To simulate the general patterns of crack propagation in rock-like materials, Huang Bingxiang (HUANG, 2009) investigated the mechanical properties of cement mortar specimens with various mix ratios through a series of uniaxial compression tests. It was determined that the mix ratio of cement to fine sand = 1:3.5 produced a complete stress–strain curve most closely resembling that of natural rock materials in key aspects such as the elastic stage, plastic yielding near peak strength, and post-peak softening behavior. This ratio also yielded compressive strength, elastic modulus, and failure modes representative of common natural rocks like sandstone or limestone. Based on these findings, the study adopted a cement-to-fine sand ratio of 1:3.5 to ensure the specimens closely simulated the mechanical characteristics of natural rock masses in subsequent experiments. This study is based on similarity theory and employs similar material samples as substitutes for natural rocks. The substitution is achieved by adjusting the mix proportions, modifying the sample dimensions, and optimizing the molding process, while maintaining consistency in key mechanical properties.

Figure 1 illustrates the three-dimensional structure and positioning details of the sample. To meet the size constraints of the testing machine’s loading chamber, cubic cement-mortar samples with 300 mm edge length were fabricated. Each sample includes a centrally positioned through-hole measuring 6 mm in diameter and 12 mm in length. PVC pipes were embedded during casting to ensure the dimensional and positional accuracy of the injection hole.

This study explores fracture propagation under true triaxial stress conditions using a self-developed hydraulic fracturing system. Figure 2 presents the experimental setup. At the core of the system is an 80 cm cubic, thick-walled steel chamber, designed to maintain pressure integrity. A bolted cover plate with rubber gaskets and multi-stage sealing assemblies around the embedded loading rams ensure effective sealing. The system is equipped with cartridge heaters and fluid-line interfaces and is capable of sustaining confining pressures σ 3 up to 50 MPa. Hydraulic fracturing is simulated by precisely injecting fluid into the wellbore through a high-pressure fluid injection module. The robust chamber structure and advanced sealing design guarantee both operational safety and leak-tight performance during high-pressure experiments.

The hydraulic workstation is an integrated system designed for applying true triaxial stress and injecting high-pressure fracturing fluid. It enables the establishment of a controlled triaxial stress environment, execution of hydraulic fracturing, and maintenance of high-pressure sealing integrity.

The confining pressure loading system, supported by a four-column structural frame, applies horizontal stresses to the sample via dual lateral hydraulic rams. At the same time, the minimum principal stress is applied vertically through hydrostatic pressure, allowing for independent control of all three principal stress directions. True triaxial stress states are achieved through the coordinated application of hydraulic pressures   σ 1 and σ 2 , along with hydrostatic confining pressure σ 3 .

The loading rams deliver bidirectional pressure transmission, while the pressure vessel accommodates unidirectional stress transfer. The entire process is controlled via dedicated software, integrating hydraulic control with AE win-based acoustic emission monitoring. This allows for precise load application and real-time analysis of fracture dynamics during the hydraulic fracturing process.

The hydraulic fracturing experimental protocol in this study consists of four core components: acoustic emission (AE) monitoring, true triaxial confining pressure control, water injection flow regulation, and AE signal acquisition. Based on in-situ stress data from the 11,211 working face of the Kekegai Coal Mine, the experiment closely replicates actual underground stress conditions and simulates various stress states. Figure 3 illustrates the configuration of the AE sensors. In this setup, σ_H and σ_h represent the maximum and minimum horizontal principal stresses, respectively, while σ_v denotes the vertical stress. Fluid injection was performed at a constant, controlled flow rate.

The rock-like samples used in the true triaxial tests were designed to simulate the mechanical behavior of natural rock. Given the difficulty of processing deep, hard roof strata into full-scale test samples, a physical simulation approach was adopted. This method enables the controlled analysis of hydraulic fracturing effects on fracture propagation and the spatial distribution of AE energy within a laboratory setting.

The sample body is subjected to hydrostatic pressure from all directions, achieved by superimposing the confining pressure and the external cylinder loads. However, the overlapping intermediate portion-specifically, the reaction force from the rigid loading rod-must be deducted. Therefore, the calculation formula for the reaction force of the rigid loading rod is: f r e a c t i o n = π × 0.06 × 0.06 × σ v = 0.011304 σ v σ H = T H − 0.011304 σ v l 2 + σ v σ h = T h − 0.011304 σ v l 2 + σ v

Where: f r e a c t i o n is the reaction force of the rigid loading rod, in N; T H and T h are the servo cylinder loads in the σ H and σ h directions, respectively, in N; l is the side length of the specimen, in mm. 0.011304 is the empirical coefficient for rigid loading in this experiment.

Based on calculation, the servo cylinder load setting parameters corresponding to the stress states of the triaxial confining pressure schemes are presented in Table 1. Accordingly, the confining pressure loading for specimens in the triaxial rock mechanics experiments was performed.

Triaxial hydraulic fracturing experiments, which simulate in-situ stress conditions, are essential for understanding fracture propagation mechanisms and their coupled effects. These tests reveal that while stress conditions primarily control fracture initiation and growth, rock heterogeneity significantly influences fracture path deviation. By quantifying the influence of stress on fracture network morphology, the experiments offer valuable insights for optimizing fracturing parameters. Pressure and injection flow rate are identified as key diagnostic indicators that determine fracture initiation, propagation capacity, and geometry. In composite rock strata, fracture behavior varies distinctly due to differences in interface properties and material heterogeneity. Such experiments provide a unique opportunity to explore the fundamental mechanical interactions at lithological interfaces.

This physical similarity simulation experiment employed an acoustic emission (AE) monitoring system. The system utilized a MICRO-II EXPRESS acoustic emission monitor, developed by Physical Acoustics Corporation (PAC) of the United States, to monitor the energy released and the intensity of failure during specimen fracturing. The AE system comprised AEwin software, developed by Physical Acoustics Corporation (PAC), USA, and hardware components including AE sensors, preamplifiers, and an AE acquisition unit. The primary AE monitoring parameters configured for this experiment are listed in Table 2 below.

ABAQUS enables advanced simulation of hydraulic fracturing through fully coupled thermo-hydro-mechanical models, capturing fracture propagation via XFEM/cohesive methods and supporting complex heterogeneity, natural fractures, and adaptive meshing. It facilitates large-scale 3D modeling with high-performance computing and provides detailed post-processing of fractures and stresses for optimizing multi-stage stimulation. With superior nonlinear convergence and multiphysics integration over alternatives like FLAC3D or ANSYS, it is widely preferred for research and design of deep reservoir fracturing.

Fracture simulation approaches such as the dynamic lattice element method used in (Rizvi et al., 2020) offer an alternative micro-mechanical framework to model stress redistribution in heterogeneous rock.

As shown in Figure 12, laboratory hydraulic fracturing samples were modeled in the ABAQUS visualization module to simulate the roof rock specimens. The objective was to reproduce the modal response and stress characteristics observed in the physical experiments through numerical simulation.

To validate the material model, two-dimensional numerical simulations were performed using the same dimensions as the physical samples (300 mm × 300 mm). To improve accuracy near critical regions such as the fracture hole and corners, a refined mesh was applied, with element sizes gradually increasing away from the hole surface. Additionally, nodal singular elements were introduced at the crack tips to better capture stress concentration and enable accurate analysis of crack propagation paths.

The numerical simulations were conducted using the same physical parameters, Darcy-law seepage model, and loading conditions as in the laboratory experiments. To replicate observed failure modes, identical injection holes and pre-existing fracture paths were incorporated into the simulation setup. As shown in Figure 13, the simulated fracture propagation and failure patterns closely resembled those from the experiments, confirming the effectiveness of the Darcy model in reproducing key fracture behaviors.

However, natural rocks contain numerous non-uniform microfractures that are difficult to replicate numerically. In physical experiments, these microcracks undergo compaction, contributing to nonlinear deformation. In contrast, the numerical models-despite incorporating preset fractures-cannot fully capture this behavior. As a result, simulated deformations tend to be more linear and less pronounced, leading to an underestimation of actual specimen deformation.

Numerical simulation is employed to analyze stress redistribution following hydraulic fracturing, thereby evaluating its effectiveness. Based on the geological conditions of a 400 m extra-long working face, a simplified numerical model was established, as shown in Figure 14. The model, measuring 500 m × 50 m, features compressive stress boundaries. The upper 45 m represents the roof strata, while the lower 5 m corresponds to the coal seam. As illustrated in Figure 15, water is injected in segmented stages from designated injection points to promote fracture propagation.

To better visualize stress distribution during hydraulic fracturing, stress contour plots were generated and analyzed. Figure 16 compares the simulated stress distribution around fractures during water injection with the corresponding experimental results (Rizvi et al., 2020).

As shown in Figure 17, high-pressure water injection creates unloading zones at the fracture tips, leading to localized stress reduction and displacement in the surrounding coal and rock. This induces plastic deformation near the fracture tips, thereby weakening the local structure. Simultaneously, stress concentrations develop on both sides of the fracture due to the pressure exerted by the injected fluid, although the resulting stress levels remain lower than the injection pressure. This tip unloading effect promotes fracture propagation along pre-existing weak planes.

Ideally, fractures continue to extend until fluid injection stops. However, in this simulation, the assumption of a homogeneous rock mass limited control over fracture direction—highlighting the challenges of modeling fracture propagation in geologically complex conditions.

This stress redistribution and pore connectivity enhancement are also evident in thermal stress field studies conducted on underground power systems, as shown in (Ahmad et al., 2025a; Ahmad et al., 2021).

In addition, stress concentrations on both sides of the fracture during hydraulic fracturing introduce complexity to the overall stress distribution. If fracture propagation is not properly controlled, elevated stress levels near the working face may increase operational risks. Therefore, appropriate personal protection measures should be implemented, and, where feasible, real-time monitoring of fracture propagation and stress variations is recommended.

Although the duration of fracturing is relatively short, the induced stress changes primarily occur within the coal and rock mass. While these changes generally have limited immediate impact on resource recovery, the altered post-fracturing stress field can significantly influence subsequent mining operations.

To further evaluate these effects, stress variation profiles ahead of the roadway and geostress contour plots following cessation of water injection were generated. These analyses provide deeper insights into the redistribution of stress within the coal and rock mass and its potential implications for mine stability and safety.

Figure 18 illustrates the stress reduction in the coal and rock mass ahead of the roadway, with stress levels decreasing as they approach the roadway and gradually returning to their original state with increasing distance. This redistribution causes the surrounding rock to move toward the roadway, resulting in greater displacement near the roadway and lesser displacement further away. After hydraulic fracturing, the previously elevated stress on both sides of the fracture is significantly reduced, leading to a substantial drop in stress and localized plastic deformation.

These results confirm that hydraulic fracturing is an effective technique for pressure relief. The formation of unloading zones reduces the strain energy stored in the coal and contributes to the mitigation of coal and gas outburst risks. Additionally, the reduction in ground stress facilitates fracture development, allowing gas to migrate along fracture paths and be released into roadways, thereby lowering gas pressure within the coal seam.

However, hydraulic fracturing can also introduce new stress concentrations at fracture tips, creating localized high-stress zones that may pose safety risks during subsequent recovery operations. If such stress cannot be adequately controlled, the application of hydraulic fracturing may be limited.

At the early stage of fracturing, the surge in water pressure quickly generates a strain concentration zone, reflecting the transition of pore wall microfractures from elastic deformation to rupture. As pressure begins to fluctuate or decline, the strain field disperses radially, indicating the initiation of primary fractures and branching of secondary cracks. Once pressure stabilizes during the holding phase, the strain field becomes more uniform, with primary fractures extending along the maximum principal stress direction and secondary fractures forming a low-strain extension zone.

In homogeneous rock layers with poorly developed primary fractures, minor pressure fluctuations lead to a continuous and smooth expansion of the strain field. This results in a gentle strain gradient at the leading edge of the primary fracture and stable crack propagation. In contrast, rock layers with significant permeability variations or distinct structural surfaces exhibit asymmetric strain field distributions under fluctuating pressure. In such cases, abrupt jumps or bifurcations occur within high-strain zones, indicating that fractures either change direction upon encountering barriers or follow weaker structural planes. These zones often feature strain concentration and clustering of secondary cracks, closely influenced by local geological heterogeneity.

Following multiple fracturing events within a single borehole, the strain field evolves into a multi-core, radiating, and interwoven network. The interaction between primary and secondary cracks forms a globally weakened zone, reflecting the combined effects of water-induced softening and crack expansion. During the pressure-holding stage, constant-flow water injection facilitates deeper strain penetration, further reducing rock mass strength through fluid–solid coupling. After support installation, near-surface strain is redistributed, and the strain gradient near the borehole decreases.

To prevent premature shallow crack closure, dynamic monitoring of stress redistribution is necessary. Overall, the observed strain characteristics confirm the effectiveness of hydraulic fracturing in generating a controllable fracture network within the roof rock layer.

The horizontal principal stress difference ( Δ σ ) plays a critical role in determining fracture geometry and propagation behavior during hydraulic fracturing. When Δσ exceeds 9 MPa, fractures tend to grow asymmetrically at angles of 65°–75° relative to the maximum horizontal stress ( σ H ). In contrast, when Δ σ is approximately 6 MPa, fractures exhibit more symmetrical, wing-shaped patterns. This trend was confirmed in the high-stress environment of the 11,211 working face at the Kekuai Coal Mine ( Δ σ > 9 MPa), where the average fracturing pressure for A/B-type boreholes (oriented perpendicular to σ H ) was significantly higher than that of -type boreholes (parallel to σH). A larger Δ σ suppresses secondary fracture branching, guiding fracture propagation along the σ H direction, and elevates the required injection pressure by 17%–22% due to increased resistance from rock tensile strength. These findings highlight the necessity of accurate stress field assessment in fracturing design.

In-situ acoustic emission (AE) monitoring further reveals that the confining pressure ( σ 3 ) significantly influences damage evolution in roof rock masses. When σ₃ increases from 11.47 MPa to 14.90 MPa-simulating deeper geological conditions-the number of AE events increases by 28.15%, and cumulative energy release rises by 31%. This indicates that under higher confining pressure, micro-fracture nucleation, propagation, and coalescence are accelerated. Although fracture opening is constrained, energy is released more intensely through concentrated micro-fracturing.

Field borehole observations provide spatial validation: post-fracturing, newly formed fracture networks were identified at depths between 12 and 42 m. Shallow sections (12–20 m) exhibited dense, hydraulically connected fractures, while deeper zones (20–42 m) showed stable main-fracture extension. This zonal pattern aligns with the radial distribution of AE events, indicating a depth-dependent damage mechanism: shallow regions are governed by shear–tensile hybrid failure along weak planes, whereas deeper regions are dominated by tensile-driven fracture propagation.

A fluid–solid coupling numerical model developed using ABAQUS reveals the dual stress redistribution effects induced by hydraulic fracturing. Specifically, it highlights the formation of unloading zones at fracture tips and stress concentration zones on both sides of the fractures. These findings have two key engineering implications.

First, The in-situ monitoring at borehole B7 critically validates the efficacy of the high-flow injection scheme. The detection of 15 L/min instantaneous flow in an adjacent anchor-bolt hole at 6-m depth confirms successful hydraulic connectivity and provides field-scale confirmation of the laboratory-derived similitude principles for fracture penetration rate. This significant flow not only indicates the achievement of a fracture radius exceeding 10 m but also demonstrates the creation of a highly conductive flow pathway within the stimulated fracture network. The results validate that the 160 L/min injection protocol effectively overcame permeability barriers, enhanced subsurface fluid migration, and corroborated the predictive models for controlled fracture propagation.

Second, the results inform improvements in fracturing strategy. To minimize the risk of stress-induced dynamic events, a segmented fracturing approach with 3-m intervals was implemented. This method effectively reduced localized stress concentrations. During the pressure-holding phase, pressure fluctuations remained within 5%, indicating quasi-static fracture propagation and a lowered risk of impact-related ground pressure.

Similar thermally-driven fluid–solid coupling behavior in subsurface materials has been reported in (Ahma et al., 2019), where cyclic thermal loading significantly influenced subsurface stress fields and fracture evolution in granular backfills.

This study suggests that lithological interface heterogeneity may play a crucial role in determining fracturing efficiency. Field observations from -type boreholes appeared to exhibit low-pressure fracturing characteristics, potentially indicating a natural weakening effect at the interface between the immediate roof siltstone and the underlying sandstone. These observations seem to align with numerical simulation results derived from interface cohesion parameters, implying that weak interlayer bonding could significantly reduce the pressures required for both fracture initiation and propagation.

In light of these findings, the research team explored the optimization of borehole geometry—including dip angle, azimuth, and penetration depth—to leverage the inferred interface weakening effect and potentially guide fracture propagation along targeted strata or interfaces. This approach, rooted in geomechanical analysis, may offer a pathway toward layered, sequential, and controlled collapse of hard roof strata, which could improve both the efficiency and safety of fracturing-based weakening interventions.

Looking forward, the study underscores a closed-loop technical framework integrating multi-physics monitoring, numerical simulation, and real-time field control. The process spans from micro-damage evolution to macro stress-field reconstruction, culminating in dynamic parameter adjustment during testing. This integrated approach offers an adaptive, science-based strategy for managing fracturing in complex geological conditions. Such a multi-scale and multi-physics methodology might not only help verify the effectiveness of hydraulic fracturing in challenging roof conditions but could also provide a adaptable and reproducible strategy for future engineering applications.

Future AE monitoring systems could be augmented with AI-driven signal classification methods, similar to those used in subsurface corrosion detection shown in (Chungui et al., 2021).