Introduction

Coal remains the dominant primary energy resource in China and continues to play a critical role in national energy production and consumption1,2. Ensuring the stability of underground roadways is a fundamental prerequisite for safe and efficient coal extraction3,4. However, with increasing mining depth, underground roadways are subjected to complex in situ stress environments, pronounced stress redistribution, and degradation of surrounding rock mechanical properties. As a result, more than 90% of deep roadways experience severe deformation, roof collapse, rock bursts, and support failure, among which roof-related accidents account for over half of all roadway disasters5,6. Roof instability control has therefore attracted extensive research attention from both engineering and rock mechanics perspectives. Previous studies have focused on optimizing support structures7,8, modifying stress environments9,10, and improving the mechanical properties of surrounding rock masses11,12. Among these approaches, surrounding rock modification has proven particularly effective in enhancing roof bearing capacity, with grouting reinforcement serving as the most widely applied technique.

Grouting reinforcement improves the overall mechanical behavior of fractured coal and rock masses by increasing cohesion, internal friction angle, and stiffness13. By mitigating deformation, collapse, and water inrush hazards during excavation, grouting substantially enhances the safety and reliability of underground roadway construction and has thus been extensively applied in coal mining engineering14. Chen et al.15 proposed an integrated bolt–grouting support technique and optimized support parameters based on real-time monitoring of shallow and deep surrounding rock responses. Salimian et al.16 systematically investigated the shear behavior of grouted fractured rock joints. Jafarpour et al.17 developed a zeolite–cement-based grouting material to address potential environmental concerns associated with conventional cementitious grouts. Xiao et al.18 studied the zoned failure characteristics of surrounding rock and the mechanism and technology of graded grouting reinforcement. Wang et al.19 studied the migration law of grout in wedge-shaped boreholes and the characteristics of grouting reinforcement of surrounding rock. Xu et al.20 employed an enhanced numerical manifold method to investigate the combined effects of grouting range and timing on reinforcement effectiveness in fault rupture zones. Wang et al.21 proposed a whole-section bolt–grouting reinforcement method for loose and fractured surrounding rock, achieving significant stability improvements. Prior to large-scale engineering application, laboratory testing and physical similarity modeling are essential for evaluating grouting reinforcement mechanisms and effectiveness. Physical similarity modeling has been widely adopted to reproduce deformation, stress redistribution, and failure processes under controlled conditions. Wang et al.22 designed a grouting simulation apparatus to investigate the reinforcement performance of hardened grouted rock masses with varying diffusion radii. Ma et al.23 conducted horizontal lateral grouting model tests to analyze deformation repair mechanisms in operating tunnels.

Grouting is the method employed to improve the physical and mechanical properties of a fractured rock by injecting a solidifiable liquid material into fractures through a grouting pump or other means. As has been proved by engineering practices, grouting is an effective method to improve the strength, integrity, and stability of fractured rock24. Grouting exhibits strong adaptability and high controllability, enabling precise reinforcement for various geological conditions. It causes minimal construction disturbance and offers good durability. Combining safety and economy, grouting technology serves as an efficient means to enhance the bearing capacity and long-term stability of surrounding rocks25,26.

Existing research has yielded a wealth of achievements in the areas of grouting material properties, fracture rock mass reinforcement mechanisms, as well as numerical simulations and model tests. However, most existing studies tend to focus on single grouting parameters, singular reinforcement methods, or local surrounding rock response characteristics. Especially in deep mining conditions, where the roof structure is complex and highly fragmented, the surrounding rock bearing system is prone to instability. Further research is still needed to explore how to achieve overall reconstruction of the surrounding rock bearing structure through rational grouting structure design. Therefore, it is necessary to combine physical similarity simulation experiments with on-site engineering practices to systematically reveal the mechanism of how grouting reinforcement affects the structural stability of roadway surrounding rocks. In this study, roof grouting physical similarity model tests and field-scale industrial application were conducted for the 3806 fully mechanized top-coal caving face ( ZF3806 ) mining roadway of Shuiliandong Coal Mine. In response to the severe deformation of the coal–rock composite roof, a layered grouting reinforcement concept is proposed. By integrating physical modeling results with field observations, a support strategy centered on “shallow bolt grouting combined with deep cable grouting” is developed and validated. The study aims to elucidate the mechanical mechanism by which layered grouting reconstructs the roof load-bearing system and enhances roadway stability in deep mining environments.

Engineering background and experimental scheme

Engineering status

A project was carried out at Shuiliandong Coal Mine in China. The Shuiliandong Coal Mine primarily extracts the No. 4 coal seam, with a thickness ranging from 3.3 to 9.4 m and an average thickness of 6.4 m. The coal seam exhibits well-developed cleats. The ZF3806 working face is selected as the study area. Its basic roof mainly consists of siltstone and fine sandstone, while the immediate roof is dominated by mudstone. The working face has a strike length of 952 m and a burial depth ranging from 320 to 370 m, with an average mining depth of approximately 330 m. The roadway layout and borehole columnar section are shown in Figs. 1 and 2. Borehole coring was conducted to characterize the roof structure above the mining roadway. Owing to severe fragmentation of the top coal and the presence of approximately 3 m of top coal, coring commenced 2 m within the top coal. The results indicate that the top coal is highly fractured with loose cores, while the overlying 1 m mudstone layer is also significantly damaged. In contrast, the sandstone above exhibits progressively higher integrity, with no visible macroscopic fractures. These observations suggest that the fractured zone is primarily confined to the top coal and immediate roof, whereas the overlying sandstone serves as a relatively stable bearing stratum.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Roadway layout of ZF3806 working face.

Standard cylindrical specimens (50 mm × 100 mm) were prepared for mechanical testing. Uniaxial compression tests were performed using a 150 C rock mechanics test system ( RMT-150 C ). Rock samples are named using the format R1, R2, R3. Coal samples are named using the format C1, C2, C3. More than three sets of parallel experiments were conducted for two types of samples. The average uniaxial compressive strength of coal specimens was 14.31 MPa, while that of sandstone specimens reached 35.36 MPa. The mechanical parameters of the surrounding rock strata are summarized in Table 1.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Sample preparation and loading process.

Table 1 Mechanical parameters of surrounding rock.
Full size table

Physical similarity modeling scheme

A plane-stress physical similarity modeling system was employed, with overall model dimensions of 2.5 m × 2.0 m × 2.0 m. Uniformly distributed loading was applied to simulate overburden stress, using a self-developed lever-loading device capable of providing compensation loads ranging from 7.22 to 62.64 kN. Physical similarity simulation experiments involve creating similarity models of real-world objects and phenomena that are much smaller than the actual engineering scale, and reproducing phenomena on these models. The materials, dimensions, and stress states used in the similarity models must maintain similarity with the actual engineering situation. Once the geometric similarity ratio and bulk density similarity ratio between the model and the prototype are determined, the similarity scales of other physical quantities can be calculated based on these two ratios. Similarity constants were determined based on geometric and bulk density similarity requirements. Refer to previous studies27,28, the geometric similarity ratio was set to 1:50, the bulk density similarity ratio to 0.7:1, the time similarity ratio to 0.05, and the dynamic similarity coefficient to 0.007. Based on these constants and prototype parameters, the mechanical properties of each model stratum were calculated (Table 2).

Table 2 Mechanical parameters of physical model.
Full size table

The model was constructed using river sand as aggregate and cement–gypsum mixtures as binding materials. Mica powder was applied between layers to simulate bedding planes. Material mix proportions were calibrated through pre-tests to ensure consistency with target mechanical properties (shown in Table 3).

Table 3 Model ratio of similar materials.
Full size table

Excavation, support, and loading scheme

Based on the roof failure characteristics observed at the ZF3806 working face, a layered grouting technique was proposed to reinforce the roadway roof. A physical similarity model test was conducted to verify the effectiveness of the proposed layered grouting reinforcement method. To comparatively evaluate the control performance of the roadway under the original support scheme and that after layered roof grouting, two roadways (Roadway A and Roadway B) were arranged within the coal seam. Roadway B was designed to simulate the actual support configuration adopted in the ZF3806 mining roadway, whereas Roadway A was reinforced by layered grouting on the basis of the original support scheme, aiming to validate the rationality and effectiveness of the proposed reinforcement strategy. The cross-sectional dimensions of both roadways were 50 mm × 40 mm.

According to the geometric similarity ratio and the bulk density similarity ratio, the compensation load required for the physical model was calculated to be 48.02 kN. The experimental setup and procedure are as follows:

  1. 1.

    Model loading: The similarity model was air-dried to ensure stable mechanical properties of the simulated materials. Subsequently, the loading device was lowered parallel to the top of the model, and a uniformly distributed load was applied to simulate the overburden stress. During the loading process, safety protection measures were implemented. Protective baffles were installed on the manually operated side of the model to ensure experimental safety, as shown in Fig. 3(a).

  2. 2.

    Data monitoring: The simulated stratigraphic distribution and the detailed layout of the two roadways are illustrated in Fig. 3(b). During model construction, pressure cells were embedded around both simulated roadways to monitor the internal stress evolution of the surrounding rock during roadway excavation and subsequent support installation. After air-drying of the model, reflective markers were attached around the two roadways to measure the deformation of the surrounding rock. Geophones were installed at the top and bottom of the model to detect microseismic activity during excavation and loading.

  3. 3.

    Data collection: The reflective markers around Roadway A were numbered from A1 to A9 from top to bottom. 8 monitoring points are set up in the horizontal direction. The pressure cells were labeled AS1 to AS9 in the order from upper right to lower left. Some reflective markers are located within the contour of the tunnel, and after excavation, the data of these reflective stickers will no longer be recorded. The numbering scheme for Roadway B followed the same convention. The overall layout of the monitoring instruments is shown in Fig. 3(c), and the monitoring control system and data acquisition system are presented in Fig. 3(d).

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Model setting and monitoring scheme.

A miniature drilling machine was used to synchronously excavate the two roadways. The excavation advance was controlled at 5 cm per step to simulate progressive roadway excavation. Installation of simulated rock bolts, anchor cables, and grouting was carried out after the roadways were fully excavated. Extendable bolts with pre-embedded holes were used to simulate boreholes for rock bolts and anchor cables. First, a simulated anchor mesh was installed, followed by sequential installation of simulated rock bolts and anchor cables. A two-component mixed hardening adhesive was used as the anchoring agent. After the adhesive had fully cured and developed anchoring capacity, bearing plates were installed and nuts were tightened to mobilize the load-bearing function of the bolts and cables. The simulated anchor mesh consisted of a steel mesh with an aperture of 3 mm and a wire diameter of 0.5 mm. The simulated rock bolts were M2 bolts with a length of 110 mm, equipped with nuts and bearing plates. The simulated anchor cables consisted of 2 mm steel wire ropes with a length of 310 mm, combined with 2 mm compression-type tensioners with nuts and bearing plates. Layered grouting reinforcement was simulated using a W-999 L alloy grouting pump to reinforce the roof surrounding rock of Roadway A. Grouting holes were drilled from the side of the model, and grouting packers were installed to prevent slurry leakage during the grouting process. During grouting, a constant grouting pressure of 0.2 MPa was maintained to ensure stable slurry diffusion. The overall experimental procedure and the simulated materials used in the physical similarity model test are shown in Fig. 4.

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

Model test steps.

According to the geometric similarity ratio and the bulk density similarity ratio, the compensation load required for the physical model was calculated to be 48.02 kN. After the similarity model was air-dried, the loading device was lowered parallel to the top of the model, and a uniformly distributed load was applied. Subsequently, simulated roadway excavation and support installation were carried out. During the loading process, safety protection measures were implemented, and protective baffles were installed on the manually operated side of the model.

Results analysis

Deformation analysis

Roadway excavation and support were conducted according to the time similarity ratio, and displacement monitoring was performed throughout the experiment. The deformation characteristics of the two roadways are shown in Fig. 5. The monitoring points A6, A7, B6, and B7 are located inside the roadway. They were excavated before loading, so there is no corresponding monitoring data. Under the two support schemes, the maximum roof subsidence values were 2.0 mm and 8.4 mm, respectively, while the maximum floor heave values were 0.8 mm and 0.7 mm, respectively. After grouting reinforcement, the maximum roof subsidence was reduced by 6.4 mm, whereas the maximum floor heave increased slightly by 0.1 mm. After layered grouting reinforcement, the displacement variations among different roof strata became relatively uniform. In contrast, under the original support scheme, deformation of the roof surrounding rock was mainly concentrated in the shallow strata. This observation indicates that the primary control objective for roof stability lies in suppressing deformation and failure of the shallow roof layers. Layered grouting reinforcement effectively enhanced the support capacity of the roof and significantly restrained deformation and damage of the shallow strata. Floor deformation monitoring results show that floor deformation after grouting reinforcement was slightly larger than that under the original support scheme; however, the overall magnitude remained small. After roof reinforcement, stress was transferred toward the floor, resulting in stress concentration and a moderate increase in floor deformation. By comparing roof displacement under different support schemes, it is evident that the new support scheme significantly reduced roof deformation, while floor deformation remained within a controllable range. Overall, the enhanced support scheme exhibited a pronounced reinforcement effect, and roadway stability was substantially improved.

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

Vertical displacement of roadway roof and floor under different support schemes.

The horizontal displacement curves of the two roadway ribs under different support schemes are shown in Fig. 6. The maximum horizontal displacement at the innermost monitoring points of the ribs ranged from 1.5 mm to 2.3 mm, whereas no obvious displacement was observed at the outermost monitoring points. This result indicates that the influence range of rib deformation was limited and largely independent of the support scheme. However, the deformation range of the ribs under the original support scheme was larger than that under the new support scheme. From the overall deformation characteristics, rib displacement under both support schemes remained relatively small and within acceptable limits compared with roof deformation. Under the original support scheme, rib displacement was larger than that observed under the new support scheme. This difference can be attributed to the enhanced roof support, which led to a more uniform stress distribution within the surrounding rock, altered the location of stress concentration, and consequently reduced rib deformation. The comparative results demonstrate that even without modifying rib support conditions, changes in roof support configuration can significantly influence rib deformation behavior.

Fig. 6
Fig. 6The alternative text for this image may have been generated using AI.
Full size image

Horizontal displacement of two sides of different support schemes.

Stress analysis

Stress monitoring was divided into two stages: the excavation stage and the post-support loading stage. The stress monitoring curves of the roadway floor during excavation and after support installation are shown in Fig. 7. During roadway excavation, unloading occurred in the floor, and stress decreased progressively with excavation advance, exhibiting a stepwise reduction pattern. Subsequently, the floor stress transitioned from compressive stress to tensile stress. After the tensile stress reached a stable state, its magnitude was approximately equal to the compressive stress level prior to excavation. After completion of roadway support, floor stress under both support schemes exhibited a decreasing trend with loading. During the early loading stage, stress variation was relatively slow, whereas during the later loading stage, stress changes accelerated markedly. The timing of stress variation under the two schemes showed strong consistency; however, the stress magnitudes differed significantly. After layered grouting reinforcement, the strength and bearing capacity of the roof surrounding rock were substantially enhanced. With continued loading, stress was transferred toward the floor, resulting in increased stress concentration and a continuous rise in tensile stress within the floor.

Fig. 7
Fig. 7The alternative text for this image may have been generated using AI.
Full size image

Vertical stress curve of roadway floor. (a) Excavation process, (b) After support.

The stress monitoring curves of the roadway ribs are shown in Fig. 8. During excavation, rib stress increased in both roadways. Stress rose rapidly during the early excavation stage and gradually stabilized in the later stage. Excavation-induced unloading of the surrounding rock led to stress concentration in the ribs, and rib vertical stress gradually stabilized after stress redistribution was completed. After support installation, significant differences in rib stress evolution were observed between the two support schemes. Under the original support scheme, rib stress increased only slightly and rose slowly with loading, indicating that the original support configuration did not substantially alter the stress environment of the surrounding rock. Under the new support scheme, rib stress exhibited a stepwise increase with loading and showed a much larger growth amplitude. This behavior indicates that the load-bearing structure formed by layered roof grouting played an effective reinforcing role and significantly influenced the overall stress environment of the roadway, promoting gradual stress transfer from the roof toward the ribs.

Fig. 8
Fig. 8The alternative text for this image may have been generated using AI.
Full size image

Vertical stress curve of roadway sides. (a) Excavation process, (b) After support.

Layered grouting reinforcement was implemented in the roadway roof; therefore, particular attention was paid to the stress evolution of the roof surrounding rock during excavation and after support installation, as shown in Figs. 9 and 10. During excavation, roof stress generally exhibited an initial increase followed by a decrease. As excavation progressed and the exposed area of the roadway increased, roof stress rose until reaching a critical state, after which unloading occurred. Stress relief developed progressively from shallow to deep roof strata, with different unloading rates observed at different depths. After completion of roadway support, under the original support scheme, stress in the deepest roof layer increased with loading, while stress in the remaining roof strata decreased. In particular, the immediate roof exhibited an abrupt stress drop during the later loading stage. This phenomenon indicates that, with continued loading, stress release and deformation failure occurred progressively from shallow to deep roof strata, leading to loss of bearing capacity. Consequently, the stress concentration zone migrated upward until reaching stable strata with sufficient bearing capacity.

Under the new support scheme, the stress environment of the roof surrounding rock was markedly altered after layered grouting reinforcement. The grouted layers formed a stable load-bearing structure with high bearing capacity. Stress increase was mainly concentrated within the shallow grouted layers, particularly in the immediate roof, where stress continued to increase and gradually stabilized with loading. In contrast, no significant stress increase or decrease was observed in the deep roof strata, and stress values remained relatively stable throughout the loading process. By integrating stress monitoring results with displacement monitoring data, it can be concluded that layered grouting effectively reconstructs the load-bearing structure of the top coal and roof strata, significantly improves the overall stress environment of the surrounding rock, enhances roof stability, and enables the superimposed beam structure of the roof to fully mobilize its load-bearing capacity.

Fig. 9
Fig. 9The alternative text for this image may have been generated using AI.
Full size image

Vertical stress curve of roadway roof during excavation. (a) Roadway A, (b) Roadway B.

Fig. 10
Fig. 10The alternative text for this image may have been generated using AI.
Full size image

Vertical stress curve of roadway roof after support. (a) Roadway A, (b) Roadway B.

Deformation and failure characteristics

The deformation and failure characteristics of the roof surrounding rock under different support schemes are shown in Fig. 11. After completion of roadway support, roof deformation and failure gradually developed with continued loading. Under the new support scheme, roof damage was mainly concentrated within the immediate roof coal layer and was characterized by sparse microcracks. The cracks were discretely distributed, and no crack penetration or large-scale fragmentation of the roof was observed. In contrast, under the original support scheme, surface spalling and roof collapse occurred, accompanied by extrusion and failure of the anchor mesh. A large number of cracks developed within the immediate roof, coalesced, and propagated upward into the mudstone roof layer, resulting in severe roof fragmentation. Comparative analysis of roof failure characteristics under different support schemes indicates that the original support configuration failed to fully mobilize the reinforcing effect of rock bolts and anchor cables, leading to severe roadway deformation and damage. In contrast, under the new support scheme, layered grouting reinforcement enabled the top coal and roof strata to form a stable superimposed beam load-bearing structure. Both the bearing capacity and the intrinsic strength of the coal–rock mass were significantly enhanced, thereby markedly improving the stability of the roadway roof.

Fig. 11
Fig. 11The alternative text for this image may have been generated using AI.
Full size image

Failure characteristics of simulated roadway surrounding rock.

Microseismic events characteristics

Eight simulated artificial test sources were arranged within the coal seam, and microseismic signals were generated using manual hammer impacts. After propagating through the internal rock mass structure of the model, the microseismic signals were recorded by seven geophones. To reduce experimental error, each artificial source was excited three times. Representative waveform data obtained from the tests are shown in Fig. 12.

Fig. 12
Fig. 12The alternative text for this image may have been generated using AI.
Full size image

Original waveform of microseism. (a) No.1, (b) No.2, (c) No.3, (d) No.4.

The Modified Akaike Information Criterion (M-AIC) algorithm is one of the most widely used seismic phase first-arrival picking methods, characterized by high picking accuracy, fast computational efficiency, and strong noise robustness29. In this study, the M-AIC algorithm was employed to extract the arrival times of microseismic signals. To further improve the accuracy of arrival time picking, a simple optimization was applied to the traditional M-AIC algorithm. Without altering the original mathematical formulation of the M-AIC function, the calculation interval was adjusted such that only the M-AIC function values prior to the maximum sampling point were considered. This modification effectively enhanced the accuracy of arrival time determination. The M-AIC function can be expressed as follows:

$${\text{AIC}}(i)=\sum\limits_{i}^{{\hbox{max} {\text{ }}index}} {i\lg (\operatorname{var} (x\left[ {1,i} \right]))+(L - i - 1)\lg (\operatorname{var} (x\left[ {i+1,L} \right]))}$$
(1)

where x[1, i] denotes the microseismic signal amplitude from the first sampling point to the i-th sampling point. var(x[1, i]) represents the variance of the microseismic signal amplitude over this interval. L is the total number of sampling points of the recorded signal, and max index denotes the sampling point corresponding to the maximum amplitude.

The method is computationally simple and efficient, satisfying the real-time requirements for microseismic arrival time extraction. Using the M-AIC method, the first-arrival times of microseismic signals at each geophone were obtained, allowing the propagation velocity of microseismic waves within the rock strata to be calculated. Typical results of first-arrival time picking are shown in Fig. 13.

Fig. 13
Fig. 13The alternative text for this image may have been generated using AI.
Full size image

Original waveforms and M-AIC values received by some sensors. (a) No.1, (b) No.2, (c) No.3, (d) No.4.

When the distance between the artificial source and the geophone changed, the arrival time recorded by the sensor varied accordingly, and the extracted arrival time increased gradually with increasing distance. Owing to anisotropy among rock strata and lithological differences between layers, equivalent wave velocities for each geophone were determined based on the artificial sources. These velocities were subsequently used to locate microseismic events generated during the loading process. The microseismic event localization results are shown in Fig. 14. In Fig. 14, since this model is a physically similar model, the quantitative representation of microseismic events is insufficient to reflect actual engineering cases. Therefore, a qualitative analysis method is adopted to classify microseismic events into three levels, reflecting the distribution of microseismic events.

Fig. 14
Fig. 14The alternative text for this image may have been generated using AI.
Full size image

Microseismic source positioning results.

Due to the resolution limitation of the monitoring system, low-energy events may not be stably identified. Excavation and support are carried out continuously, and there is a significant overlap in microseismic signals, making it difficult to reliably separate events. For clarity, the detected microseismic events were classified into three energy levels and marked using circles of different colors and sizes. During the roadway excavation stage, microseismic sources were mainly concentrated in the roof region of the roadway. These events were induced by excavation-related disturbance and changes in the stress environment of the roof surrounding rock. Although sporadic microseismic sources were also observed in other regions, no high-energy events occurred, indicating the absence of large-scale microseismic activity. After completion of roadway support, microseismic activity in the roof surrounding rock intensified significantly with continued loading. The number of microseismic events increased markedly, accompanied by a pronounced rise in high-energy events. Comparison of microseismic source locations shows that events were primarily concentrated in the roof region under the original support scheme, indicating intensive crack development and frequent fracture activity in this area. This observation indirectly corroborates the roof failure characteristics discussed in the previous sections.

In addition, several high-energy microseismic events were detected near the upper boundary of the model. This phenomenon is attributed to stress non-uniformity in regions adjacent to the loading device during loading, which likely induced the development of relatively large fractures in these strata. Overall, the results demonstrate that microseismic monitoring technology can not only detect crack initiation and propagation in the surrounding rock and provide guidance for roadway support design, but also effectively evaluate the reinforcement performance of layered grouting in the roadway roof.

Engineering application

As discussed above, severe deformation and damage were observed in the ZF3806 mining roadway of the Shuiliandong Coal Mine after the adoption of the original support scheme. Based on field observations, the specific failure characteristics can be summarized as follows:

  1. 1.

    The top coal was severely fragmented, accompanied by pronounced bulking deformation.

  2. 2.

    Owing to uneven support resistance, non-uniform deformation and failure occurred in the roof surrounding rock, resulting in localized bending and fracture of steel strips.

  3. 3.

    Local roof fall incidents occurred in some sections of the roadway, leading to failure of rock bolts.

  4. 4.

    Significant overall roof subsidence was observed, resulting in heavy maintenance requirements during the later service stage and posing potential safety risks.

Key control techniques for fractured roof

By integrating the results of the physical similarity model tests with field observations, three key principles governing the stability control of fractured coal–rock composite roofs were identified:

(1) Timely active reinforcement to ensure integrity and stability.

Grouting is employed to improve the mechanical properties of the roof coal–rock mass, providing a reliable anchoring foundation for rock bolts and forming a shallow primary load-bearing beam structure.

(2) Differentiated grouting pressure and grout strength.

By adopting different grouting pressures and grout strengths, a superimposed beam load-bearing structure with graded strength and internal stress distribution is formed, thereby reducing stress gradients within the roof surrounding rock.

(3) Rational design of support parameters and structural configuration.

Layered grouting enables the reinforced body and surrounding rock to act synergistically, forming an effective superimposed beam load-bearing structure and a stabilized support-induced stress field, which improves the overall stress state of the roof surrounding rock.

Based on these principles, the specific control strategy was formulated as follows:

(1) After roadway excavation, the fracture state and fracture extent of the top coal and immediate roof were detected to determine the support system design and grouting range, while the grouting effect was monitored in real time.

(2) A sequential grouting strategy of “shallow first, deep later” was adopted. Rapid bolt grouting reinforcement was first applied to the shallow fractured roof to improve the stability of the shallow top coal. After a relatively stable shallow load-bearing structure was formed, cable grouting was subsequently conducted in the deep roof strata.

(3) Layered grouting was implemented separately in the shallow and deep roof zones. Different grout strengths and grouting pressures were applied to the fractured top coal and the coal–rock composite roof layers, respectively. Combined with rock bolts and anchor cables, a superimposed beam load-bearing structure with sufficient strength and bearing capacity was formed, thereby enhancing the integrity and stability of the roof support system.

According to the above control strategy, a reinforcement support scheme was developed for the ZF3806 mining roadway based on the original support configuration. A combined support method centered on “shallow bolt grouting + deep cable grouting” was adopted, integrating rock bolts, anchor cables, steel mesh, steel strips, and grouting into a unified support system. The specific layout is shown in Fig. 15.

Fig. 15
Fig. 15The alternative text for this image may have been generated using AI.
Full size image

Support scheme for ZF3806 mining roadway.

Support performance

Six monitoring points (A, B, C, D, E, and F) were arranged in both the layered grouting reinforcement section and the original support section, with a spacing of 35 m between adjacent points. The grouted section covered a 100 m roadway segment including points D, E, and F, while points A, B, and C were located within a 100 m segment supported by the original scheme, resulting in a total monitored length of 200 m. The layout of the monitoring points is shown in Fig. 16. Roadway deformation was measured using a cross-shaped arrangement, with monitoring conducted at one-day intervals. Rib displacement monitoring points were selected in relatively intact roof–rib zones. Roof–floor convergence was measured at the central position of the roof, while rib convergence was measured at the roadway waistline.

Fig. 16
Fig. 16The alternative text for this image may have been generated using AI.
Full size image

Layout of monitoring points.

The measured rib convergence and roof–floor convergence of the ZF3806 mining roadway are shown in Fig. 17. In the non-grouted section (points A–C), significant rib convergence was observed, with large displacement variations and a maximum displacement of 0.42 m. In contrast, the grouted section (points D–F) exhibited noticeable deformation only during the initial stage, followed by gradual stabilization, with a maximum displacement of approximately 0.20 m. During the early monitoring stage, the working face was far from the monitoring points, and mining-induced effects were insignificant, resulting in minimal rib displacement. As the working face advanced and approached the monitoring points, mining-induced disturbance became pronounced. Under these conditions, continuous rib convergence occurred in the non-grouted section, whereas rib displacement in the grouted section remained relatively small. This observation demonstrates that layered grouting reinforcement effectively enhances rib stability. Only during periodic weighting events did both grouted and non-grouted sections exhibit relatively obvious displacement fluctuations.

The roof–floor convergence trends at points A–C in the non-grouted section were generally similar, characterized by large deformation and damage. The closer the monitoring points were to the working face, the faster the subsidence occurred, and the deformation rate increased significantly with face advance. In contrast, at points D–F in the grouted section, small deformation occurred during the initial stage, followed by gradual stabilization with limited deformation magnitude. Only minor overall deformation was observed when the working face approached. A clear contrast was observed between the two sections: the maximum roof–floor convergence in the non-grouted section exceeded 0.53 m, whereas that in the grouted section was approximately 0.12 m. After grouting reinforcement, roof deformation in the grouted section remained small, and no roof fall or bolt failure occurred, indicating good overall roadway integrity. In the non-grouted section, roof fractures were well developed with high fragmentation, and severe deformation and damage occurred during periodic weighting. In contrast, the grouted section formed an integrated roof structure after grouting, exhibiting good integrity. Minor deformation was observed in the early stage due to incomplete grout strength development, followed by small overall deformation under mining influence.

Fig. 17
Fig. 17The alternative text for this image may have been generated using AI.
Full size image

Displacement of ZF3806 mining roadway.

Fig. 18
Fig. 18The alternative text for this image may have been generated using AI.
Full size image

Support effects after grouting.

As shown in Fig. 18, after adopting the reinforcement method centered on “shallow bolt grouting + deep cable grouting,” the roadway gradually reached a stable state. Roof deformation was significantly reduced, with a maximum roof displacement of approximately 0.12 m. The roof surrounding rock maintained good integrity, with no roof falls or bolt failures observed, and maintenance requirements were substantially reduced. These results indicate that the superimposed beam load-bearing structure formed by layered grouting effectively maintained long-term roof stability.

Discussion

The physical similarity modeling and field observations consistently indicate that the instability of coal–rock composite roofs in deep roadways is not merely a consequence of insufficient support resistance, but fundamentally arises from the disruption of stress transfer continuity caused by severe fragmentation of the top coal and immediate roof. Once the roof strata lose structural integrity, the overburden load can no longer be effectively transmitted to competent overlying strata, resulting in progressive failure dominated by shallow fractured zones. Under the original support scheme, the roof behaves as a discontinuous and hierarchically disordered load-bearing medium. The anchoring system mainly acts on fragmented shallow strata, while the mechanical interaction between the fractured roof and the intact sandstone above remains weak. As a result, stress redistribution during excavation and loading is characterized by localized concentration, crack coalescence, and gradual upward migration of the effective bearing horizon, as evidenced by abrupt stress drops in the immediate roof and intensified microseismic activity.

Layered grouting fundamentally alters this mechanical behavior by reconstructing the internal load-bearing structure of the roof. Shallow bolt grouting rapidly consolidates fractured top coal and immediate roof into a mechanically continuous body, restoring cohesion and stiffness at the shallow roof level. This grouted layer functions as a primary load-bearing beam, capable of redistributing stress laterally and vertically, thereby suppressing localized deformation and crack propagation. Deep cable grouting further anchors the reconstructed shallow bearing layer to the stable sandstone roof, forming a vertically coordinated superimposed beam system. Through this configuration, load transfer continuity across different strata is effectively restored, enabling the overburden load to be shared by multiple roof layers rather than being concentrated within a single fractured zone. Consequently, the roof system transitions from a fragmented, deformation-dominated regime to an integrated, load-controlled regime.

The mechanical insights obtained in this study highlight that effective roof control in deep coal mines should prioritize the reconstruction of internal load-bearing systems rather than solely increasing support density or strength. Layered grouting achieves this objective by simultaneously addressing shallow roof fragmentation and deep roof anchorage, thereby forming a multi-level bearing structure with enhanced load transfer capability. Compared with conventional support strategies, the proposed layered grouting approach provides a systematic framework for coordinating reinforcement parameters with roof stratification and damage characteristics. This framework is particularly suitable for roadways with coal–rock composite roofs exhibiting strong mechanical heterogeneity and pronounced fracture development.

Conclusions

(1) Instability of coal–rock composite roof roadways in deep coal mines fundamentally originates from severe fragmentation of the top coal and immediate roof, which significantly weakens structural integrity and disrupts internal stress transfer paths. As a result, fractured shallow strata are unable to effectively transmit loads to stable overlying strata. Physical similarity model tests demonstrate that, under conventional support conditions, the roof fails to form a continuous and effective load-bearing structure, with fractures progressively coalescing from shallow to deep strata, ultimately leading to progressive roof instability.

(2) A layered grouting reinforcement technique for coal–rock composite roofs, centered on “shallow bolt grouting + deep cable grouting,” was proposed. Shallow bolt grouting rapidly consolidates fractured top coal and the immediate roof into a continuous load-bearing body, forming a shallow primary bearing beam. Deep cable grouting effectively anchors the shallow reinforced layer to the stable roof strata, constructing a vertically coordinated superimposed beam load-bearing structure that enables coordinated deformation and joint load bearing among different strata.

(3) Physical similarity modeling results indicate that layered grouting fundamentally alters stress redistribution paths within the surrounding rock. Roof stress concentration shifts from deep strata to shallow grouted layers, while crack propagation and coalescence are effectively suppressed. Compared with the original support scheme, the maximum roof subsidence is reduced by approximately 76%, and both the intensity and frequency of microseismic activity are significantly attenuated, indicating a substantial improvement in overall surrounding rock stability.

(4) Industrial-scale application in the ZF3806 mining roadway of the Shuiliandong Coal Mine confirms the engineering applicability of the proposed technique. Compared with the original support scheme, the maximum rib deformation is reduced from 0.42 m to 0.20 m, and roof–floor convergence is reduced from 0.53 m to 0.12 m. The roadway roof maintains good integrity without roof falls or bolt failures, ensuring effective long-term stability.