Experimental Study on the Influence of Roof/Floor Conditions on Pillar Burst

Abstract

Coal measures in the central and western regions of Mainland China are mainly composed of weakly cemented strata. However, some coal mines in these regions are characterized by hard roof and floor formations. To investigate the effects of roof and floor conditions on pillar burst and failure behavior, uniaxial compression tests were conducted on roof–coal–floor composite specimens with different lithological configurations. The failure characteristics, acoustic emission (AE) response, energy evolution, and fragmentation size distribution were comprehensively analyzed. The results show that the strength of the roof and floor significantly affects both the overall load-bearing capacity of the composite specimens and the local failure severity of the coal section. For specimens containing mudstone roof and/or floor, the burst energy index ranged from 2.30 to 2.93, indicating weak burst propensity and progressive failure controlled by the weak mudstone section. In contrast, the sandstone–coal–sandstone (SCS) specimens exhibited the highest UCS of 34.46 MPa and a typical Class II stress–strain response, indicating high burst propensity. The postpeak AE energy release rate and ring count rate of the SCS specimens reached 7.2 × 10^–3 and 4.9 × 10^–3, respectively, and the fragmentation fractal dimension reached 2.24. These quantitative results indicate that hard roof and floor conditions enhance the overall energy storage capacity of the composite system and intensify the sudden burst failure of the coal section. The findings provide an experimental basis for understanding pillar burst mechanisms under different roof and floor geological conditions.

1. Introduction

With the depletion of coal resources in the eastern region of China, the development of coal resources has gradually shifted to the central and western regions in recent years. The coal measures in the central and western regions are mainly weakly cemented strata, while some coal mines have hard roof and floor formations. For example, there are obvious weakly cemented strata such as mudstone and sandy mudstone in most mines of the Northwestern Ordos Basin. Some mines, such as Menkeqing, Huajiahe, and Nalinhe No. 2, often exhibit simple geological structures with hard roof and floor strata. Extensive research has shown that the geological conditions and rock properties have a significant influence on mine pressure during the mining process. As an essential support structure in underground mining spaces, coal pillars are prone to dynamic failure due to the supercritical stress caused by mining disturbances, structural factors, pillar layout, and roof/floor properties. Pillar bursts refer to the dynamic instability of coal pillars formed by roadway cutting, goaf isolation, or fault clamping during coal mining. Pillar bursts are a typical hazard in coal mines in China, especially in those at great depth or with complex geological conditions.

In underground mining conditions, the roof and floor stiffness can affect the pillar stability as the load is applied to the coal seam through the roof and floor. Considering coal and its surrounding rock as a composite model is a commonly used method for studying pillar stability. Various studies have been conducted on coal-rock composite models. Petukhov et al. conducted early studies to examine the stability of the coal-rock composite system based on a two-dimensional model composed of the roof, floor, and coal. Zhang et al. studied the mechanical behavior of coal-rock composite specimens with different configurations and found that the rock-coal-rock composite specimen demonstrated the greatest strength. Liu developed a damage constitutive model for coal in a composite structure by serially connecting a damage body with one or two Newtonian bodies, which revealed the influence of rock on the mechanical behavior of coal pillars in the composite structure. Poulsen studied the effect of water saturation on the strength of coal pillar in the roof–coal–floor structure. Other studies examined the burst propensity of composite specimens with different configurations and found that the burst propensity of composite specimens was greater than that of coal or rock specimens. Zuo and Song concluded that coal is the main medium for elastic energy accumulation and release in the composite specimens based on an analysis of the evolution of elastic energy density in coal and rock within composite specimens. Chen et al. proposed a method to determine the burst propensity of coal-rock composite specimens based on the residual energy release rate. Chai conducted uniaxial cyclic tests of coal-rock composite specimens with different roof lithologies to study the mechanical response and damage evolution process of the coal-rock composite structure under cyclic load disturbance. Zhao derived and validated the general compression-shear failure criterion of an equivalent model considering the interface cohesive strength between coal and soft rock based on the equivalent principle of strain energy. Zuo and Song conducted confining pressure unloading experiments at different unloading rates of coal-rock composite specimens to explore the energy evolution patterns under confining pressure unloading conditions. Zhao et al. studied the effects of different coal section thicknesses on the mechanical and energy characteristics of the composite specimens, which found that the compressive strength decreases, and the energy storage limit and energy dissipation conversion decrease with increasing coal thickness ratio. Ma analyzed the stress-deformation characteristics, acoustic emission characteristics, and failure process of the coal-rock composite specimens under different coal fracture dip angles based on uniaxial compression tests of fissured coal-rock composite specimens.

The above studies indicate that research on coal-rock composite specimens has mainly focused on three aspects. First, considerable attention has been paid to the mechanical behavior and stability of coal-rock composite systems, including the effects of the combination mode, coal thickness, interface properties, and roof lithology on strength and deformation characteristics. Second, energy accumulation, dissipation, and energy-based burst propensity indices have been investigated to explain the instability process of the composite specimens. Third, the damage evolution of coal-rock composite structures under cyclic loading, unloading paths, mining-induced disturbance, and initial defects has also been explored. These studies have provided important insights into the mechanical response and instability mechanism of coal-rock composite specimens.

However, several issues remain to be insufficiently addressed. Most previous studies focused on the overall mechanical properties and failure behavior of the composite specimen, whereas the local burst failure characteristics of the coal section within the roof–coal–floor structure have received relatively limited attention. In addition, many existing studies were mainly based on composite specimens consisting of hard rock and coal, while the comparative influence of weak mudstone and hard sandstone roof and floor conditions has not been fully clarified. Therefore, the role of the roof and floor strength contrast in controlling the transition from weak-section-dominated progressive failure to coal-section-dominated burst failure still requires further investigation.

To address this issue, this study prepared four types of roof–coal–floor composite specimens using mudstone, coal, and sandstone to represent different roof and floor conditions. Uniaxial compression tests were conducted, and the mechanical response, acoustic emission characteristics, energy evolution, and fragmentation size distributions were comprehensively analyzed. The objective of this study is to clarify how the roof/floor strength contrast affects the burst failure severity of the coal section and to provide an integrated experimental basis for identifying pillar burst behavior under different roof and floor geological conditions. In particular, the AE release-rate indicators, including the postpeak AE energy release rate and ring count rate, were used to characterize the dynamic intensity of coal-section failure. By combining these AE indicators with stress–strain response, energy evolution, and fragmentation fractal dimension, this study establishes a multiparameter identification approach for evaluating coal pillar burst severity under different roof and floor conditions.

2. Experimental Section

2.1. Specimen Preparation

The specimens were collected from a coal mine in the Ordos mining area, China, at a sampling depth of approximately 700 m. The immediate roof and floor of the coal seam at the sampling site are both sandstone layers with thicknesses greater than 5 m. Therefore, the sandstone–coal–sandstone configuration represents the typical hard roof and floor conditions at the sampling site, while the mudstone-bearing configurations were designed as comparative cases to investigate the influence of the roof/floor strength contrast on coal-section burst failure behavior. The selected component materials were mudstone (M), coal (C), and sandstone (S), representing weak roof/floor rock, coal seams, and hard roof/floor rock, respectively. In the specimen notation, the first, second, and third letters represent the roof, coal section, and floor, respectively. For example, MCM denotes a mudstone–coal–mudstone composite specimen, whereas SCS denotes a sandstone–coal–sandstone composite specimen. Their basic physical and mechanical parameters are listed in Table . As shown in Table , the three materials exhibit clear differences in strength and stiffness. Sandstone has the highest uniaxial compressive strength (UCS) and elastic modulus, whereas mudstone has the lowest strength and stiffness. Coal shows intermediate mechanical properties and is the main burst-prone component in the roof–coal–floor composite specimens. Therefore, the selected materials provide a suitable basis for investigating the influence of roof and floor lithology on the coal section failure behavior.

1. Basic Physical and Mechanical Parameters of Mudstone, Coal, and Sandstone.

Material	UCS/MPa	Elastic modulus, E/GPa	Poisson’s ratio, ν	Tensile strength/MPa	Density/kg·m–3
Mudstone	9.5	2.38	0.24	0.65	2473
Coal	24.7	3.05	0.28	1.34	1463
Sandstone	40.2	12.29	0.21	3.88	2580

Four groups of roof–coal–floor composite specimens, namely, MCM, MCS, SCM, and SCS, were prepared to represent different roof and floor stiffness contrasts.

Each composite specimen had a size of 50 mm × 50 mm × 100 mm. The middle coal section was 40 mm thick, and the upper and lower rock sections were each 30 mm thick. The schematic configuration and physical appearance of the roof–coal–floor composite specimens are shown in Figure . All rock and coal sections were cut from intact blocks and processed to the designed dimensions. The end-face nonparallelism was controlled within 0.05 mm, and the axial deviation was less than 0.25°. The rock and coal sections were bonded using high-strength adhesive to form integrated composite specimens before testing.

1.

Rock-coal composite specimen.

2.2. Equipment

The tests were performed on a WAM-200 servo-controlled loading system with the application of a MISTRAS PCI-2 AE (acoustic emission) monitoring system, a DH5960 linear variable differential transformer (LVDT) set, and a digital camera, as shown in Figure . The four components of the test system were synchronized to ensure the ongoing collection of stress, strain, and AE data at the same time for further data analysis. The specimens were loaded at a displacement-controlled rate of 0.2 mm/min according to previous coal-rock composite specimen tests, which was adopted to maintain a quasi-static loading condition and ensure the comparability of different specimen groups. The AE parameters are listed in Table . Vaseline was applied between the sensor and the specimen to reduce the acoustic impedance difference at the contact surface and minimize energy loss due to reflection at the interface, which ensures that the acoustic emission signals can be effectively received by the sensor. The lead break method was used to test the sensitivity of the AE system to verify that the amplitude signal of each sensor is above 90 dB. The sampling rate of the LVDT was 100 Hz. The specimen fracture process was continuously recorded with a digital camera.

2.

Test equipment for the composite specimen.

2. AE Parameters.

Items	Parameters	Items	Parameters
Main amplifier	40 dB	AE sensor type	Nano 30
Threshold value	45 dB	Resonant frequency of the AE probe type	100–600 kHz
Floating threshold	6 dB	Sampling frequency	1 MHz

2.3. Procedure

Before testing, the dimensions, end-face parallelism, and bonding quality of each roof–coal–floor composite specimen were checked. The specimen was placed at the center of the loading platen, and its axial alignment was adjusted to reduce eccentric loading. The AE sensors were attached to the specimen surface using Vaseline as the coupling agent. The lead break test was conducted before loading to verify the sensitivity and consistency of the AE acquisition channels. The loading system, LVDT system, AE monitoring system, and digital camera were then synchronized, and the load, displacement, deformation, and AE channels were zeroed or calibrated.

The specimens were loaded under a displacement control at a rate of 0.2 mm/min until failure. During loading, the axial load, axial displacement, AE signals, and surface crack evolution were recorded continuously. The test was terminated when the specimen lost its bearing capacity, when the postpeak stress decreased to a stable residual level, or when severe fragment ejection occurred. After unloading, the failed specimen was removed carefully, and the coal fragments were collected separately from roof and floor rock fragments as much as possible. The collected coal fragments were weighed and classified according to the equivalent side length. Large fragments were measured using a vernier caliper, fragments between 1 and 10 mm were classified by sieving, and fine fragments smaller than 1 mm were analyzed using the heavy-medium flotation method. The resulting mass data were used for the fragmentation size distribution and fractal dimension analysis.

3. Results

3.1. Mechanical Properties and Burst Energy Index

The stress–strain curves for different composite specimens were plotted in Figure . The MCM specimens showed the lowest UCS of 9.60 MPa, whereas the SCS specimens showed the highest UCS of 34.46 MPa. The UCS values of SCM and MCS were 14.97 and 13.08 MPa, respectively. The results indicated that the UCS of the composite specimen increased with the proportion of sandstone sections. The composite specimen demonstrated a good linear stress–strain relationship before the peak stress. The composite specimens with mudstone have a relatively gentle increasing trend before the peak and a gentle decrease after the peak. The composite specimens without mudstone, i.e., both roof and floor sections are sandstone, demonstrated a sharp increase before peak and a sudden drop after peak, a behavior that is characteristic of brittle failure.

3.

Stress–strain curves for composite specimens.

The UCS of composite specimens with mudstone are less than the UCS of standard coal specimens (24.7 MPa) but is closer to the UCS of mudstone specimens (9.5 MPa). The strength of the composite specimen depends on the weak section. In addition, the changing trend of Young’s modulus E and radial strain ε _h is similar to UCS. When both the roof and floor sections are sandstone, the above parameters demonstrated an obvious increasing trend. The absolute value of UCS, E, and ε _h increased 2–3 times. The standard deviations of the mechanical parameters are summarized in Table . The results indicate that the standard deviations of UCS, elastic modulus, and transverse strain for SCS specimens are approximately 3–4 times higher than those of specimens containing mudstone.

3. Standard Deviations of the Mechanical Parameters of the Composite Specimens.

Standard deviation	MCM, MCS, and SCM	SCS
UCS	2.04	9.26
E	0.23	1.05
ε h	0.09	0.22

Burst energy index is the indicator for the coal burst propensity to evaluate the potential coal burst risks. , The definition of burst energy index (K _E ) is the ratio between the accumulated deformation energy before the peak stress below the stress–strain curve and the dissipated deformation energy after peak stress. A higher value of K _E indicates that more elastic deformation energy was released in the form of brittle failure. The burst energy index (K _E ) was calculated from the complete stress–strain curve as follows:

$$K_E=U_{pre}/U_{post}$$ 1

where U _pre is the accumulated deformation energy before the peak stress and U _post is the dissipated deformation energy after the peak stress. In this study, U _pre was obtained by integrating the area under the stress–strain curve from the initial effective loading point to the peak strain, while U _post was calculated by integrating the postpeak stress–strain curve from the peak strain to the residual bearing stage or to the end point of the recorded postpeak curve. The initial unstable contact section of the stress–strain curve was excluded from the integration. For specimens showing a typical Class II stress–strain curve, the postpeak stress dropped sharply and the postpeak dissipated energy was very small. In such cases, the calculated K _E became very large, and the Class II curve feature was also used as supplementary evidence for identifying high burst propensity. According to the GB/T 25217.2-2010 test standard, K _E ≥ 5 is high burst propensity, 1.5 ≤ K _E < 5 is low burst propensity and K _E ≥ 1.5 is none burst propensity. The K _E for each composite specimen was calculated to compare the burst propensity. The calculated burst energy indices and related mechanical parameters of different composite specimens are presented in Figure . The average K _E for MCM, MCS and SCM are 2.93, 2.3 and 2.76, which indicated a low burst propensity. In addition, the SCS specimens exhibited a typical Class II stress–strain curve, characterized by an abrupt postpeak stress drop and an extremely small postpeak dissipated energy. This curve feature further confirms the high burst propensity of the SCS specimens.

4.

Burst energy index for composite specimens. (Note: “II” means typical high burst propensity type “II” stress–strain curve.)

It should be noted that the increase in the overall strength of the composite specimen does not simply mean that the coal section becomes more stable. For the SCS specimens, the hard sandstone roof and floor improve the load-bearing capacity of the roof–coal–floor system and allow more elastic energy to accumulate before failure. However, because the coal section is weaker than the sandstone roof and floor, the accumulated elastic energy is preferentially released through the sudden failure of the coal section, resulting in stronger local burst behavior.

3.2. Failure Characteristics

The instability of coal and rock subjected to loading conditions is the dynamic process of crack initiation, propagation, and sudden penetration. , The failure characteristics contain rich information, such as displacement and crack expansion. The failure characteristics of composite specimens are shown in Figure . Surface crack observations showed that visible cracks first developed in the mudstone section and then propagated toward the contact area between the mudstone and the coal sections. As the load increases, cracks in the floor section propagate into the coal section. The cracks in the floor section are more intensive than those in the coal section for the failed specimens. When the residual strength of the specimen is at 3.91 MPa, the floor mudstone section was fractured mostly by tensile-compressive failure. The coal section is basically intact with more tensile macroscopic cracks without burst behavior. The failure characteristics of MCS, SCM, and MCM specimens are similar. Visible surface cracks were first observed in the mudstone section of these specimens and then propagated into the contact area between the mudstone and coal sections. MCS and SCM specimens demonstrated a coal particle ejection phenomenon in some areas. More tensile cracks were observed from the mudstone and coal sections of MCS and SCM specimens with a relatively more severe failure behavior compared with MCM specimens. There was no obvious failure for SCS specimens at the early loading stage. The localized failures occurred at the corner of the specimen with coal and rock particles ejection when the specimen was loaded to 26.6 and 32.4 MPa (corresponding to points b and c in Figure d). Violent failure occurred with coal particle ejection and loud sound when the stress reached the peak, which demonstrated an obvious burst failure characteristic. The coal section of the specimen lost its bearing capacity with severe failure behavior. The failure of the SCS specimen was instantaneous to observe the crack initiation and propagation process within the coal section.

5.

Failure evolution of composite specimens.

Under hard roof and floor conditions, the strength of the coal is the dominant factor for the strength of the rock-coal composite system, which is consistent with previous studies. ,,, Coal will fail when the internal damage reaches the limit due to its low strength. As shown in Figure d, elastic rebound of floor and roof was observed from SCS specimen when the failure occurred, which indicated that the energy from the loading machine was transformed into elastic energy within the roof and floor sandstone through elastic deformation. As the strength of sandstone is higher than that of coal, the sandstone is still in the elastic state when the coal fails. The elastic energy stored within the sandstone section will be suddenly released, which potentially contributes to the dynamic failure of the coal section.

The cracks first appeared within the mudstone section for the composite specimens with the mudstone section. The cracks will expand and propagate with the increase of the applied load. The cracks will propagate to the coal section after peak load and propagate along the loading direction. The failure of these composite specimens is dominated by the mudstone section.

3.3. Energy Dissipation

During uniaxial compression, the external work done by the loading system is mainly converted into elastic strain energy stored in the specimen and dissipated energy consumed by pore closure, crack initiation, crack propagation, frictional sliding, and irreversible damage. , Before the peak stress, the energy balance of the specimen can be expressed as

$$U=U_e+U_d$$ 2

where (U) is the total input energy density, (U _e ) is the elastic strain energy density, and (U _d ) is the dissipated energy density. The total input energy density (U) was obtained by integrating the stress–strain curve before the peak stress. The elastic strain energy density (U _e ) was estimated according to elastic strain energy theory, and the dissipated energy density (U _d ) was calculated as the difference between (U) and (U _e ). The schematic diagram of energy accumulation and dissipation under uniaxial compression is shown in Figure .

6.

Schematic diagram of energy accumulation and dissipation of the specimen subjected to uniaxial compressive loading.

For a general three-dimensional stress state, the total input energy density and elastic strain energy density can be expressed as

$$U={\int }_0^{{\epsilon }_1}{\sigma }_1\mathrm{d}{\epsilon }_1+{\int }_0^{{\epsilon }_2}{\sigma }_2\mathrm{d}{\epsilon }_2+{\int }_0^{{\epsilon }_3}{\sigma }_3\mathrm{d}{\epsilon }_3$$ 3

$$U_e=\frac{1}{2E}[{\sigma }_1^2+{\sigma }_2^2+{\sigma }_3^2-2\nu ({\sigma }_1{\sigma }_2+{\sigma }_2{\sigma }_3+{\sigma }_1{\sigma }_3)]$$ 4

where (σ _i ) (i = 1, 2, and 3) is the principal stress, (ε _i ) (i = 1, 2, and 3) is the principal strain, (ν) is Poisson’s ratio, and E is the elastic modulus. In the present uniaxial compression tests, the lateral stresses were approximately zero, namely (σ₂ = σ₃ = 0). Therefore, the total input energy density and elastic strain energy density were calculated as

$$U={\int }_0^{{\epsilon }_1}{\sigma }_1\mathrm{d}{\epsilon }_1$$ 5

$$U_e=\frac{{\sigma }_1^2}{2E}$$ 6

$$U_d=U-U_e$$ 7

where (σ₁) and (ε₁) are the axial stress and axial strain, respectively. In this study, (E) and (ν) were treated as equivalent mechanical parameters of the composite specimen and were determined from the prepeak quasi-linear elastic stage.

The values of (U), (U _e ), and (U _d ) for different composite specimens are shown in Figure . The total input energy density and elastic strain energy density of the SCS specimens were approximately twice those of the composite specimens containing mudstone. The average dissipated energy density of the specimens containing mudstone was about 0.04 mJ/mm³, whereas that of the SCS specimens was 0.023 mJ/mm³. This indicates that more input energy was consumed by plastic deformation, crack development, and weak section damage in the specimens containing mudstone. In contrast, the SCS specimens stored more energy in the form of elastic strain energy before failure, which provided an energy basis for the sudden-burst failure of the coal section.

7.

Energy accumulation and dissipation of different composite specimens.

It should be noted that the energy parameters calculated in this section are derived from the global stress–strain response of the roof–coal–floor composite specimen. Therefore, the calculated total input energy density, elastic strain energy density, and dissipated energy density represent the system-level energy evolution of the composite specimen rather than the independent energy evolutions of the roof, coal section, and floor. Because the axial and lateral deformations of each individual material section were not measured separately in the present tests, the energy contributions of each section cannot be directly separated from the experimental data. Accordingly, the energy results are used mainly for comparative analysis among different composite configurations under the same specimen size, loading path, and boundary conditions.

In addition, (E) and (ν) are not strictly constant during the entire loading process. With the initiation, propagation, and coalescence of microcracks, the effective elastic modulus and Poisson’s ratio of the composite specimen may change progressively, especially near the peak stress and in the postpeak stage. If the degradation of (E) and (ν) is not considered, the calculated elastic strain energy may deviate from the actual recoverable elastic energy, and the corresponding dissipated energy may also be affected. Therefore, the calculated (U _e ) and (U _d ) should be interpreted as equivalent energy parameters based on the global stress–strain response, rather than exact time-dependent energy components.

3.4. Acoustic Emission

The damage and crack of rock/coal bodies are generally associated with acoustic signals, which is also an indicator of energy dissipation and internal fracture. , Figure is the AE energy plot for different composite specimens. The diagram shows an obvious “Calm–Active–Decay” staged variation (Stages I, II and III, as indicated in Figure ) in the accumulated AE energy. To reduce the subjectivity of stage division, the AE stages were identified by combining the normalized stress level, the variation rate of cumulative AE energy, and the occurrence of high-energy AE events. The calm stage was defined as the stage in which the cumulative AE energy increased slowly and no continuous high-energy AE events were observed. The active stage was identified when the cumulative AE energy showed a continuous increase and AE events became more frequent. The unstable failure stage was characterized by a sharp increase in cumulative AE energy near the peak stress, accompanied by localized failure or fragment ejection. The corresponding stress levels for different specimen groups are discussed below.

8.

AE energy of composite specimens.

For composite specimens containing mudstone, the natural pores and weaknesses within the specimens were compressed and closed before loading reached 70% of peak stress, generating only a small number of low-energy AE events with a minimal increase in cumulative AE energy.

As the load increases, the AE energy starts to rise noticeably, entering an active stage of acoustic emission until macroscopic failure occurs. At this stage, AE events become more frequent and cumulative AE energy sharply increases. When the stress is close to the peak, stress adjustments appear in the stress–strain curve with a surge in AE energy, indicating an abrupt change in AE cumulative energy. This suggests that cracks interconnect to form a rupture surface, causing significant internal damage within the composite specimen. At the postpeak stage, the number of high-energy events decreases with the residual stress for MCM specimens. The slope of the AE cumulative energy curve gradually decreased. However, the MCS and SCM composite specimens demonstrate higher values at the same stage compared to the MCM, due to more severe damage of the coal section in the MCS and SCM specimens. Cracks in the coal sections of MCS and SCM specimens continue to develop, releasing a considerable amount of energy during the damage process.

Before 85% of the peak stress, no obvious change of the cumulative AE energy was observed for SCS specimens with acoustic emission remaining in a calm phase. At approximately 95% of the peak stress, a sudden localized failure occurs in the coal section, causing a sharp increase in cumulative AE energy. Then, the composite specimens enter a stress rebalancing stage with a slowdown of AE cumulative energy increase. There is a sudden increase in AE energy at the failure point, indicating that the SCS specimens stored substantial elastic strain energy in their early stages.

The duration of the AE active period for the mudstone composite specimen is 1200 to 2000 s, with the main increase in AE energy concentrated around the peak and postpeak stages. In contrast, the duration of the active period for the SCS specimens is approximately 800 s, with high-energy events primarily concentrated at prepeak and peak stages. The failure of the SCS specimen is instantaneous. Moreover, as shown in Figures and a–d, each AE energy spike corresponds to cracking in the mudstone composite specimen. The surge in AE signals corresponds to damage in the coal and floor section at prepeak and peak stages of SCS specimens. The SCS specimens exhibit relatively low AE signals with no distinct crack development process before the peak stress, suggesting that cracks within the coal section are not gradually formed by sustained mechanical energy input, instead crack penetration was formed instantaneously at the failure stage.

11.

Accumulated mass-equivalent side length curve.

Coal/rock burst is the dynamic failure of coal/rock body accompanied by the sudden release of stored elastic strain energy. The energy release rate and ring count rate per unit time during the postpeak stage were normalized to compare the severity of burst failure of composite specimens. That is, higher energy release rate and ring count rate indicate a more severe failure. As shown in Figure , both the energy release rate and ring count rate are below 1 × 10^–3 after peak stress for composite specimens containing mudstone. The energy release rate and ring count rate are 7.2 × 10^–3 and 4.9 × 10^–3 respectively, significantly higher than mudstone composite specimens. The energy accumulation level and releasing rate are lower when the strength of the floor and roof is low. The energy accumulation level and releasing rate are higher when the roof and floor are strong, which also indicates a high propensity of coal burst or pillar burst.

9.

Postpeak AE energy release rate and ring count rate of composite specimens.

3.5. Fragmentation Size Distribution

Fractal theory can quantify the fracture fragmentation size distribution (FSD), which is another indicator of burst failure severity. A higher fractal dimension means more fractured coal/rock fragmentation, which also means an intensively dynamic failure. The fragmentation size is classified based on its equivalent side length. The coal section is more fractured for the SCS specimen, while that of other specimens is more intact.

Previous studies have indicated that the fractal characteristics of fragmentation size distribution are more evident within a suitable particle-size range, whereas a small number of large fragments may significantly affect the fitting result. Therefore, different upper-threshold sizes were examined in this study. The threshold size L _eq‑max = 30 mm was finally selected because the log–log fitting coefficient satisfied R > 0.9 within this range, indicating good linearity between accumulated mass and equivalent side length. This threshold selection also follows the particle size treatment method adopted in relevant fragmentation studies. Therefore, L _eq‑max = 30 mm was used for calculating the fractal dimension in this study. It should also be acknowledged that the thickness of the coal section in the composite specimen was 40 mm; therefore, fragments with an equivalent side length of close to 30 mm may be affected by the specimen size and boundary conditions. These large fragments may not fully reflect burst-induced fragmentation alone but may also be related to the geometric constraint of the coal section. For this reason, the fractal dimension obtained in this study was mainly used as a relative indicator to compare the fragmentation degree among different roof–coal–floor composite specimens under the same specimen size and testing conditions, rather than as an absolute material parameter independent of the specimen scale. The heavy medium flotation method was used to analyze the coal fragments less than 1 mm. The vernier calliper was used to measure the coal fragments larger than 10 mm. The coal fragments between 1 and 10 mm were sieved as shown in Figure f.

10.

Fragmentation of composite specimens.

The fractal dimension D for accumulated mass-equivalent side length is calculated by

$$\alpha =\frac{\lg (M_{L_{eq}}/M)}{\lg L_{eq}}$$ 8

$$D=3-\alpha$$ 9

where α is the slope of the curve in log–log coordinate when $M_{L_{eq}}/M-\lg L_{eq}<L_{eq}$ , $M_{L_{eq}}$ is the accumulated mass of fragments less than L _eq , M is the total weight of fragments, and D is the fragmentation fractal dimension.

The better linearity between $\lg (M_{L_{eq}}/M)$ and lg L _eq indicates an obvious fractal distribution. When the D = 0–2, larger fragments account for a higher proportion. When D = 2–3, smaller fragments account for a higher proportion. The accumulated mass for each equivalent side length is listed in Table .

4. Accumulated Mass for Each Equivalent Side Length.

Equivalent side length	Accumulated mass of fragments smaller than L eq /g
L eq /mm	MCM	MCS	SCM	SCS
0.25	0.31	0.23	0.23	1.76
0.5	0.57	0.53	0.49	4.46
1	0.96	0.8	0.78	8.23
2.5	1.59	1.18	1.06	17.06
5	2.7	1.96	2.42	27.7
10	3.61	2.46	3.12	39.34
20	10.24	10.07	9.8	63.39
30	31.64	30.4	40.92	69.92
50	154.76	151.34	155.2	142.84

Based on Table , the accumulated mass-equivalent side length curve is plotted in Figure . The D for SCS specimens is the largest, at 2.24, which indicates more fractured fragmentation. The D alues for other specimens are below 2, which indicates that larger fragments constitute a greater proportion. For the composite specimens containing mudstone, internal cracks are less developed in the coal section, resulting in larger fragments. For the SCS specimens, a dense crack network rapidly initiates and propagates, resulting in a higher degree of fragmentation.

4. Discussion

This study investigated the mechanical behavior and burst failure characteristics of roof–coal–floor composite specimens from the perspectives of failure morphology, AE response, energy evolution, and fragmentation size distribution. The results indicate that roof and floor lithology have a significant influence on both the overall load-bearing capacity of the composite specimen and the local burst failure severity of the coal section.

For specimens containing mudstone in the roof or floor, visible surface cracks were first observed in the weak mudstone section and then propagated toward the coal–mudstone interface. On the basis of the surface crack evolution and postfailure morphology, the failure of these specimens can be interpreted as being mainly controlled by the weak mudstone section. Because mudstone has a lower strength and stiffness, a larger proportion of the input mechanical energy is consumed by progressive cracking, plastic deformation, and weak-section expansion. Therefore, the coal section in these specimens shows a relatively limited burst failure behavior. In contrast, for the SCS specimens with a sandstone roof and floor, no obvious surface cracking was observed in the early loading stage. The coal section became the primary weak-bearing component within the composite system, and abrupt coal fragment ejection occurred near the peak stress, indicating a more intense local burst failure.

The mechanical, AE, energy, and fragmentation results indicate a consistent mechanism linking roof and floor lithology to coal section burst failure. Under hard roof and floor conditions, the sandstone sections have higher strength and stiffness and can remain largely elastic before failure of the coal section. This increases the overall load-bearing capacity and elastic energy storage capacity of the roof–coal–floor composite system. Once the damage in the relatively weak coal section reaches its critical state, the accumulated elastic energy is released abruptly, resulting in a sudden postpeak stress drop, intense AE activity, coal fragment ejection, and a higher fragmentation fractal dimension. In contrast, when mudstone is present in the roof or floor, deformation and cracking first develop in the weak mudstone section and a larger proportion of the input energy is dissipated through progressive damage and plastic deformation. Consequently, less elastic energy is available for a sudden release in the coal section, and the burst failure severity is reduced. Therefore, hard roof and floor conditions do not simply imply a more stable coal section; rather, they enhance the overall energy storage capacity of the composite system and may intensify the local burst failure of the coal section.

It should be noted that the burst energy index, K _E , used in this study was originally proposed for standard single coal specimens according to GB/T 25217.2-2010. When this index is applied to roof–coal–floor composite specimens, its physical meaning differs from that for a single coal specimen because the roof and floor sections also participate in elastic energy storage and release. This effect is particularly significant for the SCS specimens, in which the hard sandstone roof and floor can remain largely elastic before coal section failure and may store a considerable amount of elastic strain energy. Therefore, K _E in this study should not be interpreted as an absolute burst propensity classification index for coal material itself. Instead, it is used as a relative comparative indicator to evaluate the differences in energy accumulation and postpeak energy release among different composite specimens under the same specimen size, loading path, and boundary conditions. To reduce the uncertainty caused by this limitation, the burst propensity of the composite specimens was interpreted together with the stress–strain curve type, failure characteristics, AE release rate indicators, and fragmentation fractal dimension.

The present energy analysis also has certain limitations. The roof, coal section, and floor have different elastic moduli, Poisson’s ratios, and damage evolution characteristics. In this study, the energy calculation was based on the global stress–strain response and equivalent mechanical parameters of the whole composite specimen. Thus, the calculated energy parameters represent the system-level energy evolution of the composite specimen rather than the independent energy evolution of the roof, coal section, and floor. During loading, especially near and after the peak stress, the effective elastic modulus and Poisson’s ratio of the composite system may change continuously due to crack initiation, propagation, and coalescence. If the degradation of E and ν is not considered, the calculated elastic strain energy may deviate from the actual recoverable elastic energy, and the corresponding dissipated energy may also be affected. Therefore, the energy results in this study are mainly used to compare the relative energy storage and dissipation characteristics of different roof–coal–floor configurations, rather than to provide an exact time-dependent separation of elastic and dissipated energy or exact segmental energy values. Further quantification of the energy contributions of the roof, coal section, and floor requires unloading-reloading tests, independent segmental strain measurements, digital image correlation, CT-based damage reconstruction, or numerical simulation.

It should also be noted that AE signals may be affected by the lithological differences among the mudstone, coal, and sandstone. These materials have different stiffnesses, acoustic impedances, wave velocities, and attenuation characteristics. Therefore, the absolute AE energy recorded by the sensors may not only reflect the fracture intensity but may also be influenced by wave propagation and attenuation through different material sections. In this study, all specimens were tested under the same sensor layout, acquisition parameters, and load conditions. Thus, the AE energy release rate and ring count rate were mainly used as relative indicators for comparing the failure intensity among different composite configurations. The interpretation of AE results was further combined with stress–strain behavior, macroscopic failure characteristics, energy evolution, and fragmentation fractal dimensions.

The failure evolution shown in Figure was mainly interpreted from surface crack observations recorded by a digital camera. This method provides direct visual evidence of macroscopic crack development and fragment ejection but cannot fully capture the three-dimensional initiation, propagation, and coalescence of internal cracks. Therefore, the statements regarding crack initiation in the weaker section should be understood as interpretations based on surface observations, postfailure morphology, and AE responses, rather than direct three-dimensional evidence of internal cracking. Further studies using CT scanning, three-dimensional AE localization, or digital volume correlation are required to more accurately reveal the internal fracture evolution of the roof–coal–floor composite specimens.

In addition, the present experiments were conducted under uniaxial compression and should be regarded as a fundamental investigation of the influence of roof and floor lithology on coal section burst failure. In actual underground coal mines, coal pillars are subjected to three-dimensional in situ stress, lateral confinement, mining-induced dynamic disturbance, structural discontinuities, and scale effects. These factors may further influence energy accumulation, crack propagation, and dynamic failure intensity. Therefore, the conclusions obtained in this study should be combined with in situ stress conditions, mining layout, roof movement, and field monitoring data when they are applied to coal pillar burst risk assessment.

5. Conclusions

• The strength of the roof and floor has a significant influence on the mechanical response of the roof–coal–floor composite specimens. For specimens containing mudstone, the overall failure is mainly controlled by the weak mudstone section, and the stress–strain curves show relatively gentle prepeak and postpeak responses. For the SCS specimens with sandstone roof and floor, the composite specimens exhibit higher UCS, greater stiffness, and a sudden postpeak stress drop, indicating more evident brittle failure characteristics.

• The roof and floor strength strongly affects the burst propensity of the coal section. The burst energy indices of the MCM, MCS, and SCM specimens are relatively low, indicating weak burst propensity. In contrast, the SCS specimens exhibit a typical Class II stress–strain response, suggesting high burst propensity and sudden energy release in the coal section.

• The acoustic emission response further confirms the difference in the failure severity among the composite specimens. For specimens containing mudstone, the AE activity develops gradually and lasts for a relatively long period, indicating progressive damage evolution. For the SCS specimens, AE activity remains relatively weak before the peak but increases sharply near failure, and both the postpeak AE energy release rate and ring count rate are the highest among all specimen groups.

• The fragmentation size distribution indicates that the coal section in the SCS specimens experienced the most severe dynamic failure. The fractal dimension of SCS reaches 2.24, whereas those of the other specimens are less than 2.0. This indicates that hard sandstone roof and floor conditions lead to a higher proportion of small coal fragments and a more intense burst failure degree.

The data are available from the corresponding author upon reasonable request.

Z.Y.: Data curation; formal analysis; roles/writingoriginal draft; investigation. Z.Z.: Conceptualization; funding acquisition; resources; supervision. W.W.: Writingreview and editing. X.Y.: Methodology; writingreview and editing. P.W.: Visualization. J.H.: Project administration. S.G.: Validation.

This work was supported by the National Science and Technology Major Project for Deep Earth (No. 2024ZD1000705), the Xingliao Talent Program Leading Talent Project of Liaoning Province (No. XLYC2502016), and the Basic Scientific Research Project of the Liaoning Provincial Department of Education (Key Independent Project; No. LJ212410147007).

The authors declare no competing financial interest.