator of each part setting shows green.

2.3.2. Installation of the Coal Sample

The coal samples after different high-temperature treatments were placed in a carbon fiber latex protective sleeve with a special loader. The coal samples were loaded into the triaxial stress gripper and fixed on the scanning turntable. The coal sample was centered in the detection area, and the resolution was adjusted to display it with a high quality on the detection screen. Finally, the axial pressure and confining pressure lines of the triaxial stress system were checked to ensure a proper connection.

2.3.3. Triaxial Stress CT Scan Experiments

Starting the TRS10030–50/25 triaxial stress tester involves inputting the basic parameters of the coal sample and setting the confining pressure to 5 MPa. Axial stress can be applied continuously until a sample is completely broken, and the axial stress can be stopped by the system at any stage before a sample is broken according to the stress–strain curve. At each selected loading stage, the axial loading was paused according to the stress–strain curve and the target stress level was maintained constant during CT scanning. Under these load-holding conditions, CT scanning was performed to obtain the fracture morphology corresponding to the selected loading stage. Each scan acquired 1200 projection images with an exposure time of 800 ms per image. Based on the stress–strain records corresponding to the selected loading stages, the stress variation around the target stress levels was estimated to be within 1.6–4.8%, indicating a relatively stable stress condition during scanning. Therefore, the reconstructed CT images represent fracture states under quasi-static load-holding conditions and are considered representative of the fracture evolution characteristics at the corresponding loading stages.

2.3.4. 3D Digital Image Reconstruction and Data Analysis

Using the dedicated software of the CT scanning system, the scanning region of the coal sample was selected and beam-hardening correction was applied. The reconstructed data sets were then imported into VG Studio MAX for fracture extraction and quantitative analysis. Fractures were first identified using grayscale threshold segmentation based on the contrast between fractures and the coal matrix. Subsequently, manual correction was performed to remove image artifacts and nonfracture features and to refine the fracture boundaries. The same extraction procedure was applied consistently to all samples. The extracted fracture network was then used to calculate fracture volume, fracture density, porosity, and 3D fractal dimension. Finally, fracture parameters were analyzed together with the corresponding stress–strain data.

3. Results and Analyses

3.1. Stress–Strain Analysis

The TRS10030–50/25 triaxial stress tester was used to obtain the full stress–strain curves of coal samples after different temperature treatments in real time. Stress and strain values were recorded at 1 s intervals, and the full stress–strain curve was obtained as shown in Figure . The mechanical parameters during triaxial loading are shown in Table . The elastic modulus was calculated from the slope of the linear elastic segment of the stress–strain curve. According to the stress–strain characteristics of the specimens, the strain ranges used for modulus determination were 0.5–2.1% for RC, 1.5–4.6% for C200, and 1.6–3.6% for C400, corresponding to the linear elastic deformation stage after crack compaction and before yielding.

4.

Stress–strain curves of triaxial loading tests on coal samples at different temperatures.

2. Mechanical Parameters of Triaxial Loading Test for Coal Samples.

Coal samples	Diameter (mm)	Height (mm)	Loading method	Confining pressure (MPa)	Strength (MPa)	Elastic modulus (GPa)
RC	25.01	50.02	Triaxial stress loading	5	44.13	1.41
C200	25.01	50.07		5	63.54	0.77
C400	25.04	50.16		5	58.21	0.74

As shown in Figure , the stress–strain curves of the coal samples treated at different temperatures after the fixation of the confining pressure are consistent with the basic compression process of coal and rock masses. In the compaction stage, the strain growth interval of C200 and C400 coal samples after high-temperature treatment is wider than that of RC coal samples, and the rate of change of strain is larger than that of stress. This is because the high-temperature pretreatment generated thermal stress and caused moisture desorption, producing microcracks and increasing the initial porosity of the C200 and C400 samples. After cooling and mounting in the testing apparatus, these preexisting features result in a more pronounced compaction behavior during the initial loading stage, manifesting as a wider strain growth interval. The stress–strain curve then increases rapidly, entering the elastic deformation stage. From the slope of the rising curve as well as Table , it can be seen that the lower elastic modulus of coal samples after high-temperature treatment is associated with both thermal stress and the thermal degradation of organic macromolecular structures. As shown in Table , the mass of the coal samples decreased after heating, indicating the loss of volatile components and the consumption of part of the organic macromolecular structures. This degradation weakens the internal bonding structure of the coal body and contributes to the reduction in elastic modulus, making the coal samples more susceptible to destabilization and deformation. It is also noted that the elastic modulus decreases sharply from RC to C200 (1.41 GPa → 0.77 GPa, a 45% reduction) but then decreases only marginally from C200 to C400 (0.77 GPa → 0.74 GPa, a further 4% reduction), forming a plateau. Meanwhile, the peak stress increases from RC (44.13 MPa) to C200 (63.54 MPa) by 44%, then decreases from C200 to C400 (58.21 MPa) by 8%. Both trends suggest that significant mechanical changes occur around 200 °C under the tested conditions. The plateau in modulus is explained by the mass loss data in Table : the C200 sample loses 4.21 g (13.5%) at 200 °C, accounting for about 80% of the total mass loss of C400, indicating that major physical changes are largely completed by 200 °C, leaving limited room for further modulus reduction. The peak strength increased from 44.13 MPa for RC to 63.54 MPa for C200 and then decreased to 58.21 MPa for C400. This nonmonotonic trend may be associated with different dominant mechanisms at different temperature levels. At moderate temperature (200 °C), moisture evaporation and the release of adsorbed water may alter the internal pore–fracture system and reduce the weakening effect of some preexisting defects, resulting in a temporary enhancement of the load-bearing capacity. Previous studies have shown that heating can induce moisture loss, gas desorption, and significant changes in the pore–fracture structure of coal. , However, as the treatment temperature further increased to 400 °C, thermally induced damage, including the development of microcracks and the degradation of the coal matrix, became dominant, leading to a reduction in peak strength. Therefore, 200 °C may represent a transition region under the present experimental conditions. Since no direct microstructural or chemical characterization was conducted in this study, this interpretation is presented as a possible explanation rather than a definitive mechanism. With the increasing axial stress, the curve shows the plastic flow characteristics (the stage of plastic deformation), and the plastic deformation stage initially shortens and then lengthens with increasing temperature. This lengthening occurs because coal is a porous medium, and high temperature consumes part of the coal matrix, reducing its brittleness and leading to a gradual increase in strain at this stage. From the peak stress to the complete destruction of the coal sample, the strain remains constant, and this process is called the stage of complete destruction. It can be seen that the RC coal sample and C400 coal sample decline rapidly at this phase, while the C200 coal sample decreases slowly, indicating that the coal sample still maintains residual mechanical properties at this time. This contrasting behavior is attributed to differences in fracture penetration. As shown in Figure , the through-going fractures in C200 do not fully penetrate the coal body, allowing the unbroken portion to sustain stress through friction and interlocking. In contrast, fully penetrating fracture networks in RC and C400 lead to a complete loss of integrity and a rapid stress drop.

12.

(a) Evolution of the 3D fractal dimension at different scan stages; (b) relationship between the 3D fractal dimension and temperature after complete failure (fifth scan). Error bars represent the standard deviation of three parallel specimens (n = 3).

3.2. 2D Coal-Body Fracture Evolution Analysis

The development of external fractures in the high-temperature coal-rock mass under triaxial stress was examined by observing the CT-scanned 2D transverse and longitudinal sections of the coal samples under triaxial compression at each temperature treatment. As well as the structural morphology of the internal fractures, the dynamic process of initiation, fracture sprouting, and interpenetration in the 2D sections is shown in Figures –.

5.

(a, b) Transverse section and longitudinal section of the RC coal sample under triaxial loading.

7.

(a, b) Transverse section and longitudinal section of the C400 coal sample under triaxial loading.

3.2.1. RC Coal Sample

A 2D transverse section of the triaxial stress of the RC coal sample is shown in Figure (a). The presence of numerous white spots indicates a higher mineral content in the RC coal sample. During the first to fourth scans, no obvious fractures were found, but a slight reduction in the area of minerals could be found by comparing each scan, indicating that the coal samples were compacted at this time. With the increasing axial stress, the coal body is destabilized and damaged after exceeding the peak stress. The coal body shows complex through fractures from the upper part to the lower left and lower right, and the fractures bypass the minerals and shear damage only in the coal body.

As shown in Figure (b), it can be seen that the RC coal sample only has very few original fractures at the lower edge of the coal body and in the direction parallel to the laminae at the first scan. Closure of the original fractures occurs when the axial stress rises to the second scan. The third scan revealed new fractures, appearing in the upper left end face, upper right end face, and lower right section, oriented parallel to the main stress direction. Approaching peak stress at the fourth scan, the fractures that appeared in the third scan began to develop and extend into the interior of the coal body, and two new fractures appeared in the coal core area, parallel and perpendicular to the main stress direction, respectively. At the fifth scan, all the fractures that appeared in the fourth scan widened and grew, there were multiple through fractures parallel to the main stress direction, and the microfracture direction was mostly perpendicular to the main stress direction.

3.2.2. C200 Coal Sample

As shown in Figure (a), no obvious fractures were found in the cross-section at the first to fourth scans, and it can be seen that there are fewer minerals inside the coal sample than in the RC coal sample, and the fifth scan is the stage of complete destruction, and four obvious fractures appear and penetrate each other through the midpoint of the coal core, among which the fractures at the lower left end intersect with minerals. Compared to the RC coal sample, the fractures in the fifth scan’s transverse section are narrower.

6.

(a, b) Transverse section and longitudinal section of the C200 coal sample under triaxial loading.

As shown in Figure (b), it can be seen from the first scan that the temperature has caused the coal to produce many marginal fractures and air pores, and at the bottom end, fractures appear in the direction of the vertical principal stress and are close to the minerals. The second and third scans showed little change and were the same as the RC coal sample, with only fracture compaction closures occurring. After the peak stress, a fourth scan was performed, and the fractures appeared at the upper left and upper right ends, extending toward the coal body proper in the direction of the parallel principal stress and connected to the high-temperature air pores. At the same time, fractures parallel to the axial stresses appear in the axial line of the coal body. The fifth scan is the complete failure stage. The fractures that originate from the through expansion of the fracture of the fourth scan, which is parallel to the main stress direction and terminates at the center of the coal body, showing a “V”-shaped through fracture.

3.2.3. C400 Coal Sample

The transverse section in Figure (a) shows fewer white areas, indicating that the coal sample contains only a small number of minerals. From the transverse sections of the first CT scan, the cylindrical coal sample had the largest number of fractures at the circumferential surface compared with other coal samples due to the effect of high temperature, and the fractures extend continuously from the outside to the inside, showing a change trend of wide outside and narrow inside. From the development location of fractures, the fractures of coal samples under the action of high temperatures of 400 °C mostly stop at the white minerals, and the fractures mostly exist near the minerals appearing. When in the second and third scans, the fracture length remained unchanged, and there was a significant decrease in width. From the results of the first to third scans, no macroscopic fractures were produced during the process from the initial scan to the plastic deformation stage, indicating that this stage is in the compaction closure phase. With the increasing axial stress, the fractures that were compacted in the fourth scan were re-expanded, and the width of the fractures was narrower than that in the first scan, while many new fractures sprouted in the area closer to the center of the circle. By the fifth scan, macroscopic through-going narrow fractures had appeared inside the coal sample. The lower end of the through-going fracture corresponded to the fracture generated by high-temperature treatment, while the upper end was a new fracture induced by loading. At this stage, the coal sample had entered the complete failure stage.

As shown in Figure (b), the longitudinal section better characterizes the dynamic evolution of the overall rupture of the coal sample from the top to the bottom as it passes through the coal core. From the results of the first scan, there are many minerals distributed inside the coal sample, and the color imbalance also indicates that the experimental coal sample is a nonhomogeneous coal and rock mass. The top and bottom of the C400 coal sample are distributed with dense microfractures, and a penetrating fracture with parallel laminae appears near the center. More obvious fractures were present on both sides perpendicular to the laminae, and these were mostly connected to minerals. A high degree of fracture closure in the vertical and parallel laminae directions was observed during the second and third scans. The axial stress continues to increase, and in the fourth scan, the fractures produced by the high-temperature effect perpendicular to the laminae direction begin to develop and extend, among which the end corners of the coal samples are more obvious. When the axial stress reached its peak, the compressive resistance of the coal sample decreased dramatically, the axial stress then decreased, and obvious shear damage and penetration fractures appeared. The starting and ending ends of the fractures were in the same direction as the new fractures sprouted at the fourth scan, and a small number of tensile fractures appeared at the lower left end.

3.3. 3D Coal-Body Fracture Evolution Analysis

3.3.1. RC Coal Sample

The fracture-related data were exported and analyzed by using VG Studio MAX software. The calculated values of fracture density and porosity at each scanning stage are presented in Table .

3. Calculation Results of Fracture Density and Porosity of Coal Samples at Different Temperatures.

	RC		C200		C400
Number of scans	Fracture density/mm–1	Porosity/%	Fracture density/mm–1	Porosity/%	Fracture density/mm–1	Porosity/%
1	0.037	0.0007	0.066	0.0028	0.364	0.0070
2	0.027	0.0005	0.064	0.0027	0.403	0.0079
3	0.298	0.0053	0.247	0.0043	0.442	0.0086
4	0.316	0.0059	0.333	0.0089	0.489	0.0193
5	1.451	0.0921	0.903	0.0314	0.733	0.2593

As shown in Figure , most of the fractures perpendicular to the main stress are distributed inside the RC coal sample, and there are continuous microfractures on the axial line of the sample. The initial porosity values are 0.07% for RC, 0.28% for C200, and 0.70% for C400 (see Tables and , first scan). At a constant confining pressure of 5 MPa, with the increase of the axial direction, it is obvious in the second scan at the elastic deformation stage that most of the fine original fractures are compacted and closed, and the volume of the remaining part of the original fractures is significantly reduced. New fractures sprouted near the top and middle of the sample, and the total volume increased by 3.4 mm³ when the sample was in the compaction phase. At the third scan, the fractures seen in the second scan had different degrees of expansion, while many new fractures appeared, mainly at the upper end edge and the lower right edge, at which time the strain increased by 1.01, and the fracture volume grew to 129.14 mm³, which is the plastic deformation stage. With the gradual increase of axial stress, the fourth scan was carried out in the plastic deformation stage, and some of the fractures that were compacted and closed were reopened, the fractures out of the third scan were less dilated, some fractures were closed in the upper part, the range of fractures at the top and low end edges increased significantly, and the volume of fractures was 144.63 mm³ at this time. When the axial stress exceeded the peak, the fifth scan was carried out, at which time the coal sample was completely destabilized and damaged, the axial stress decreased significantly, the fracture volume increased substantially to 2260.13 mm³, the fracture further expanded and penetrated, and the width, length, and fracture abundance increased to the maximum, which was the stage of complete destruction and residual deformation of the coal sample.

8.

3D fracture evolution of the RC coal sample under different stress levels.

3.3.2. C200 Coal Sample

As shown in Figure , when the coal sample was treated at 200 °C, the fracture distribution was more uniform from the first scan. Compared with the RC coal sample, the fracture volume increased by 52.03 mm³ to 68.07 mm³, and the surface appeared with nearly round pores and flat elongated fractures, which further verified that the oxidation reaction would generate pores in the coal sample at 200 °C. Heat and mass transfer cause coal sample quality degradation, and thermal stress causes coal sample expansion and cracking. When the coal sample is scanned for the second time, it can be seen that the number of fractures is significantly reduced. At this time, the fracture volume of the coal body is 65.98 mm³, and the strain is increased by 3.22%, indicating that there is a significant reduction of fracture volume in the compacting stage. The higher volume reduction and increased strain were due to more fracture closures formed at high temperatures. In the third scan, the closed fracture opened again, and a new fracture appeared on the lower middle surface, and the volume of the fracture increased to 105.115 mm³ at this time, with a more obvious increase. With the increasing axial stress, when the peak stress phase was reached, the upper part of the coal sample fractures developed rapidly, expanded, and connected, and a large number of obvious concentrated new fractures developed, some fractures closed in the lower part, and new fractures appeared in the lower left side, at which time the volume of fractures was about twice that of the third scan, reached 218.56 mm³, and the strain also changed greatly, increasing by 1.99% relative to the third scan. When the stress increases to the peak stress, the coal as a whole destabilizes and breaks down, the upper fracture further extends to the middle and lower part, but does not penetrate the whole coal body, at this time is the complete failure stage, forming a complex fracture network channel in the upper 2/3 of the coal body, and the fracture volume increases to the maximum value of 771.13 mm³.

9.

3D fracture evolution of the C200 coal sample under different stress levels.

3.3.3. C400 Coal Sample

As shown in Figure , the first scan of the coal sample was performed in the relaxed state after 400 °C, and the fracture volume was 172 mm³ at this time. This initial volume was more than 10 times that of the RC sample (16.04 mm³) and approximately 2.5 times that of the C200 sample (68.07 mm³), because the high temperature of 400 °C made the fracture and pore structures of the coal sample more abundant. The thermal damage, heat transfer, and thermal stress simultaneously consumed and expanded the coal body, resulting in abundant fractures and well-developed pores. When the coal body was scanned for the second time, the fracture volume was 192 mm³ at this time, which had a small increase compared with the first time, but this scan stage is the compaction stage, because it can be seen from the figure that the red main large-volume fracture at the bottom has a significant reduction in the second scan. At the same time, under the scorching temperature of 400 °C, the coal sample can be seen from the outer surface in Figure (b) that the coal body is severely damaged and a large number of lamellar fractures appear, which will make part of the damaged coal sample fall off and fracture when stress was applied and form a fracture structure. When the third and fourth scans enter the plastic deformation stage, the deformation of the coal sample increases rapidly, the fractures extended continuously along the red high-temperature scorching fractures in the lower left corner to the upper right corner, and there is an obvious penetration between the internal fractures and the surface fractures, accompanied by a large number of newborn fractures. As the axial stress continues to increase, the coal sample is destabilized and is completely destroyed (fifth scan result), and the fractures extend through from the bottom to the top, forming a complex network of fractures and a typical type of shear damage.

10.

3D fracture evolution of the C400 coal sample under different stress levels.

The over 15-fold increase in fracture volume from Scan 4 to Scan 5 (e.g., RC: 144.63 mm³ → 2260.13 mm³) provides quantitative evidence for the failure mode. Such a dramatic increase cannot be attributed to gradual crack propagation alone; instead, it indicates rapid through-going fracture coalescence at the postpeak stage. As shown in Figures –, the fractures become interconnected across the entire sample, forming macroscopic failure surfaces. Based on these observations, the failure mode of RC and C400 is characterized as shear-dominated through-going failure, while C200 exhibits a localized V-shaped shear fracture that does not fully penetrate the coal body, consistent with its smaller postpeak volume increase and retained residual strength.

4. Discussion

4.1. Fracture Density and Porosity Evolution Law

During an underground coal fire, the coal mass is subjected to both high temperature and compressive stress, leading to a dynamic evolution of fracture networks with changing stress and temperature. Therefore, in order to investigate the variation of fractures with temperature and stress during this process in depth, porosity and fracture density are used to characterize the effect of temperature on fracture development during the full stress–strain process. The calculated values of fracture density and porosity at each scanning stage are presented in Table (Section ).

The variation law of fracture density and porosity with strain is clearly shown in Figure , and combined with the data in Table , it can be seen that the coal samples with different temperature treatments all went through the compaction stage, elastic deformation stage, plastic deformation stage, and complete failure stage during triaxial loading. The change trends of porosity and fracture density of the RC coal sample and the C200 coal sample are the same. First, there is a slow decrease, mainly because the fracture closes and reduces the fracture volume. Then, there is a slow increase because the stress gradually increases, the generation of new fractures, and the expansion of original fractures, resulting in a small increase in fracture volume and the fracture area. Finally, a sudden rise occurs because the axial stress of the loaded coal body exceeds the peak and a penetrating fracture network appears, leading to a sharp increase in porosity and fracture density.

11.

Relationship between fracture density and porosity with strain for RC coal sample, C200 coal sample, and C400 coal sample.

Porosity, defined as the ratio of fracture volume to the total volume of the coal sample, increased with temperature. From the initial scan onward, higher temperatures resulted in greater fracture development. Compared to the porosity of the RC coal sample, the porosity of samples treated at 200 and 400 °C was four times and ten times higher, respectively. This indicates that high temperature enhances the penetration of preexisting fractures within the coal sample. In the second scan, the porosity of the RC coal sample and the C200 coal sample decreased by 0.0002% and 0.0003%, respectively, indicating that some of the microfractures were compacted in the compaction stage. For the C400 sample, the porosity increases slightly from Scan 1 to Scan 2 (from 0.0070% to 0.0079%), which contrasts with the compaction-induced porosity reduction observed for RC and C200. This counterintuitive trend reflects a competition between compaction and disintegration. After 400 °C treatment, the coal sample is severely damaged, with extensive loose thermal cracks and fragmented particles on the surface (Figure (b)). When stress is initially applied, compaction closes existing fractures (as evidenced by the reduced volume of the large red fracture in Figure ), while disintegration causes loose particles to spall off and microcrack edges to break, creating new small fracture spaces. For C400, disintegration slightly outweighs compaction, resulting in a net porosity increase. Thus, for the severely degraded C400 sample, the “compaction stage” no longer represents true compaction but rather localized early-stage failure. We refer to this as the “compaction-disintegration competition stage.” At the third to fourth scans, RC, C200, and C400 exhibit different degrees of increase, by 0.0006%, 0.0046%, and 0.0107%, respectively, indicating that during the new fracture rapid growth stage, the stress causes the loading capacity of the coal sample to decrease continuously, and the new fractures and thermally expanded fractures continue to penetrate and converge, resulting in an increase in the penetration degree of the fractures with increasing temperature. At the fifth scan, all coal samples were destabilized and destroyed with very few new fractures, and the coal samples started to slip and destroy along the through fractures, and the fracture volume increased dramatically in the scan results.

4.2. 3D Fractal Dimension

To ensure the reliability of the results, three parallel specimens were tested under each temperature condition. The relative standard deviation (RSD) of the calculated 3D fractal dimension among parallel specimens was less than 1.5%, indicating the good repeatability of the experimental results. The 3D fractal dimension calculated by the 3D box dimension method can better characterize the pore and fracture morphology and distribution of the 3D model, , so the 3D box dimension method was used to process the test results of coal samples treated at different temperatures. Figure (a) presents the 3D fractal dimension of the coal sample at different scan numbers, and Figure (b) shows its relationship with temperature after complete failure under load destabilization (fifth scan).

From Figure (a), it can be seen that the 3D fractal dimension follows the same trend as porosity and fracture density, and the three coal samples show different increasing trends of fracture variation with increasing temperature at 1–4 scans. The three fracture evolution stages are aligned with the characteristic points on the stress–strain curve (Figure ). The smooth development stage (Scans 1–3) corresponds to the compaction and elastic deformation stages. The slow growth stage (Scans 3–4) corresponds to the plastic deformation stage (from yield point to peak point). The rapid growth stage (Scans 4–5) corresponds to the postpeak failure stage. The stage division was further supported by the variation characteristics of the three-dimensional fractal dimension between successive scans. As shown in Table , only minor fluctuations in fractal dimension were observed during Scans 1–3 for all coal samples, indicating relatively stable fracture development. Between Scans 3 and 4, the increase in fractal dimension became noticeably larger, suggesting the onset of accelerated fracture propagation. A much greater increase was observed between Scans 4 and 5, indicating rapid fracture coalescence and connectivity development. Therefore, the stage boundaries were identified according to the observed transitions in fractal-dimension growth rate rather than solely by qualitative observation. The increase in fracture density, porosity, and 3D fractal dimension is consistent with the reduction in elastic modulus and peak strength observed in Section .

4. Calculation Results of 3D Fractal Dimension of Coal Samples at Different Temperatures.

Coal samples	First scan	Second scan	Third scan	Fourth scan	Fifth scan
RC	2.04369	2.03171	2.04816	2.09951	2.58319
C200	2.05955	2.04414	2.04791	2.12537	2.27471
C400	2.09401	2.09817	2.09741	2.21765	2.83251

To further evaluate the internal consistency among the fracture characterization parameters, a correlation analysis between the porosity and three-dimensional fractal dimension was conducted (Figure ). The results show that the relationship between the porosity (P) and the three-dimensional fractal dimension (D) can be well described by a quadratic function, D = 3.3259 + 0.9212P + 0.1621P ², with a coefficient of determination of R ² = 0.993. This strong correlation indicates that the increase in the fracture volume is accompanied by a corresponding increase in fracture-network complexity. Moreover, the abrupt increase in the three-dimensional fractal dimension during the rapid-growth stage is accompanied by a sharp increase in porosity, demonstrating that both parameters identify the same transition point in fracture evolution. This consistency provides quantitative support for the proposed three-stage fracture-evolution model. In particular, the C400 sample exhibits the highest fracture complexity and the greatest degradation of mechanical properties, indicating that the deterioration of the mechanical behavior is closely associated with the development and coalescence of internal fractures.

13.

Quantitative relationship between porosity and three-dimensional fractal dimension.

Analysis of the rapid growth stage (fifth scan) shows that the 3D fractal dimensions of RC, C200, and C400 are 2.58319, 2.27471, and 2.83251, respectively. The value first decreases from RC to C200 and then increases from C200 to C400. It should be noted that this observation is based on only three temperature points (room temperature, 200 °C, and 400 °C). The absence of intermediate temperatures (e.g., 100 and 300 °C) means that the exact evolution pattern between these points remains uncertain. Therefore, the trend shown in Figure (b) should be considered indicative rather than predictive. Future work with more temperature intervals is needed to establish a reliable quantitative relationship.

The lower 3D fractal dimension of C200 (2.27471) compared to RC (2.58319) does not indicate that its fracture network is necessarily simpler in terms of crack abundance. Instead, it reflects a reduction in geometric irregularity and directional complexity after 200 °C treatment.

As shown in Figure , the termination point of the fifth shear-slip damage of C200 is located at a dense fracture zone in the lower part of the fourth scan, and the through-going fractures do not traverse the entire coal body. This indicates that the fracture network in C200 is more localized, directionally aligned, and structurally simpler than that of RC, where natural fractures are randomly distributed and oriented.

In contrast, the RC sample contains abundant natural fractures (cleats, grain boundaries, and random microcracks) that are geometrically irregular and spatially dispersed, which naturally yield a higher fractal dimension.

For C400, extensive high-temperature treatment promotes the formation of highly branched, interconnected, and tortuous microcracks, resulting in the highest fractal dimension (2.83251) among all three samples.

Therefore, the 3D fractal dimension in this study is not a proxy for fracture volume, but a measure of fracture network complexity and spatial irregularity. The nonmonotonic trend (RC → C200 → C400) reflects the competition among natural fracture complexity, thermal-induced fracture regularization (dominant at 200 °C), and thermal-induced fracture branching (dominant at 400 °C).

C400 coal samples exhibited an increase because a high temperature weakens the mechanical properties of coal and rock masses, making the coal body more susceptible to damage during triaxial loading. This can be seen in Figure (b). The higher the temperature generated when an underground coal fire occurs, the easier it is for the area to collapse within the mining area, the more fractures there are, and the more developed the fracture network connected to the surface becomes.

The three-stage fracture evolution identified in this study provides practical guidance for coal fire prevention. In the smooth development stage (Scans 1–3, corresponding to compaction and elastic deformation), fractures grow slowly with limited connectivity, making early warning systems and preventive sealing of oxygen entry points effective. , In the slow growth stage (Scans 3–4, corresponding to plastic deformation), fractures begin to interconnect, providing a critical window for grouting and sealing operations using colloidal materials or gel foam technologies; , temperature monitoring points should be placed in fracture-dense zones to capture early thermal anomalies. In the rapid growth stage (Scans 4–5, corresponding to postpeak failure), through-going fracture networks form dominant oxygen supply pathways, requiring emergency fire-fighting measures such as inert gas injection. , The coupled mechanism between fracture evolution and fire hazards has also been discussed. Thus, identifying the current stage of fracture evolution in a coal fire zone can guide the timing and type of intervention, improving the efficiency of fire prevention and control strategies.

5. Conclusion

Underground coal fires represent complex geological processes, and the generation of coal and rock mass fractures will increase the hazard of underground coal fires. The dynamic evolution law of fractures during the loading process of three coal and rock masses was analyzed by a triaxial stress loading device and a CT scanner, and the relationship between the 3D fractal dimension and the treatment temperature after the complete destruction of the coal sample by load destabilization was analyzed. The main conclusions were obtained as follows.

• The basic mechanical parameters of high-temperature coal and rock masses are governed by temperature. For coal specimens under the same confining pressure, as the temperature increases, the compaction stage shortens, the elastic modulus decreases, and the peak stress first increases and then decreases. Meanwhile, the durations of the plastic deformation stage and the postpeak failure stage first decrease and then increase with increasing temperature.

• As the temperature increases, more fractures and pores develop on the surface of thermally treated coal, and diagonal shear damage becomes more pronounced. The number and width of fractures produced by high-temperature heating in the parallel laminae direction are more compared with those in the vertical laminae. High temperature mainly weakens the mechanical properties of coal and rock masses, and the main fractures develop along the top and bottom edges to form a through-going fracture network.

• With the increase in temperature, tiny new fractures sprouted in the high-temperature-treated coal at 400 °C in the compaction stage, and the increase in fractures in the high-temperature coal body was more obvious in the plastic deformation stage. In the postpeak residual deformation stage, fractures did not fully penetrate the coal body when the C200 sample was finally destroyed, resulting in the lowest porosity (3.14%). In contrast, the RC and C400 samples exhibited higher porosities (9.21% and 25.93%, respectively), with C400 showing the most extensive fracture network development after failure.

• Temperature significantly affects the evolution of the fracture density and porosity during the loading process. The development characteristics vary among RC, C200, and C400 samples, reflecting the combined effects of the thermal treatment and mechanical loading. On the basis of the 3D fractal dimension, which exhibited a trend of initial decrease followed by an increase with temperature, and combined with the analysis of fracture density and porosity, the fracture evolution of the loaded high-temperature-treated coal rock can be divided into three stages: the smooth development stage, the slow growth stage, and the rapid growth stage.

L.W.: Data curation, Formal analysis, WritingOriginal draft, Writingreview and editing. X.L.: Data curation, Formal analysis, WritingOriginal draft, Writingreview and editing. Y.X.: Data curation, Formal analysis, Writingreview and editing. W.W.: Data curation, Formal analysis. Q.Z.: Supervision, Methodology. H.X.: Conceptualization, Validation.

The authors declare no competing financial interest.