Experimental Investigation of Fracture Behavior in Coal-Seam Hard Roofs Using Different Fracturing Fluids

Abstract

In fully mechanized mining faces with large mining heights, thick and hard roofs present significant challenges, including extensive overhang areas, difficult roof control, and frequent roof failures. Hydraulic fracturing is a crucial technique for roof weakening and mine pressure control, and the performance of fracturing fluids directly determines the effectiveness of pressure relief. This study conducted true triaxial hydraulic fracturing experiments using three media: clear water and low-viscosity and high-viscosity fracturing fluids. Fracture propagation patterns under varying media and roof strength conditions were systematically investigated through acoustic emission (AE) monitoring, pump pressure analysis, and rock strain measurements. The results show that both fracturing fluid properties and roof compressive strength significantly influence hydraulic fracture initiation, AE characteristics, and ultimate fracture morphology. Compared to conventional clear water, high-viscosity fracturing fluid exhibits superior performance in fracture initiation efficiency (34% higher peak pressure), propagation intensity (3.7 times more AE energy), and influence range (34% greater fracture length). These advantages make it particularly suitable for hard roof conditions requiring precise fracture management. The results provide a theoretical foundation for optimizing hydraulic fracturing parameters in hard roof control engineering applications.

1. Introduction

The progressive mechanization of mining operations has led to a substantial expansion in coal production capacity, along with accelerated advancement rates of working faces, intensified mining activities, and continuous expansion of mining areas [1,2,3]. However, these developments have been constrained by persistent challenges in roof control for fully mechanized mining faces, particularly the pronounced manifestation of mine pressure in large-mining-height operations, which has significantly impeded the advancement of large-mining-height coal extraction technologies [4,5]. These issues are especially acute in conditions with thick, hard roof strata, where mining activities induce substantial disturbances to overlying formations. Such conditions frequently result in severe roadway deformation and recurrent roof failures, including working face pressure frame incidents, posing serious threats to both mining equipment and personnel safety [6,7].

The hard roof stratum, characterized by its high integrity and bearing strength, typically resists the development of joint fissures and weak surfaces when disturbed by underlying thick coal seam extraction. This geological behavior often results in extensive roof overhang at the working face. When the overhang reaches a critical size, the roof may collapse suddenly and dramatically, potentially causing rock bursts [8]. Hydraulic fracturing technology addresses this challenge by utilizing high-pressure fluid to create controlled fracture networks, effectively weakening the roof structure and promoting gradual, multi-layered collapse in the goaf area [9,10]. This technique has emerged as a critical method for both roof weakening and mine pressure mitigation in coal seam operations [11,12,13]. The effectiveness of hydraulic fracturing in relieving pressure is influenced by a combination of geological and engineering factors [14,15]. Key geological factors include the structure of the rock mass, the distribution of in situ stress, the strength of cemented natural fractures, and the mechanical properties of coal and rock. Engineering considerations involve borehole design and the specific parameters of the fracturing process. These factors collectively govern fracture propagation dynamics. Optimal fracturing outcomes consequently require comprehensive understanding of roof characteristics and fracture development patterns.

Numerous studies have studied the mechanisms of hydraulic fracture propagation in coal seam roofs by employing laboratory-scale physical experiments with theoretical analysis and numerical simulation. Wasantha et al. [16] developed a fully coupled hydro-mechanical model to examine fracture closure behavior, comparing single-stage versus multi-stage fracturing under varying in situ stress conditions. Liang et al. [17] developed a two-dimensional model to analyze stress fields, introducing a new rock failure criterion that considers multiple natural fractures. This model provided an analytical method to predict the initial fracture pressure during horizontal staged fracturing in fractured reservoirs. Li et al. [18] introduced a three-dimensional discontinuous fracture network model that combined computed tomography (CT) scan data with nonlinear mechanical-seepage theory. Their model enables detailed analysis of the key factors that affect fracture propagation across coal/rock interfaces.

Laboratory-scale physical simulations, particularly true triaxial hydraulic fracturing experiments, have proven invaluable for investigating fracture morphology and propagation mechanisms in coal seam roofs [19,20,21,22]. Daneshy [23] demonstrated through layered media experiments that interlayer cementation strength predominantly controls fracture height, while coal-roof lithological differences showed negligible impact. Jiang et al. [24] conducted hydraulic fracturing experiments using coal and rock samples with different lubrication states, and the results showed that the initiation behavior of interfacial fractures was mainly controlled by vertical stress and the interfacial friction coefficient. Silva et al. [25] pointed out that differences in rock physical properties and ground stress can affect the complexity of fractures, especially at weakly cemented interfaces. Gao et al. [26] conducted hydraulic fracturing experiments on simulated coal seams with roof and floor under different stress conditions and found that in situ stress state strongly controls fracture orientation and propagation. However, most physical simulation experiments rely on splitting the fractured samples for visual inspection and using tracers to identify fracture surfaces, making it difficult to obtain their complete three-dimensional morphology.

To address this issue, acoustic emission (AE) monitoring technology has been used for fracture monitoring in hydraulic fracturing experiments. By tracking the formation process of the fracture network in three-dimensional dynamic through acoustic emission event points, the initiation and propagation processes of hydraulic fractures have been clarified to a certain extent [27,28]. CT tomographic technology has also been used in fracture morphology monitoring [29,30], which can obtain three-dimensional fracture distribution morphology with the help of three-dimensional reconstruction technology, but this technology is limited by sample size.

Fracturing fluid serves as the critical medium in hydraulic fracturing operations, with its primary functions encompassing pressure transmission, fracture initiation, proppant transport, and friction reduction between the fluid and rock formations [31]. The performance of hydraulic fracture propagation is highly influenced by fracturing fluid properties, including viscosity, frictional resistance, and fluid loss [32]. Consequently, optimal selection of fracturing fluids can substantially enhance operational efficiency. In current practice, clear water is widely adopted as the fracturing medium for roof treatment in coal mines due to its abundant availability. However, when clear water is used for hydraulic fracturing to weaken the roof strata, its inherently high frictional resistance leads to significant pressure losses within the fracturing pipeline [33]. This results in two major technical challenges: inadequate bottom-hole pressure and excessive equipment pressure during water fracturing. Furthermore, the restricted fracture length fails to meet the control requirements for diverse coal seam roof conditions [34,35].

To overcome the limitations of clear water fracturing in controlling coal seam roofs, the slippery water system from the petroleum industry has been adapted to develop specialized fluids with improved viscosity and lower friction, leading to more effective stimulation [36,37]. Despite advances in coal seam roof fracturing, the influence of fracturing fluid properties has not been extensively addressed in physical experiments. Current research on how fluid viscosity affects fracture propagation is largely based on theoretical models and numerical simulations, with limited support from controlled physical testing [38,39]. Moreover, few studies have systematically compared traditional water-based fracturing with slippery water-based fracturing techniques. This gap suggests two key directions for future research: understanding the mechanistic influence of fluid viscosity in hydraulic fracture propagation through coal seam roof strata and identifying optimal fluid parameters for diverse geological conditions.

To address these challenges, this study performed physical simulations using true triaxial hydraulic fracturing equipment under varying fracturing fluids and roof strength conditions. Three fracturing fluid types including clear water, low-viscosity fracturing fluid, and high-viscosity fracturing fluid were tested, combined with two distinct roof strength levels. Fracture propagation characteristics were systematically evaluated and quantitatively characterized through AE monitoring, real-time pump pressure curves, rock strain distribution, and post-experiment fracture morphology. This work elucidates the fracture propagation mechanisms of novel fracturing fluids in coal seam roofs with heterogeneous mechanical properties, providing theoretical insights for enhancing hydraulic fracturing models and field parameter optimization.

2. Materials and Methods

2.1. Specimen Preparation

The Shendong mining area is situated at the tri-province boundary of Shanxi, the Inner Mongolia Autonomous Region, and Shaanxi in northern China. Characterized by thick, hard roof strata, this region faces significant mining challenges of high ground pressure manifestations, severe roof subsidence, and support difficulties in fully mechanized long-wall faces. To mitigate these risks, this study conducted hydraulic fracturing experiments specifically targeting roof weakening in the Shendong mining area. The mechanical properties of the coal seam roofs are summarized in

Table 1. The mechanical properties of the coal seam roofs.

Mine Name	Level	Young’s Modulus (GPa)	Poisson’s Ratio	Uniaxial Compressive Strengths (UCS) (MPa)
				Maximum	Minimum	Average
Shangwan Mine	Coal seam 22	8	0.2	50.9	18.6	40.8
Shigetai Mine	Coal seam 22	-	-	45.47	9.77	30.41
Buertai Mine	Coal seam 42	-	-	69.5	30.0	50.2
Liuta Mine	Coal seam 22	22.4	0.22	71.5	20.8	39.7
Cuner Mine	Coal seam 31	1.23	0.16	49.0	24.9	43.7

.

To investigate fracture propagation characteristics and AE responses during hydraulic fracturing in the Shendong mining area, this study employed scaled physical simulations. Based on the UCS of the local coal seam roof, two sets of specimens with distinct UCSs were prepared using cement and 20-mesh quartz sand as analogous materials, following established similarity principles. These cubic specimens (300 mm × 300 mm × 300 mm) were designed to replicate low-strength (14.1 MPa UCS) and high-strength (24.4 MPa UCS) roof conditions. A true triaxial hydraulic fracturing system was utilized to simulate fracture propagation under varying fracturing fluids. The mechanical parameters of the specimens are detailed in

Table 2. The mechanical parameters of the specimens.

Group	Material Mass Ratio (Sand: Cement)	UCS (MPa)	Poisson’s Ratio	Young’s Modulus (GPa)	Density (g/cm3)
1	1:2	14.1	0.266	1.73	2.09
2	1:1	24.4	0.198	3.87	2.26

.

The specimen preparation followed a standardized procedure: (1) Mixing and Molding: Cement and quartz sand were homogeneously mixed at a predetermined mass ratio and poured into a pre-cleaned mold. (2) Borehole Formation: A drill tube was inserted into the mixture center and removed after cement curing to create a centralized borehole. (3) Fracturing Tube Installation: A 12 mm diameter stainless steel fracturing tube was epoxy-sealed into the borehole, ensuring hydraulic isolation. (4) Configuration: All specimens featured a consistent fracturing borehole depth of 250 mm, with a 20 mm unfilled section at the borehole bottom to initiate fractures.

Three fracturing media were employed in this study: clear water (laboratory tap water as the control group), fracturing fluid A (0.05% slippery water-based solution), and fracturing fluid B (0.1% slippery water-based solution). Compared to plain water, both fracturing fluids exhibited higher dynamic viscosity and reduced frictional resistance. The rheological properties of all three media were measured at standard conditions (20 °C, 1 atm), with detailed viscosity data provided in

Table 3. Experimental design.

Number	Fluid Properties	Dynamic Viscosity (×10−3 Pa·s)	Injection Rate (mL/min)	Compressive Strength (Mpa)	In-Situ Stress (Mpa)
					σh	σH	σv
#1	Clean water	1.01	53	14.1	3.2	5.7	4.0
#2	Low viscosity	4.2		14.1
#3	High viscosity	12.1		14.1
#4	Clean water	1.01		24.4
#5	Low viscosity	4.2		24.4
#6	High viscosity	12.1		24.4

.

2.2. Experimental Equipment

The true triaxial hydraulic fracturing experimental system (Figure 1 and Figure 2) consists of three integrated subsystems: a true triaxial servo loading system, a pump injection system, and a data acquisition system. Each subsystem plays a crucial role in replicating the complex conditions of subsurface fracture propagation under controlled laboratory conditions. The true triaxial servo loading system applies isotropic confining pressure via hydraulic oil with maximum principal stress of 25 MPa in three orthogonal directions and control accuracy of 0.25% F.S. in full scale. The system’s ability to uniformly apply pressure in multiple directions is critical for simulating the triaxial stress state in underground environments. The pump injection system has a maximum pressure of 50 MPa, maximum flow rate of 100 mL/min, and flow measurement accuracy of ±1% F.S. The pump pressure and flow rate are monitored and stored with a sampling interval of 10 ms. The data acquisition system includes an AE monitor and a dynamic strain monitor. The AE monitor has eight channels, with a frequency range of 1 kHz to 2.5 MHz and accuracy of ±0.01% F.S. The dynamic strain monitor has 64 channels with a maximum sampling rate of 1 kHz. The spatial arrangement of strain bricks and AE sensors is illustrated in Figure 3. These monitoring measurements are carefully designed to capture the full spatial distribution of strain and acoustic activity during the experiment.

2.3. Experimental Procedures and Parameters

The hydraulic fracturing test procedure is as follows: (1) Specimen Preparation and Instrumentation: Position the specimen in the fracturing chamber, mount eight AE probes on all four specimen sides with coupling agent, and connect strain gauges to the strain monitoring system; (2) Fracturing Fluid Preparation: Add red tracer dye to the fracturing fluid for fracture visualization; (3) Confining Pressure Application: Apply triaxial confining pressure to predetermined values (σ_h, σ_H, σ_v); (4) Fracturing Initiation: Simultaneously activate pump injection system (constant flow rate: 53 mL/min), AE monitoring system, and strain monitoring system; (5) Fracture Propagation Monitoring: Continue injection until either fracturing fluid breakthrough occurs or pump pressure stabilizes (±2% fluctuation over 30 s); (6) Termination Criteria: Physical fracture is confirmed when pump pressure drops abruptly to the minimum principal stress level or hydraulic fractures fully propagate through the specimen; then, immediately shut down all monitoring and injection systems.

The experimental parameters (Table 3) were determined through dimensional analysis using similarity theory, with reference to field conditions at the Shendong mining site. The specimen size is 300 mm and the segment spacing in engineering is 30 m, so the geometric similarity ratio is 1:100. The similarity ratios of compressive strength and ground stress are both 1:2. The complete set of scaled parameters maintains consistency with the Buckingham π theorem requirements for hydraulic fracturing simulations [40,41].

3. Results and Discussion

3.1. Analysis of Pump Pressure

Figure 4 presents the pump pressure curves of high- and low-compressive-strength specimens under varying fluid medium conditions. All curves exhibit three distinct stages: pressure rise stage, pressure drop stage, and pressure stabilization stage. Fracturing is considered complete when the pump pressure reaches a minimum and stabilizes, indicating fracturing fluid leakage.

In the first stage, the fluid is injected into the pressure borehole, causing the pressure in all curves to rise until reaching its peak. Subsequently, a sharp pressure drop occurs as cracks initiate and propagate within the specimen. During this stage, crack propagation accelerates rapidly, often penetrating the specimen within a very short time, leading to a drastic decline in pump pressure. The dotted line in Figure 4 represents a least-squares fit of the first-stage curve, with its slope indicating the pressure rise rate per unit time. A steeper gradient corresponds to a shorter time required for the pressure to peak, facilitating the attainment of fracture initiation pressure in the roof strata. The numbered points in Figure 4 denote pressure peaks, which reflect the fracture initiation pressure. As illustrated in Figure 4a, Specimen #1 exhibits a peak pressure of 8.12 MPa, while Specimens #2 and #3 show peak pressures of 7.10 MPa and 9.28 MPa, respectively. The pressure rise rates, ranked from smallest to largest, are 0.10 MPa/s (Specimen #1), 5.52 MPa/s (Specimen #2), and 5.55 MPa/s (Specimen #3).

Compared to Specimens #1 and #2, #3 exhibits the highest peak pressure and pressure rise rate. This suggests that high-viscosity fracturing fluid not only elevates fracture initiation pressure but also accelerates the pressure rise stage, which facilitates rapid fracture propagation in hard roof strata. In contrast, Specimens #2 and #3 display a distinctly different pressure behavior from Specimen #1. After peaking, their pump pressure curves show repeated fluctuations—characterized by rapid decreases, increases, and decreases—before stabilizing. This phenomenon may result from the initial fracture opening, which releases pressure and causes fluid to leak along existing fractures. The fracture tip may temporarily close and then reopen under continued pumping. Stability is reached when fluid loss equals the injection rate, allowing the fracturing process to stabilize. The low viscosity of clear water in Specimen #1 increases filtration loss, prolonging the time required to reach equilibrium with the injection flow. Consequently, this specimen demonstrated the slowest pressure decline.

Figure 4b presents the pump pressure curves for specimens with higher compressive strength. The peak pressures, in ascending order, were 8.04 MPa (Specimen #4), 8.23 MPa (Specimen #5), and 10.80 MPa (Specimen #6). High-viscosity fluids may promote gradual and sustained opening of roof fractures, thereby extending the fracture initiation time and reducing the pressure rise rate. As a result, the injected fluid volume is primarily consumed in fracture volume expansion rather than in rapid pressurization. Peak pressure increases by 34%. The corresponding pressure increase gradients ranged from 4.75 MPa/s (Specimen #6) to 13.80 MPa/s (Specimen #4), with Specimen #5 exhibiting an intermediate gradient of 5.46 MPa/s. Our experiments reveal that increasing fluid viscosity leads to a progressive rise in peak pressure, accompanied by a more gradual reduction in pressure gradient. This behavior demonstrates that high-compressive-strength roof strata follow patterns like their lower-strength counterparts but with significantly more pronounced peak pressure elevation. The fluid medium’s influence on fracturing pressure appears amplified by the roof’s high-strength characteristics. This phenomenon may be attributed to two key material properties of high-strength roofs: greater brittleness and lower permeability. These characteristics promote enhanced fluid pressurization and energy storage, ultimately increasing the fracturing pressure. Furthermore, under high-compressive-strength conditions, the injected fluid requires substantially more time to accumulate sufficient energy to reach the specimen’s rupture pressure compared to low-strength conditions.

3.2. Analysis of Characteristic Parameters of AE

AE monitoring serves as a widely adopted technique for characterizing hydraulic fracture propagation at the laboratory scale [42,43]. To investigate the fracturing processes across different specimens, we analyzed three key AE parameters: amplitude, energy, and cumulative energy (Figure 5). Among these, AE energy represents a particularly significant parameter as it directly correlates with the extent of internal rock damage. The unit of AE energy is mV×ms, which is calculated by integrating the voltage signal over time. Typically, high-energy AE events correspond to macroscopic fracture formation or significant structural failure, while low-energy events are associated with microcrack initiation or localized damage [44,45,46]. This energy-based classification enables quantitative assessment of damage progression during the fracturing process, providing valuable insights into the specimen’s failure mechanisms.

Figure 5 presents a comparative analysis of AE characteristics during hydraulic fracturing of specimens with varying UCSs and under different fluid media. The monitoring reveals three distinct stages of AE activity: (1) Initiation Stage: AE signals become active immediately upon fluid injection, and both amplitude and energy show gradual increase corresponding to rapid pump pressure rise; (2) Fracture Propagation Stage: When fluid pressure reaches the specimen’s rupture threshold, instantaneous surge in AE peak amplitude and energy occurs. Vigorous AE activity indicates active crack propagation and specimen failure; (3) Stabilization Stage: The pump pressure plateaus with corresponding decrease in AE signal activity observed.

Figure 5a presents the amplitude, energy, and cumulative energy curves of AE events for Specimens #1–#3 during hydraulic fracturing. The data show a clear viscosity-dependent trend: higher fluid viscosity leads to increased AE activity, greater event amplitude, and higher energy release. The cumulative energy increases from 1151 mV×ms (#1) to 2808 mV×ms (#2) and 4246 mV×ms (#3), and the average amplitude increases from 168 mV (#1) to 195 mV (#2) and 955 mV (#3). Cumulative energy and average amplitude increased by a factor of 3.7 and 5.7, respectively. Clear water fracturing (#1) exhibited the lowest AE activity and fewest events. Specimen #2 exhibited two distinct bursts of concentrated AE activity, with a longer AE duration and roughly twice the number of AE events compared to Specimen #3. However, the amplitude and energy of the AE events in Specimen #2 were relatively lower. This pattern indicates more dispersed, uniform fracture propagation in fracturing fluid A. In contrast, Specimen #3 showed temporally concentrated AE events, predominance of high-energy signals, and more vigorous local fracture propagation. These contrasting behaviors highlight the fundamental difference in fracture mechanisms between low and high-viscosity fracturing fluids.

Figure 5b displays the amplitude, energy, and cumulative energy curves of AE events for Specimens #4–#6. The viscosity-dependent AE characteristics observed in these high-strength specimens mirror those shown in Figure 5a, which demonstrate that both the fracturing fluid viscosity and UCS of the formation have significant effects on fracture propagation. Progressive increase in AE event amplitude with rising fluid viscosity corresponds to enhancement in energy release per event and cumulative energy growth. A marked reduction in overall AE activity levels, significant decrease in both AE event counts (approximately 55% reduction), and substantially lower cumulative energy (from 8205 to 5705 mV×ms in all three) are observed for high-UCS specimens compared to lower-strength specimens. These observations suggest that enhanced compressive strength fundamentally alters fracture propagation mechanics. Under identical pumping conditions, expansion of hydraulic cracks is limited. The combined effects of fluid viscosity and rock strength demonstrate a transition from distributed microcracking to localized macro-fracture development as the mechanical strength increases.

3.3. Analysis of the Strain and Hydraulic Fracture Width

The experimental setup incorporates a strain monitoring system with four embedded strain gauges to collect triaxial strain data during the fracturing process. The principal stress of the Y direction is defined as maximum horizontal principal stress (σ_H), and the principal stress of X direction is defined as minimum horizontal principal stress (σ_h). Consistent with fracture mechanics theory, hydraulic fractures propagate preferentially along the Y-direction (σ_H). Consequently, we utilized the minimum principal strain (ε_X) measurements to quantitatively characterize hydraulic fracture width development, as this strain component directly reflects fracture opening perpendicular to the primary propagation direction.

Figure 6 presents the temporal evolution of minimum principal strain (ε_X) in the fracture width direction for all six specimens. The strain evolution exhibits three distinct stages corresponding to different fracturing mechanisms: (1) Pre-fracturing Stage: Encompasses initial monitoring and borehole pressurization. It shows minor strain fluctuations attributable to the system initialization artifacts and operational variances in fluid injection, which has less relevance for fracture propagation; (2) Fracture Initiation and Propagation Stage: The fluid pressure keeps accumulation within the borehole and the rock formation, the fracture begins to initiate and expand, and the strain shows a downward trend (ε_X of #2 decreases by 2.5 × 10⁻⁵); (3) Fracture Closure Stage: a sudden pressure release and strain curve rebound (ε_X recovery of #2 is 30%) caused by the fracture runs through the specimen and then the fracture progressively closes. The stage transitions are particularly evident in high-strength Specimens #4–#6, where the strain rebound magnitude exceeds that of low-strength specimens by approximately 50%, indicating more pronounced potential energy of elastic deformations.

Figure 6a reveals significant differences in the evolution of the minimum principal strain (ε_X) among Specimens #1 to #3. In Specimen #1, the strain variation is relatively subtle, with the minimum strain reaching approximately 3.6 × 10⁶. In contrast, Specimens #2 and #3 exhibit more substantial strain changes, particularly during the crack initiation and propagation stages, where the strain decline rate is higher. The minimum strains in Specimens #2 and #3 are both below 2 × 10⁷. After reaching their minima and entering the crack closure stage, the strain curves rebounds. These observations suggest that for low-strength specimens, fractures generated by clear water fracturing tend to be narrower, whereas those produced using low- and high-viscosity fracturing fluids exhibit greater fracture widths. This phenomenon may be attributed to the higher viscosity of the fracturing fluid, which increases fracture initiation pressure, thereby promoting wider fracture formation.

In Figure 6b, the strain variation of Specimen #6 (high-strength rock) is the most pronounced, exhibiting the steepest decline rate during fracture initiation and propagation, along with the smallest extreme strain value (<−4 × 10⁷). The strain extreme value of Specimen #6 is roughly twice that of Specimen #3, suggesting that the fracture width in the high-strength specimen is approximately double that of the low-strength specimen. When combined with the pump pressure variation analysis in Figure 4, this phenomenon may result from the high-strength rock enhancing the fluid’s pressure buildup effect, thereby increasing fracture initiation pressure. Furthermore, compared to Specimen #2, Specimen #5 demonstrates smaller strain curve fluctuations and a slower strain decline rate. This implies that the fracture width produced by the low-viscosity fracturing fluid decreases under the confining effect of the high-strength rock formation, whereas the high-viscosity fracturing fluid’s fracture width remains largely unaffected.

3.4. Analysis of AE Source Locations

When combined with the analysis of the fracturing medium, a progressive increase in location points is observed from clear water (Figure 7a) to fracturing fluid A (Figure 7b) and then to fracturing fluid B (Figure 7c) in the location map, accompanied by gradually intensifying acoustic emission activity. This demonstrates that the fracture propagation process induced by the fracturing fluid is more vigorous than that achieved with the clear water medium. Moreover, the positioning sites in Figure 7a are predominantly concentrated in small areas. In contrast, Figure 7b,c show an increasing number of positioning points near the boundary with broader spatial distribution, indicating that the new fracturing fluid increases the fracture extension area. These results clearly show that, compared to the clear water medium, the new fracturing fluid produces a greater number of fracture points, more intense acoustic emission activity, and more vigorous fracture propagation.

The roof strength significantly influences the propagation pattern and fracture characteristics. Based on the roof strength analysis, as the roof strength increases, the number of AE events gradually rises (Figure 7a–f). However, the energy associated with these additional AE events is lower, and most occur during the middle and late stages of AE monitoring. These observations suggest that crack propagation in high-strength roofs proceeds more slowly, with a higher frequency of rock fractures but less intensity. This implies that during hydraulic fracturing, fractures in high-strength roofs propagate at a slower rate and exhibit more frequent but less severe rock fracturing. This phenomenon may be attributed to the greater tensile strength and toughness of high-strength roofs, which increase resistance to fracture propagation. Consequently, more time is required to accumulate sufficient energy prior to rupture. Thus, hydraulic fracturing parameters should be optimized according to roof strength to ensure effective fracture propagation and enhance overall fracturing performance.

4. Fracture Morphologic Evolution

4.1. Fracture Morphology

After completing the triaxial hydraulic fracturing experiment, the specimens were dissected along the direction of the maximum principal stress. Upon splitting, the fracture morphology was recorded immediately to prevent the fracturing fluid’s water marks on the specimen surface from drying out and becoming unrecognizable. The hydraulic fracture morphology can be traced using tracers to investigate the macroscopic fracture propagation under different fracturing media and coal-seam roof-strength conditions. In this study, the vertical stress (σ_z) is applied in the Z-direction, the maximum horizontal in situ stress (σ_H) in the Y-direction, and the minimum horizontal in situ stress (σ_h) in the X-direction. Specimens #1 to #3 have a UCS of 14.1 MPa, while Specimens #4 to #6 exhibit 24.4 MPa. All specimens are subjected to the same fracturing fluid flow rate.

Figure 8 presents the physical specimen and three-dimensional reconstruction of hydraulic fracture morphology for Specimen #1. The fractures, initiated by clear water fracturing, originated from the borehole bottom. Primary fracture propagation occurred along the X-axis (minimum horizontal principal stress direction), completely penetrating the specimen radially from the borehole center. The hydraulic fractures breached both upper and lower specimen surfaces, forming a dominant vertical fracture with significant height extension. During upward propagation, fracture deflection occurred, resulting in an offset between the upper surface crack path and the borehole location. The overall fracture geometry approximates a rectangular form with dimensions of 24 cm (length) × 30 cm (width) and is oriented approximately 10° off the direction of the maximum principal stress direction (Y-axis).

As shown in Figure 9, Specimen #2, treated with fracturing fluid A, developed relatively simple hydraulic fractures. The fractures initiated near the borehole and propagated primarily under in situ stress control, ultimately forming a single vertical fracture aligned with the maximum horizontal principal stress direction (Y-axis). The fractures were predominantly concentrated around the borehole region, with their width slightly exceeding the borehole length. The fractures extended approximately 24 cm in length and were oriented at about 15° to the maximum principal stress direction. Due to rapid fracture penetration through both the XZ upper surface and XZ side of Specimen #2, significant fracturing fluid leakage occurred. This resulted in only localized fracture development near the borehole, with no hydraulic fracture extension into the lower strata beyond the borehole region.

Figure 10 presents the hydraulic fracture morphology in high-strength Specimen #3. The physical image reveals that the hydraulic fracture completely penetrates the specimen, with the fracturing fluid propagating from the borehole bottom to all four specimen boundaries. This resulted in an extensive fracture network primarily aligned with the maximum horizontal principal stress direction (Y-axis). The dominant fracture exhibits a rectangular geometry, measuring 30 cm in length and 32 cm in width, with a 10° deviation from the vertical direction. Notably, the fracture maintains perfect alignment (0°) with the maximum principal stress direction. Fracture propagation analysis shows multiple deflection events near specimen boundaries, leading to irregular surface fracture patterns. This behavior likely results from decreasing fracturing fluid pressure at the boundaries, requiring additional energy expenditure to overcome rock resistance and causing subsequent fracture path deviations.

Figure 11 presents the hydraulic fracture morphology of Specimen #4, which exhibits higher compressive strength. Results indicate that two main vertical hydraulic fractures were formed after fracturing with clear water, both symmetrically distributed about the borehole axis. The first fracture initiates from the bottom of the borehole and extends vertically downward, ultimately penetrating the lower boundary of the specimen. This fracture measures 18 cm in length and 15 cm in width, with an inclination angle of 55° relative to the direction of the maximum principal stress. Morphologically, the hydraulic fracture exhibits a trapezoidal shape, forming a single vertical fracture with significant propagation distance. The second fracture originates from the borehole and extends laterally toward one side of the specimen. It propagates vertically upward, fully penetrating the XY upper surface of the specimen, resulting in a single vertical crack with an approximately square morphology. Compared to the first fracture, this crack is shorter, measuring 15 cm in length, while its width matches the borehole length (15 cm). Notably, this fracture propagates nearly parallel to the maximum principal stress direction, with a deviation angle of 0°.

As illustrated in Figure 12, the hydraulic fractures in Specimen 5# using fracturing fluid A exhibit a relatively simple morphology, manifesting as a pair of butterfly-shaped single vertical fractures. These fractures initiate from the bottom of the borehole and propagate vertically upward, primarily confined to the upper section of the specimen. Their propagation direction aligns perfectly with the orientation of the maximum principal stress (0° deviation). The fractures measure 20 cm in length, with a width equivalent to the borehole length (15 cm). Notably, the fractures fully penetrated the XY upper surface of Specimen #4 yet induced only localized fracturing in the specimen’s upper portion. Critically, the fracturing fluid failed to infiltrate the lower rock formation, indicating limited vertical propagation.

As illustrated in Figure 13, the hydraulic fracture network in Specimen #6 (fracturing fluid B) exhibits relatively complex morphology, displaying an overall inverted triangular pattern. The fractures propagate upward from the borehole base, predominantly concentrating in the specimen’s upper section. Their propagation is governed by in situ stress conditions, primarily aligning with the maximum principal stress direction. The fractures are mainly clustered around the borehole periphery, with crack widths slightly exceeding the borehole length. The fracture length measures approximately 24 cm, forming an angle of about 25° with the maximum principal stress direction. Furthermore, the fracture morphology observed on the XY surface of the specimen indicates an initial slight deviation near the borehole, followed by continued propagation along the maximum principal stress direction under in situ stress control. Owing to the high-strength roof formation, the fracture only partially penetrated the XY upper surface of Specimen #3, resulting in localized fracturing confined to the upper section. The hydraulic fracture failed to propagate into the underlying strata.

4.2. Effect of Fracture Fluids on Crack Extension

Hydraulic fracturing relies on fracturing fluid pressure to overcome the rock formation’s fracture resistance. The viscosity of the fracturing fluid is a critical factor influencing fracture propagation morphology as it affects fluid loss and pressure transmission, thereby altering hydraulic fracture geometry and seepage area. Selecting an appropriate fracturing fluid is therefore essential in designing effective fracturing operation plans.

Table 4. Statistics of crack parameters.

Specimen	Crack Length (cm)	Crack Width (cm)	Angle with the Direction of σH (°)	Crack Shape Features
#1	24	30	10	Single vertical crack, turning occurs
#2	24	18	15	Single vertical crack with simple shape
#3	30	32	0	Single square crack, large in extent
#4	18 (15)	15 (15)	50 (0)	Two vertical cracks with complex shapes
#5	20	15	0	Single vertical crack with simple shape
#6	24	18	25	Inverted triangular cracks, turning occurs

presents the fracture parameters for two groups of coal seam roof specimens. By analyzing fracture morphology, this study investigates how fracturing media influence fracture propagation behavior during roof hydraulic fracturing. The results demonstrate significant differences in fracture propagation patterns when using different fracturing media. As indicated in Table 4, both fracture length and width exhibit a positive correlation with fracturing fluid viscosity. For low-strength roof specimens, fracture length increased from 24 cm (clear water) to 30 cm (fracturing fluid B). Similarly, high-strength roof specimens showed an increase from 18 cm (clear water) to 24 cm (fracturing fluid B). The crack length increases by 34%. This behavior occurs because high-viscosity fracturing fluids generate greater viscous resistance, resulting in reduced fluid filtration loss. Consequently, the net pressure at the fracture tip increases significantly, enhancing the fracture’s ability to penetrate hard roof formations. Furthermore, at constant flow rates, higher fluid viscosity leads to lower pressure loss along the fracture. This effect provides more sustained driving force for fracture propagation and promotes the formation of greater propagation area.

Hydraulic fractures generated by clear water fracturing exhibit greater complexity than those produced by viscous fracturing fluids. Specimens #2 and #5 (treated with fracturing fluid A) both developed single vertical fractures with simple geometries. The fracture pattern of Specimen #3 (using fracturing fluid B) was also relatively simple, while Specimen #6 displayed local deflections but maintained overall consistent propagation direction. In contrast, the vertical fracture in Specimen #1 (Figure 8) underwent significant directional changes during propagation, deviating from the original fracture path. Specimen #4 (Figure 11) developed two vertical fractures with a wider intersection angle and divergent propagation directions, resulting in more complex fracture networks. This phenomenon may be attributed to the enhanced pressure conduction capability of low-viscosity clear water, which activates more natural fractures and thereby forms complex fracture networks.

Combined with fracture morphology and AE analysis, the results demonstrate that during the fracturing process, the fracture initiation pressure of clear water is lower than that of viscous fracturing fluid, while producing more complex hydraulic fractures. The AE activity observed with viscous fracturing fluid is more intense than with clear water, and the resulting fracture propagation area is also larger compared to clear water.

4.3. Effect of Roof Strength on Crack Extension

The propagation of hydraulic fractures requires overcoming formation strength, making it essential to investigate the influence of compressive strength on hydraulic fracturing behavior. As presented in Table 4, comparative analysis of hydraulic fracture propagation in two coal seam roof strengths reveals significantly reduced fracture extension with increasing roof strength. This occurs because higher-strength rock formations require greater fracturing pressure for fracture initiation. Conversely, low-strength formations more readily reach initiation pressure, resulting in more extensive rock damage and wider fracture zones. Taking Specimens #2 (Figure 9) and #5 (Figure 12) as examples, both employed the same fracturing fluid, yet Specimen #2 exhibited greater fracture length and width than Specimen #5. Additionally, fractures in Specimen #2 extended to both specimen surfaces, while those in Specimen #5 only penetrated the upper surface. This difference likely stems from the elevated critical stress required for fracture propagation in higher-strength formations, which increases propagation difficulty. Among all specimens, Specimen #3 (Figure 10) demonstrated the most extensive fracture propagation, with fracturing fluid penetrating the entire specimen and the most intense fracturing process. This outcome reflects the combined effects of fracturing fluid B’s high viscosity and low frictional resistance working in conjunction with the roof’s low strength.

When the roof has high strength, the weak areas in the heterogeneous layer are more likely to become fracture initiation points, leading to fracture turning and branching, thus enhancing fracture complexity. In Specimen #4, the fracture branched, and the new fracture path deviated from the direction of the maximum horizontal principal stress, propagating toward the area with lower initiation pressure. Conversely, for low-strength roofs (as shown in Figure 8, Figure 9 and Figure 10), the critical stress required for fracture initiation and propagation is reduced, promoting rapid extension along the maximum principal stress direction without significant deflection to overcome rock resistance. These observations demonstrate that higher roof strength promotes fracture turning and branching, resulting in increased fracture complexity. Integrated analysis of fracture morphology and AE data reveals that increased roof strength results in (1) higher fracture initiation pressure, (2) enhanced fracture complexity, (3) reduced fracture propagation distance, and (4) decreased AE activity.

5. Conclusions

Currently, clear water is commonly used as the fracturing medium for controlling coal seam roofs. However, its low viscosity, high fluid loss, and high frictional resistance often lead to suboptimal fracturing performance in hard roof conditions. By contrast, slick-water fracturing fluid, with its higher viscosity and lower frictional resistance, shows greater potential for enhancing fracture propagation efficiency and improving control effectiveness. This study investigates the influence of different fracturing media on fracture propagation behavior through true triaxial hydraulic fracturing simulation experiments under varying fluid viscosities and roof strengths. A comprehensive analysis of fracture initiation characteristics and distribution patterns was performed using integrated monitoring techniques, including AE, pump pressure, strain measurement, and tracer methods. The principal findings are as follows:

(1)

Pump pressure monitoring showed that high-viscosity fracturing fluid not only increased the fracture initiation pressure but also shortened the pressure buildup time required, enabling faster fracture expansion in hard roofs. In comparison, the low viscosity and high fluid loss of water cause delayed fracture initiation and longer stabilization times for both injection pressure and flow rate. While high-strength roofs further prolong initiation time, high-viscosity fracturing fluid maintains superior efficiency under these conditions.

(2)

AE monitoring indicated that higher fracturing fluid viscosity significantly increased the activity, amplitude, and energy of AE events. Clear water produced the fewest and least active AE events. Although low-viscosity fluid generates more events, they exhibit greater temporal dispersion, indicating more uniform fracture propagation. Conversely, high-viscosity fluid induces concentrated events with intense energy release, suggesting localized vigorous fracture propagation. Additionally, increased roof strength correlates with reduced AE activity and constrained fracture propagation.

(3)

Strain monitoring results indicated that high-viscosity fracturing fluid produced significantly wider fractures than clear water. With increasing roof strength, low-viscosity fluid produces narrower fractures, whereas high-viscosity fluid maintains a relatively constant fracture width.

(4)

Fracture morphology analysis showed that clear water had lower initiation pressure and produced smaller propagation areas than high-viscosity fracturing fluid but led to more complex, branched fractures. A higher UCS of the roof resulted in higher fracture initiation pressure, greater fracture complexity, and a smaller propagation range.