Mechanism by Which Backfill Body Reduces Amount of Energy Released in Deep Coal Mining

Deep coal mining is unavoidable, and the complex mining environments and the increasing dangers associated with ultrahigh energy accumulation and release from mining disturbances renders it extremely difficult to maintain a safe and stable stope. Solid backfilling technology directly uses coal gangue and other solid wastes in the mining area to fill the gob after mining. Support from the backfill body can inhibit themovement of overlying rock strata and significantly alleviate the influence of mining. In this study, the correlations between the deformation of gangue filling material and the characteristics of energy dissipation were examined under lateral uniaxial compression. 1e strain energy density distributions of backfilling and caving mining methods were simulated using numerical modeling. 1e results showed that the strain energy density distribution of backfilling mining was less concentrated, and its peak value was lower than that of caving mining by 51.0%, indicating that backfilling could effectively reduce the amount of energy released from mining rocks. 1e dense backfill mining area of the No. 9301 face in Tangkou Coal Mine was used as a case study. Measures for controlling the backfill body compaction for reducing the amount of energy released from mining rocks were proposed. 1ese measures include optimizing the support structure and filling material formula, controlling the preroof subsidence, and ensuring an appropriate number of tamping strokes.1emonitoring results of the backfilling quality, surface subsidence, and microseismic energy of No. 9301 working face in Tangkou Coal Mine showed that when the backfill body filling ratio control value was 82.28%, the total number of microseisms and the amount of energy released from the mining working face were significantly lower compared to those of the caving method. 1is study demonstrated that the backfill body could effectively reduce the amount of energy released frommining rocks, thereby realizing management of mine earthquake and sustainable deep coal mining.


Introduction
With the continuous and high-intensity mining of shallow coal resources and the depletion of coal resources in the eastern region, the resource development in China continues to move toward deep-earth mines at 1000 m to 2000 m. According to the incomplete survey, there are more than 140 deep coal mines, of which 50 have a depth of more than over 1000 m, especially the mining depth of Suncun Coal Mine has exceeded 1500 m. At these depths, the high concentration of stope stress results in the accumulation and release of ultrahigh mining energy. e frequency and intensity of the rockburst also increase significantly. e prevention and control of rock energy in deep coal mining are facing severe challenges [1][2][3][4][5]. Scholars over the world have proposed methods to control the energy accumulation in deep mining. Blake et al. [6] discovered an abnormal increase in microseismic (MS) activity before rock bursts induced by roof fracturing in the process of monitoring test in a hard rock mine. Ortlepp [7] provided strong evidence of an extremely violent fracturing that induced significant rock bursts in the faulting process during stoping of a highly stressed remnant in a deep gold mine by MS monitoring. Zhang et al. [8] proposed a method for hazard assessment in mines based on seismic energy distribution. Abdul-Wahed et al. [9] set up a close correlation between the location of seismic activity and induced stresses in the ground surface of the working areas by comparing the seismic activity and results of a numerical modeling of an advancing coalface with observations. Chen et al. [10] found that seismic energy and event count steadily increased accompanied by stress increase. Lu et al. [11][12][13] analyzed the frequency-spectrum evolutionary rule and precursory characteristics by experimental tests for combined coal and rock sample rock burst failure and in situ measurements of a strong rock burst in a coal mine.
As one of the core green mining technologies, solid backfilling mining technology (SBMT) [14][15][16] directly places coal gangue and other solid wastes into the gob after mining. It is widely applied in resolving many frontier scientific issues in the field of coal mining, such as deep mining [17], no-waste mining [18,19], safe mining of coal and associated resources [20], and safe hard roof mining [21][22][23]. Compared with conventional caving mining, the roof of the backfilling stope is only subjected to bending deformation, and the key layer also only undergoes a slight bending deformation from the load of the overlying strata. erefore, the support from the backfilling body showed a clear controlling effect on reducing the mining influence. However, the research mainly focuses on rock formation and mining pressure of SBMT, in-depth studies on the mechanism by which the backfill body helps reduce the amount of energy released during mining have not been conducted.
erefore, based on a new perspective of energy, this paper mainly studies the evolution of energy during filling material compression and coal seam excavation.
In this study, the correlations between the deformation and acoustic emission (AE) energy dissipation of the gangue filling material were tested under confined compression. e evolvement of mining energy distribution, while using backfilling and caving mining methods, was analyzed by numerical simulation of FLAC 3D . Based on the analysis of the energy release results, measures for reducing the amount of energy released from mining rocks by controlling the backfill body compaction were proposed. At the same time, the backfilling quality, surface subsidence, and mining energy between backfill mining working face and caving mining working face were measured, verifying the accuracy of the result. e results showed that backfill body could effectively control the stope energy accumulation and release, which is of great significance for realizing a safe, efficient, and green mining of deep coal resources.

Effect of Backfilling on Mining Influence
At the caving roof control working face, the overlying strata supporting system consists of only the coal wall before roof fracture. As the working face advances, the formed cantilever beam structure continues to expand until the suspended area becomes extremely large and causes fractures of the overlying strata. e support system transforms into a composition of "coal wall-hydraulic support-fractured swelling gangue in the collapsed gob" (Figure 1(a)). en, the roof fractures periodically as the working face advances. e bearing stress and the overlying strata displacement both change significantly, showing an obvious mining influence.
At the backfilling mining working face, the gob is constantly filled by filling materials. erefore, the overlying strata supporting system always consists of "coal wallbackfilling mining hydraulic support-backfill body" (Figure 1(b)). Supports from the backfill body inhibit the bending and subsidence of the roof, which will not be subjected to periodical fractures. ere is no obvious initial and periodic pressure phenomenon. e bearing stress and overlying strata displacement only change slightly. e mining influence is not obvious. e stress concentration coefficient is significantly smaller than that of caving mining. erefore, backfilling mining can significantly reduce the movement of overlying strata and mining influence.

Experimental Materials.
e gangue samples were selected from the excavation gangue of Tangkou Coal Mine in Shandong.

Experimental Equipment.
Two sets of devices, namely, the load control system and AE monitoring system, were used for the test. e load control system uses a microcomputer-controlled electrohydraulic servo universal testing machine. e maximum pressure is 1000 kN. e experimental process can be accurately controlled, and the required data can be recorded using a computer program. e sample container is a compacted steel cylinder with a depth of 300 mm, an inner diameter of 250 mm, and a maximum loading height of 270 mm. e AE monitoring system uses the PCI-2 AE testing and analyzing system developed by the Physical Acoustic Corporation (USA). e six evenly placed Nano30 AE sensors at the top and bottom ends of the steel cylinder could automatically count and store the AE parameters and realize real-time monitoring and positioning of the AEs. e experimental equipment is shown in Figure 3.

Compaction Properties of Backfill Materials.
e compaction deformation curves of gangue samples in different gradation conditions (0-10 mm, 0-20 mm, 0-30 mm, 0-40 mm, and 0-50 mm) are shown in Figure 4. e entire deformation process of gangue can be divided into three stages: rapid deformation (0-2 MPa), slow deformation (2-10 MPa), and stable deformation (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). In the rapid deformation stage, a small stress could cause a relatively large deformation of the gangue sample. e relative states between particles are unstable, such that the voids close quickly, and the deformation resistance is relatively low. In the slow deformation stage, the large particles break into small ones, which then re ll the voids, thereby strengthening the deformation resistance of the gangue sample rapidly. In the stable deformation stage, a relatively large stress can only induce a small strain in the gangue sample.
e deformation is mainly manifested as the compression of residual voids, the rebreakage of gangue into smaller particles, and the lling of regenerated voids. e deformation resistance of gangue is signi cantly enhanced in this stage. Under the same compaction stress, the larger the particle size, the larger the strain. is is because there are relatively more voids between large particles, and the strain from void closure is larger than that from particle breakage and deformation. e stress increases exponentially with strain under lateral compression for all the gangue samples.
e collected data are tted to the equation σ a 1 e a 2 ε + a 3 .

Dissipation Characteristics of AE Energy in Compressed
Gangue.
e dissipation characteristics of energy were analyzed by studying the variations in AE energy (E), cumulative counts (ΣN), cumulative energy (ΣE) with strain, and the spatial distribution of AE events at di erent time nodes. In the particle size range of 0-50 mm, the variations in stress, AE energy, cumulative energy with strain, and the spatial distribution of AE events are shown in Figure 5. e entire AE energy dissipation process can be divided into three stages: slow energy dissipation, accelerated energy dissipation, and attenuated energy dissipation. ese stages correspond to rapid deformation, slow deformation, and stable deformation stages in the deformation process, respectively. In the slow energy dissipation stage of AE activities, the AE energy was at a relatively low level. e compaction force mainly worked on rotating and adjusting the gangue particles and closing the voids between them. e AE event locations were scattered within the gangue samples. Most of the events occurred at the contacting points on the edges of the particles, which consumed relatively less energy. As the strain of the gangue samples entered the slow deformation stage, the corresponding AE activities entered the accelerated dissipation stage. More AE signals were recorded, and their spatial distribution became denser. e corresponding hopping signals also increased. e AE event locations gradually spread toward the blank area and became denser. Signi cantly, more event location points were detected in the center than at the end of the samples, and the energy dissipation rate also increased.
is was probably because the compaction force mainly worked on breaking the stable contact structure between gangue particles. e concentrated stress generated at the angular edges of the particles led to their massive breakage. As the strain continued to increase, the AE activities entered the attenuation stage, which corresponds to the stable deformation stage in the stress-strain curve. e particle sizes of gangue samples decreased signi cantly at this stage, and the dominant AE signal was the friction-induced AE, which was generated as a result of rubbing and colliding between particles during slippage and overturning. However, the occurrence frequency of hopping signals decreased significantly. Moreover, owing to the low probability of secondary breakage for gangue with small particle size, the AE activities were extremely limited.

Temporal and Spatial Distribution
(1) e compressive tangent modulus E g changes linearly with strain. At a fixed stress value, E g can be considered as the elastic modulus of the filling material subjected to a constant stress. e bulk modulus K and shear modulus G required for the Mohr-Coulomb model can be calculated. By using the inbuilt FISH language in FLAC 3D , the nonlinear compaction procedure is programmed accordingly: e strain energy density, which is strain energy per unit volume, in a spatially varying stress state is given by the following equation: Combined with the generalised form of Hooke's law, the strain energy density in the coal rock mass is obtained as follows: where σ 1 , σ 2 , and σ 3 denote the three principal stresses, while ε 1 , ε 2 , and ε 3 are the three principal strains. Moreover, E and u represent the elastic modulus and Poisson's ratio of coal rock masses, respectively. e three principal stress values of each point in the coal body ahead of the working face could be calculated through numerical computation. e strain energy density at each point in the coal body could then be obtained.

Model Establishment and Numerical Solution.
FLAC 3D developed by Itasca Consulting Group, Inc., is widely used in mining engineering; and it is capable of three-dimensional structural force behavior simulation and energy evolution analysis of soil, rock, and other materials. In FLAC 3D software, the procedure for strain energy calculation is compiled in FISH [24], in accordance with equation (4), thus giving the distribution of strain energy density in SBMT. A model of 700 m × 270 m × 600 m (length × width × height) was established using the Mohr-Coulomb model according to the engineering-geological conditions of No. 9301 working face of Tangkou Coal Mine ( Figure 6). e rock stratum thickness was approximately rounded according to the actual conditions. e physical and mechanical parameters were obtained from laboratory tests (Table S1). Furthermore, the meshes depicting those strata surrounding the coal seams are refined, and the model was divided into 520,800 elements and 543,414 nodes. e boundary conditions were set such that the horizontal freedom was constrained by the walls, whereas the freedom in all three directions was constrained by the bottom, i.e., a fixed bottom. No boundary conditions were applied to the top. Moreover, a compensation stress of 12.5 MPa was applied. In addition, coal columns of 100 m were left at the boundaries to eliminate the influence of boundary conditions on the excavation process. First, the in situ stress fields when applying compensated stress to the top surface and under the self-weight of rock strata were, respectively, simulated, followed by simulation of the excavation and backfill processes. e variations in mining energy were simulated for the backfilling method (filling ratio of 82%) and the caving mining method, recording the strain energy density distribution at the distance from the working surface of 48 m, 72 m, 112 m, and 200 m, respectively.

Evolution of Strain Energy Density with Advancing Working Face.
e distributions of strain energy density of the surrounding rock mass at different advancing distances of the working face using backfilling and caving methods obtained by Surfer are shown in Figures 7 and 8, respectively. An obvious increase in the strain energy density around the stope working face could be observed in the initial mining phase while using the caving method. When the advancing distances of the working face were at 48 m, 72 m, 112 m, and 200 m, the corresponding peak horizontal strain energy densities at the center of the coal seam were 921 kJ/m 3 , 1,010 kJ/m 3 , 1,096 kJ/m 3 , and 1,168 kJ/m 3 , respectively, whereas the in situ strain energy density was only approximately 360 kJ/m 3 . It can be concluded that the caving mining method could lead to a significant increase in the stope strain energy density. No areas with obviously increased strain energy density formed while using the backfilling method in the initial or subsequent mining phases. Furthermore, no significant changes were observed as the working face advanced. When the advancing distances of the working face were at 48 m, 72 m, 112 m, and 200 m, the corresponding peak strain energy densities were 517 kJ/m 3 , 531 kJ/m 3 , 549 kJ/m 3 , and 573 kJ/m 3 , which did not vary significantly. Compared with the caving mining stope, the backfilling mining stope was in an environment with slight energy increase. e overall destruction of the surrounding Shock and Vibration rock mass was mitigated, and the amount of energy released was signi cantly reduced. e strain energy density distributions at di erent mining locations in the central axis of the working face can be obtained by examining the cross section along the working face advancing direction, as shown in Figure 9. e following can be observed: (1) As the working face advanced, the strain energy density in the back ll body increased gradually and reached the maximum when the working face advanced 200 m. e distribution along the working face advancing direction has the shape of an inverted basin, where the strain energy density was high in the middle and low along the side.   (2) In the case of caving mining, as the advancement of a workface, the concentration coe cient of strain energy density increased gradually, as shown in Figure 9(a). When the working face advanced to 48 m, 72 m, 112 m, and 200 m, the corresponding concentration coe cients were 2.56, 2.81, 3.04, and 3.24, respectively. A concentration coe cient in the range of 2.5 to 3.5 indicated a high level of energy accumulation and release as well as a drastic change under the caving mining condition, which led to serious destruction of the surrounding rock mass. (3) In the case of back lling mining, as the working face advanced, the concentration coe cient of strain energy density did not change signi cantly, as shown in Figure 9(b). When the working face advanced to 48 m, 72 m, 112 m, and 200 m, the corresponding concentration coe cients were 1.44, 1.48, 1.53, and 1.59, respectively. A concentration coe cient of approximately 1.5 indicated a low level of energy accumulation and release as well as limited changes under the back lling condition, where the mining gob was compactly lled by the back ll body. erefore, there was little destruction of the in situ stress eld. (4) e strain energy densities ahead of the working face of caving stope and back lling stope showed similar distribution patterns, as shown in Figure 9(c). e strain energy density increased rapidly in the beginning and then dropped slowly after reaching the peak. However, compared with the back lling method, a wider range of strain energy concentration ahead of the working face was observed when using the caving method. e peak value also occurred at a farther location, and the strain energy density concentration coe cient was relatively larger. In the case of caving method, the strain energy was concentrated within a 45 m range ahead of the working face, and a peak strain energy of 1,168 kJ/m 3 was 10 m ahead of the working face. In the case of back lling method, the strain energy was concentrated within a 20 m range ahead of the working face, and a peak strain energy of 573 kJ/m 3 was 6 m ahead of the working face. e 51% drop in the peak strain energy density indicated that the back lling method could e ectively reduce the amount of energy released from the mining rock.   Figure 10 shows the layout of the mining area and the distribution of underground layers. Because the corresponding surface above the working face was a village, dense lling was set as the controlling criterion for back ll mining.

Control of Filling Ratio to Weaken Mining Energy.
rough numerical analysis, we conclude that the back ll body can e ectively reduce amount of the energy released in deep mining. erefore, guaranteeing the lling ratio is the key factor to the release of mining energy. During the actual solid back ll mining process, the control of the back ll body lling ratio can be performed during the design and implementation stages to achieve a reduction in the amount of energy released [25,26], as shown in Figure 11. In the design stage, the appropriate tamping force and tamping angle are rst determined by optimizing the supporting structure and lling material composition. is ensures a su ciently high density of the lling material. During the implementation stage, the presubsidence magnitude of top roof, number of tamping strokes, and height of lling material were controlled carefully. More than su cient lling materials were back lled in the mining cavity for ensuring a uniform height of the back ll content. e lling ratio was also monitored at di erent mining locations to obtain instantaneous feedback of the back ll material density. Table 1 shows the main methods of improving roofcontrolled back lling ratio.  Shock and Vibration

Monitoring of Re lling Mass.
A dynamic monitor was installed on the roof inside the lling material for keeping track of the dynamic response of the roof. Speci cally, as the re lling process proceeded to di erent working areas, four roof dynamic monitors were installed as one group as shown in Figure 12(a). e measurement curves are shown in Figures 12(b) and 12(c). e following characteristics were observed from the curves: (1) e subsidence curve of the roof changes continuously without any sudden jump. is indicates that the roof used for monitoring did not break or slide o during the sampling period. An intense activity of the roof was observed during the early stage of the re lling process. As the re lling proceeds to 188 m, a stable response was observed from the roof. e maximum subsidence of the roof was found to be 620 mm at the peak point.
is suggests that the actual lling ratio was controlled at 82.28%.
(2) With increasing subsidence of the roof in the re lling area, the re lling stress increased as the re lling materials were tamped gradually. e internal stress of the lling material reached a stable value of 22.4 MPa when the working plane reached 89 m. is value approaches the original rock stress, which indicates that the bending deformation of the overlying stratum was equilibrated by the support from the re lling material.

Measurement of Microseismic Energy.
During the mining of No. 9301 working face, the MS energy was measured using an MS monitoring system. e recorded results were compared with the No. 1303 fully mechanized mining face located approximately 1,000 m to its north. Both working faces had similar burial depths and dip angles. e measured energy distributions of microseismicity are shown in Figure 13, and the MS events are presented in Table 2.

Shock and Vibration
Six MS events with energy greater than 5,000 J were detected from the No. 9301 backfill mining working face, whereas 49 such events were detected from the No. 1303 fully mechanized mining face. Among them, seven events had energy greater than 10,000 J, and the maximum MS energy reached 17,100 J. e total number of MS events and average MS energy measured from the No. 9301 working face were much smaller compared with the ones from the No. 1303 working face. No large energy release was detected from the No. 9301 working face, indicating that the solid backfilling mining method effectively reduced the amount of energy released.

Measurement of Surface Subsidence.
In order to monitor the effect of underbuilding refilling mining on the surface buildings at mining area No. 930, a monitor system was designed with a monitor point spacing of 25 m and a control point spacing of 50 m, based on the actual condition at the ground surface. Figure 14 shows the detailed installation map of surface subsidence measuring points and the monitoring results. Until now, the coal extraction activities at working areas No. 9301, No. 9302, and No. 9303 are already finished, and no significant deformation was observed on any surface buildings. e maximum surface subsidence was far smaller than the critical deformation value required for inducing permanent damage to the masonry buildings. Hence, these measurement results demonstrate that a filling ratio of 82.28% can effectively protect the surface buildings from mining activity . Ensure that the tamping counts for each backfilling are more than 10 strokes Improve the initial loaded stress on the backfill body and reduce the compression ratio Determine the initiation of the next cycle of coal extraction by the backfilling quality Dynamic monitoring of the filling ratio

Monitoring
Install equipment for monitoring support resistance, tamping force, and filling ratio to achieve real-time filling quality monitoring system Establish a step-by-step inspection system to check and rectify the filling quality

Common measures Operating specification
Establish complete training and assessment specifications and improve the technical capabilities of backfilling operators Form a five-people team to perform backfilling alternatively and minimize the influence of personal reasons on the filling quality Adjust salary based on filling quality to motivate operators

Conclusions
(1) In the compaction process of the gangue samples, the AE energy evolution corresponded well with the stress-strain curves. During the three deformation stages, namely, rapid deformation, slow deformation, and stable deformation, shown in the stressstrain curves, the monitored AE activities showed three corresponding energy dissipation stages, i.e., slow dissipation, accelerated dissipation, and attenuated dissipation. e locations of detected AE events showed the most AE activities at the edge of the gangue sample center. e compression test of gangue samples provided data to support the numerical calculations of energy distributions in the mining stope.  (2) Based on the principle of low degree of disturbance and destruction of rock mass and lack of mining pressure appearance at the backfill mining stope, the mechanism by which the backfill body reduces the amount of energy released during mining of rocks was proposed. e evolving characteristics of strain energy density with the advancing working face were revealed. Compared with the caving mining stope, the backfilling mining stope had a relatively less concentrated strain energy density distribution, and the peak value was 51% less. is indicated that the backfilling method effectively reduced the amount of energy released from the mining rocks. (3) Based on the mining geological conditions of No. 9301 solid backfilling mining working face in Tangkou Coal Mine, measures for reducing the amount of energy released by controlling the compaction of filling materials were proposed. ese measures include optimization of the support structure and filling material formula, controlling the preroof subsidence, ensuring an optimal number of tamping strokes, and dynamic monitoring of the filling ratio. Engineering practices showed that a filling ratio control value of 82.28% could effectively reduce the amount of energy released, thereby realizing safe and sustainable deep coal mining.

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this article.

Authors' Contributions
Wenyue Qi and Nan Zhou contributed equally to the work and should be regarded as co-first authors.